CN118446969A - Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud - Google Patents

Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud Download PDF

Info

Publication number
CN118446969A
CN118446969A CN202410505174.7A CN202410505174A CN118446969A CN 118446969 A CN118446969 A CN 118446969A CN 202410505174 A CN202410505174 A CN 202410505174A CN 118446969 A CN118446969 A CN 118446969A
Authority
CN
China
Prior art keywords
litchi
point cloud
dimensional point
image
tested
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202410505174.7A
Other languages
Chinese (zh)
Inventor
陈海波
向星岚
曾山
龚康业
李正心
黄顺豪
李纯煕
蔡晓峰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China Agricultural University
Original Assignee
South China Agricultural University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China Agricultural University filed Critical South China Agricultural University
Priority to CN202410505174.7A priority Critical patent/CN118446969A/en
Publication of CN118446969A publication Critical patent/CN118446969A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Geometry (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud, wherein the method comprises the following steps: acquiring two-dimensional images of litchi to be detected under multiple visual angles; mapping and processing two-dimensional images of the litchi to be detected under a plurality of view angles to obtain three-dimensional point cloud data of the litchi to be detected, calculating point cloud curvature and normal vector of the three-dimensional point cloud data of the litchi to be detected, counting curvature threshold values, and identifying mechanical injury areas of the litchi; and performing color model space conversion on the obtained RGB image of the litchi to be detected, extracting red pixel points in the litchi image under the HSI model through red pixel point threshold values to obtain the red coloring rate of the litchi, obtaining a depth image containing the thickness of the litchi according to the depth image and the background depth image, and calculating the volume of the litchi by adopting the 3D models at the top and the bottom of the litchi to be detected, which are built by the depth image. The invention can rapidly and accurately detect the phenotype parameters of the litchi and provide scientific basis for quality classification of the litchi.

Description

Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud
Technical Field
The invention belongs to the technical field of fruit phenotype parameter detection, and particularly relates to a litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud.
Background
The post-partum grading of litchi is an important process before fresh-keeping packaging, sales and deep processing, and is also a key for improving the added value of litchi. At present, the post-partum classification treatment of litchi in China is still relatively backward, and the litchi is classified mainly in a manual mode, however, the post-partum classification of litchi in a manual mode has the problems of higher labor cost, lower classification precision and low efficiency, and in addition, the litchi is extremely easy to damage in the process of manually post-partum classification of the litchi due to the thin epidermis and high water content of the litchi, so that the sales value of the litchi is greatly influenced.
With the continuous development of machine vision technology in recent years, the machine vision technology is gradually applied to quality detection of fruit products, and researchers extract phenotype parameters of litchi through three-dimensional point cloud data by collecting the three-dimensional point cloud data of the litchi, so that nondestructive postpartum grading of the litchi is realized. However, because the size of the litchi is smaller, the accuracy of the three-dimensional point cloud of the collected litchi is lower, so that the accuracy of the litchi phenotype parameter detection result is greatly influenced, in addition, the current litchi surface parameter detection still only uses the size of the litchi fruit diameter as a grading standard, the problem of single litchi grading basis exists, and the situation of uneven maturity of the same graded litchi is caused.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a litchi phenotype parameter detection method based on image processing and three-dimensional point cloud.
The invention further aims to provide a litchi phenotype parameter detection system based on image processing and three-dimensional point cloud.
In order to achieve the above object, the present invention can be achieved by adopting the following technical scheme:
A litchi phenotype parameter detection method based on image processing and three-dimensional point cloud comprises the following steps:
s1, acquiring two-dimensional images of litchi to be detected under multiple view angles, wherein the two-dimensional images comprise depth images and RGB images;
S2, mapping the two-dimensional images of the litchi to be detected under a plurality of view angles to obtain original three-dimensional point cloud data of the litchi to be detected, carrying out background segmentation and noise reduction filtering treatment on the original three-dimensional point cloud data to obtain three-dimensional point cloud data of the litchi to be detected, calculating the point cloud curvature and normal vector of the three-dimensional point cloud data of the litchi to be detected, and counting the curvature threshold value for judging the litchi surface loss area, so that point cloud of the litchi damage area is segmented, and the mechanical damage area of the litchi to be detected is identified;
S3, converting the RGB images of the litchi to be detected under the different visual angles obtained in the step S1 into an HSI color model space to obtain litchi images under the HSI models of the different visual angles, extracting respective red pixel points in the litchi images under the HSI models of the different visual angles through a preset red pixel point threshold value, taking an average value, and taking the average value as the red coloring rate of the litchi to be detected;
s4, obtaining a depth image h (x, y) containing the thickness of the litchi according to the depth images f (x, y) and the background depth images b (x, y) of the litchi to be detected under different visual angles, preprocessing the depth image h (x, y) containing the thickness of the litchi, and calculating the volume value of the litchi to be detected according to the 3D models of the top and the bottom of the litchi to be detected by adopting the 3D models of the top and the bottom of the litchi to be detected, which are established by the preprocessed depth image h (x, y).
Preferably, the plurality of viewing angles in step S1 includes four viewing angles of left view, right view, top view and bottom view.
Preferably, in step S2, the deriving process of the point cloud curvature and the normal vector of the three-dimensional point cloud data of the litchi to be detected is as follows:
Because each point in the point cloud has a neighborhood point cloud, the neighborhood point cloud is approximated by using a specific surface, and the curvature of a certain point can be represented by the curvature of a local curved surface fitted by the point and the neighborhood point thereof; therefore, k points are uniformly selected around the circumference rate with the circumference rate as a center point, and a quadric equation can be expressed as follows:
z(x,y)＝ax2+bxy+cy2
According to the least squares principle, the sum of squares of z i can be expressed as:
from this, the values of the coefficients a, b, c of the quadric equation can be found:
The curve on the curved surface can be expressed as:
r＝(x(t),y(t))
The unit normal vector at point Pi can be expressed as:
taking the normal vector at the Pi point as the normal vector of the local curved surface in the neighborhood, and then the covariance matrix of each point in the neighborhood is:
C·Xj＝λj·Xj，j＝1,2,3
Wherein P 0 is the centroid of the neighborhood points, k is the number of the neighborhood points, and lambda j and X j respectively represent the characteristic value and the characteristic vector of C.
The feature vector corresponding to the minimum feature value of the matrix C is a point (normal vector of P i, a covariance matrix is constructed, the feature value of the matrix is calculated through feature value decomposition, and when the feature value satisfies the condition that λ0 is less than or equal to λ1 is less than or equal to λ2, the curvature of the neighboring point can be expressed as:
Finally, the neighborhood curvature k i is the curvature degree of the neighborhood curved surface.
Preferably, the curvature threshold for determining the litchi surface loss area in step S2 is 1.81.
Preferably, the specific procedure of step S3 is as follows:
Firstly, respectively carrying out normalization processing on an R component, a G component and a B component in RGB images under a plurality of different visual angles, wherein the color of a pixel point positioned at a space position (x, y) in an RGB color space is represented by 3 numerical values of the R component R (x, y), the G component G (x, y) and the B component B (x, y) of the pixel point; then, the litchi area in the RGB image after normalization processing is converted into an HSI color space, and the H component H (x, y), the S component S (x, y) and the I component I (x, y) of the pixel point in the HSI color space are expressed as follows:
Wherein,
Finally, according to the formula, litchi images under the HSI model corresponding to RGB images under a plurality of different visual angles of the litchi to be detected are obtained;
Extracting respective red pixel values in the litchi images under the HSI models of the different visual angles according to a preset red pixel point threshold value, and averaging, wherein the average value is used as the red coloring rate of the litchi to be detected; the red coloring rate is the ratio of the red pixel points on the surface of the whole litchi to be detected to the total area value of all the pixel points of the litchi to be detected; the red pixel points are red H components of the litchi surface in the litchi image to be detected under the HSI model.
Preferably, the preset threshold value of the red pixel point is 35-40 degrees.
Preferably, in step S4, the preprocessing mode of the depth image h (x, y) including the litchi thickness is erosion, expansion and gaussian smoothing.
Preferably, the shooting environments and shooting heights of the depth image f (x, y) and the background depth image b (x, y) in step S4 are the same.
The litchi phenotype parameter detection system based on the image processing and the three-dimensional point cloud is used for the litchi phenotype parameter detection method based on the image processing and the three-dimensional point cloud, and comprises a litchi image acquisition platform and a cloud server;
The litchi image acquisition platform comprises a camera bellows, an LED light source, a litchi placing plate, a depth camera and a camera bracket, wherein the LED light source is arranged at the top in the box body of the camera bellows, the depth camera is arranged below the LED light source through the camera bracket, the litchi placing plate is arranged at the bottom of the box body of the camera bellows, and a lens of the depth camera faces the litchi placing plate; the litchi placing plate is used for placing litchi to be detected, and the depth camera is in communication connection with the cloud server;
The cloud server comprises a processor and a memory; the memory having stored thereon non-transitory computer instructions which, when executed by a processor, perform the litchi phenotype parameter detection method based on image processing and three-dimensional point cloud as claimed in any one of claims 1-8.
Preferably, the depth camera employs INTEL REALSENSE D a 405 depth camera.
Compared with the prior art, the invention has the following advantages:
(1) According to the litchi phenotype parameter detection method based on image processing and three-dimensional point cloud, the point cloud data of the litchi to be detected is obtained by obtaining the RGB images and the depth images of the litchi to be detected under 4 visual angles, and the curvature threshold value of the litchi surface loss area is obtained based on the point cloud normal vector and the curvature algorithm, so that the litchi mechanical damage is rapidly and accurately identified, the litchi skin is effectively protected from being damaged in the detection process, the economic value of the litchi is effectively improved, the average identification accuracy of the litchi mechanical damage area according to multiple test statistics reaches 94%, and the litchi phenotype parameter detection accuracy is improved.
(2) The litchi phenotype parameter detection method based on image processing and three-dimensional point cloud comprises the steps of performing RGB-HSI color model space conversion on RGB images of litchi to be detected under a plurality of different visual angles to obtain a litchi image under an HSI model, extracting red pixel points in the litchi image under the HSI model through a preset red pixel point threshold value, and finally calculating to obtain the red coloring rate of the litchi; compared with the conventional method for reconstructing three-dimensional point cloud of litchi based on MVS-SFM algorithm in the prior art, the method provided by the invention has the advantages that the coloring rate of the litchi point cloud is improved by 90% in speed, so that the efficiency of litchi phenotype parameter detection is greatly enhanced.
(3) According to the litchi phenotype parameter detection system based on image processing and three-dimensional point cloud, the litchi to be detected is acquired through ingenious matching of the camera bellows, the LED light source, the litchi placing plate, the depth camera and the camera support, the litchi depth image and the RGB image under 4 visual angles are acquired, the three-dimensional point cloud data of the litchi to be detected are obtained through mapping of the acquired litchi depth image and the acquired RGB image under 4 visual angles, and therefore the three-dimensional point cloud data of the litchi to be detected are more accurate, and the accuracy of litchi phenotype parameter detection is improved.
Drawings
Fig. 1 is a schematic diagram of a litchi phenotype parameter detection method based on image processing and three-dimensional point cloud provided in embodiment 1 of the present invention;
Fig. 2 is an RGB image, a depth image and a point cloud image of litchi at multiple viewing angles provided in embodiment 1 of the present invention;
Fig. 3 is a schematic diagram of comparison between an original three-dimensional point cloud and a processed point cloud of litchi provided in embodiment 1 of the present invention;
fig. 4 is a schematic diagram showing the comparison of the litchi RGB image and the HSI model image provided in embodiment 1 of the present invention;
Fig. 5 is a visual result diagram of litchi red pixel extraction provided in embodiment 1 of the present invention;
Fig. 6 is a schematic view of a depth image of the top and bottom of litchi provided in embodiment 1 of the present invention;
fig. 7 is a schematic diagram of the litchi depth image h (x, y) provided in embodiment 1 before and after processing;
Fig. 8 is a schematic diagram of a top model of litchi provided in embodiment 1 of the present invention;
Fig. 9 is a schematic view of a litchi bottom model provided in embodiment 1 of the present invention;
fig. 10 is a graph of the results of the mechanical damage experiment of litchi provided in example 1 of the present invention;
FIG. 11 is a schematic flow chart of a litchi point cloud coloring rate extraction process based on the reconstruction of three-dimensional point cloud of litchi by MVS-SFM algorithm provided in embodiment 1 of the present invention;
fig. 12 is a linear result chart of litchi tinting rate and artificially measured tinting rate extracted based on an litchi RGB graph and a color space model provided in embodiment 1 of the present invention;
fig. 13 is a linear result diagram of the extraction of the coloring rate of the litchi point cloud and the manual measurement of the coloring rate based on the reconstruction of the three-dimensional point cloud of litchi by the MVS-SFM algorithm provided in the embodiment 1 of the invention;
FIG. 14 is a graph showing the predicted volume results and the linear results of the manual measurement of the volume of litchi provided in example 1 of the present invention;
Fig. 15 is a schematic structural diagram of a litchi phenotype parameter detection system based on image processing and three-dimensional point cloud according to embodiment 2 of the present invention.
Wherein 1 is the camera bellows, 2 is the LED light source, 3 is the camera support, 4 is the depth camera, 5 is the litchi that awaits measuring, 6 is cloud ware, 7 is the litchi and places the board.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments, and all other embodiments obtained by those skilled in the art without making any inventive effort based on the embodiments of the present invention are within the scope of protection of the present invention.
The technical scheme of the invention is further described below by the specific embodiments with reference to the accompanying drawings.
In the description of the present invention, it should be understood that the directions or positional relationships indicated by the terms "upper", "lower", "top", "bottom", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention.
Example 1
As shown in fig. 1, the litchi phenotype parameter detection method based on image processing and three-dimensional point cloud comprises the following steps:
s1, acquiring two-dimensional images of litchi to be detected under multiple view angles, wherein the two-dimensional images comprise depth images and RGB images; specifically, the plurality of viewing angles includes four viewing angles of left-hand view, right-hand view, top-down view and bottom-hand view.
Specifically, INTEL REALSENSE D depth camera 4 is used in this embodiment to collect depth images and RGB images of the litchi under test under multiple viewing angles, where each pixel point of the depth image represents a distance between the pixel point and the plane of depth camera 4, and the unit of the distance is mm.
S2, mapping the two-dimensional images of the litchi to be detected under a plurality of view angles to obtain original three-dimensional point cloud data of the litchi to be detected, carrying out background segmentation and noise reduction filtering treatment on the original three-dimensional point cloud data to obtain three-dimensional point cloud data of the litchi to be detected, calculating the point cloud curvature and normal vector of the three-dimensional point cloud data of the litchi to be detected, and counting the curvature threshold value for judging the litchi surface loss area, so that point cloud of the litchi damage area is segmented, and the mechanical damage area of the litchi to be detected is identified;
specifically, the process of acquiring the original three-dimensional point cloud data of the litchi to be detected is as follows:
Firstly, transmitting the depth image and the RGB image of the litchi to be detected under the multiple visual angles acquired by the depth camera 4 in the step S1 to a cloud server 6, processing the depth image and the RGB image of the litchi to be detected under the multiple visual angles by adopting python software and combining corresponding programming driving of the depth camera 4 in the cloud server 6, so that the depth image and the RGB image are combined, converting two-dimensional pixel data into points in a three-dimensional space through calibration parameters of the depth camera 4, mapping color data in the RGB image onto corresponding three-dimensional points, enabling the generated point cloud to have real color information, and finally obtaining the original three-dimensional point cloud of the litchi to be detected through conversion of a mapping relation.
Specifically, as shown in fig. 3, since the original three-dimensional point cloud data of the litchi to be detected obtained by the depth camera 4 includes a large number of hash points and isolated points, and when the point cloud data is obtained, because of the influence of the precision of the device, the experience of an operator and environmental factors, the diffraction of electromagnetic waves, the change of the surface property of the object to be detected and the error influence in the data stitching registration operation process, some unavoidable noise and other point clouds irrelevant to the litchi will appear in the point cloud data, in order to obtain better three-dimensional point cloud data, in this embodiment, the obtained original three-dimensional point cloud data of the litchi to be detected is subjected to background segmentation by adopting cloudcompare software, and the litchi point cloud is subjected to noise reduction and filtering processing based on a bounding box algorithm and a statistical filtering (STATISTICAL OUTLIER REMOVAL, SOR) method, so that the clean and complete litchi point cloud is finally obtained.
Since the surface curvature of the point cloud is a concept of eigenvalues that is used to describe the degree of change in the surface of the point cloud. The mechanical damage formed in the processes of growing, picking and grading after picking of the litchi can enable the surface of the litchi to form a certain depression or bulge, the depression or bulge area of the surface of the litchi can change the original shape of the surface of the fruit, the point cloud curvature of the area of the litchi is changed and is usually expressed as an abnormally high value or a abnormally low value, so that after the surface curvature of the point cloud of the litchi is obtained through calculation, a threshold value can be obtained through statistics, the curvature point cloud area under the threshold value range is the point cloud area with the changed curvature of the surface of the litchi, the point cloud area is the mechanical damage area of the litchi, which is obtained through identification, of the litchi, the threshold value range is the curvature change threshold value of the point cloud of the litchi, and then the mechanical damage area of the litchi to be detected can be identified through the threshold value.
Specifically, the calculation principle of the point cloud curvature and the normal vector of the three-dimensional point cloud data of the litchi to be detected is as follows:
The degree of change of the point cloud is described by the eigenvalue due to the surface curvature. By estimating the normal vector for each point in the point cloud, the curvature of a point can be obtained. The normal vector may be calculated using the tangent normal of the surface at that point. According to the least squares method (Least Squares Method, LSM), a quadric may be used to characterize the local region and a local region surface fitting method may be used to estimate the normal vector.
Because each point in the point cloud has a neighborhood point cloud, the neighborhood point cloud is approximated by using a specific surface, and the curvature of a certain point can be represented by the curvature of a local curved surface fitted by the point and the neighborhood point thereof; therefore, k points are uniformly selected around the circumference rate with the circumference rate as a center point, and a quadric equation can be expressed as follows:
z(x,y)＝ax2+bxy+cy2
According to the least squares principle, the sum of squares of z i can be expressed as:
from this, the values of the coefficients a, b, c of the quadric equation can be found:
The curve on the curved surface can be expressed as:
r＝(x(t),y(t))
The unit normal vector at point Pi can be expressed as:
taking the normal vector at the Pi point as the normal vector of the local curved surface in the neighborhood, and then the covariance matrix of each point in the neighborhood is:
C·Xj＝λj·Xj，j＝1,2,3
Wherein P 0 is the centroid of the neighborhood points, k is the number of the neighborhood points, and lambda j and X j respectively represent the characteristic value and the characteristic vector of C.
The feature vector corresponding to the minimum feature value of the matrix C is a point (normal vector of P i, a covariance matrix is constructed, the feature value of the matrix is calculated through feature value decomposition, and when the feature value satisfies the condition that λ0 is less than or equal to λ1 is less than or equal to λ2, the curvature of the neighboring point can be expressed as:
Finally, the obtained neighborhood curvature k i is the point cloud curvature of the curved surface area.
Specifically, after iterative analysis is performed on the point cloud curvatures of different litchi point cloud samples in Cloudcompare software, the accuracy of the measured surface loss curvature of litchi is finally obtained when the curvature threshold is set to be 1.81.
S3, converting the RGB images of the litchi to be detected under the different visual angles obtained in the step S1 into an HSI color model space to obtain litchi images under the HSI models of the different visual angles, extracting respective red pixel points in the litchi images under the HSI models of the different visual angles through a preset red pixel point threshold value, taking an average value, and taking the average value as the red coloring rate of the litchi to be detected;
Specifically, the specific process of step S3 is as follows:
Firstly, acquiring the values of an R component, a G component and a B component of each pixel point in the RGB image under a plurality of different visual angles obtained in the step S3 by adopting a python+opencv writing program, and then respectively carrying out normalization processing on the values of the R component, the G component and the B component of each pixel point to obtain that the color of the pixel point positioned at a space position (x, y) in an RGB color space is represented by 3 values of the R component R (x, y), the G component G (x, y) and the B component B (x, y) of the pixel point; then, the litchi area in the RGB image after normalization processing is converted into an HSI color space, and the H component H (x, y), the S component S (x, y) and the I component I (x, y) of the pixel point in the HSI color space are expressed as follows:
Wherein,
The RGB color space is based on a three-dimensional rectangular coordinate system and comprises R, G, B original spectrum components, an HSI model is set according to a human visual system and consists of H, S, I components, wherein an H component represents the color of an object, an I component represents the brightness, an S component represents the saturation of the color, and θ (x, y) is the value of an included angle in an H component conversion formula.
As shown in fig. 4, finally, according to the above formula, obtaining litchi images under the HSI model corresponding to the RGB images under a plurality of different viewing angles of the litchi to be detected;
The smaller the H value of a pixel point in the HSI color space is, the more red the hue of the pixel point is, so that the pixel point with the H component smaller than the preset red pixel point threshold value in an image is marked as a red pixel point according to the preset red pixel point threshold value, then the respective red pixel point values in litchi images under the HSI models with different visual angles are obtained through calculation, and the average value is used as the red coloring rate of litchis to be measured; the red coloring rate is the ratio of the red pixel points on the surface of the whole litchi to be detected to the total area value of all the pixel points of the litchi to be detected;
the method for acquiring the preset red pixel point threshold value comprises the following steps:
Firstly, different litchi image samples under an HSI model are obtained, then the red coloring rate corresponding to the samples is obtained by adjusting the size of the red pixel threshold value, the obtained red coloring rate is verified by the litchi coloring rate standard, and finally the value range of the red pixel threshold value with the most accurate verification result is used as the preset red pixel threshold value.
Specifically, the red coloring rate measurement experiment is carried out on the litchi sample for a plurality of times, and the result is the most accurate when the value range of the preset red pixel point threshold value is 35-40 degrees.
S4, obtaining a depth image h (x, y) containing the thickness of the litchi according to the depth images f (x, y) and the background depth images b (x, y) of the litchi to be detected under different visual angles, preprocessing the depth image h (x, y) containing the thickness of the litchi, and calculating the volume value of the litchi to be detected according to the 3D models of the top and the bottom of the litchi to be detected by adopting the 3D models of the top and the bottom of the litchi to be detected, which are established by the preprocessed depth image h (x, y).
The specific process of step S4 is as follows:
Since a depth image is an image or image channel containing information about the distance of the surface of a scene object from the viewpoint, also called range view, directly reflects the geometry of the visible surface of the photographic subject. Each non-zero gray area pixel of the depth image represents the height of the litchi surface within one pixel area, so the thickness of the litchi can be defined as the distance from the surface of the litchi to the platform on which the litchi is placed.
Since the depth image f (x, y) of the litchi indicates the distance between the surface of the litchi and the depth camera 4, this distance does not directly reflect the thickness of the litchi. It is necessary to keep the depth camera 4 at the same height, the background depth pattern b (x, y) under the same photographing environment, and construct a depth image h (x, y) containing the litchi thickness based on the depth image f (x, y) and the background depth image b (x, y), the depth image h (x, y) containing the litchi thickness being expressed as follows:
Then using python+opencv writing program to erode, expand and Gaussian smooth depth image h (x, y), removing tiny noise in the image, filling small holes, and obtaining the result as shown in figure 7
And then, the processed preprocessed depth image h (x, y) is imported into Matlab software and built into 3D models of the top and bottom of the litchi to be detected, and at the moment, the non-zero pixel value in the image h (x, y) is the thickness of the litchi at the point, and the default pixel value unit of the depth image obtained by directly storing the processed image after shooting by a camera is a sub-millimeter, so that the unit is converted into mm. Meanwhile, the number of the non-zero pixel values represents the surface area S of the litchi projected into the two-dimensional image space, and the actual area of each pixel is calculated to be equal to 0.089mm 2.
Specifically, the actual volume acquisition process of each pixel is as follows:
Firstly, shooting a green card paper with the area of 100 square centimeters vertically downwards by adopting the same camera height and angle as those of shooting litchi data, then using python+opencv to write a program, dividing the green card paper area, calculating the number of pixels in the area in an image, and dividing the actual area of the card paper by the number of pixels to obtain the actual area of each pixel. It should be noted that, this value represents the actual area represented by each pixel point in the depth image generated by the depth camera, and does not change with the change of the external environment, so it can be used in the subsequent litchi detection.
The volume of litchi is defined as the sum of the volumes of each pixel in the litchi surface area, and because only half of litchi thickness information is contained in one depth image h (x, y), the minimum non-zero pixel value is subtracted from each non-zero pixel value in h (x, y) to obtain half of the litchi volume. And finally, calculating the volume of the whole litchi through the three-dimensional surface models of the top and the bottom of the litchi.
The volume calculation of litchi is expressed as follows:
V Total (S) ＝V Top +V Bottom
Conclusion and analysis
In order to provide efficient technical means and data support for post-partum grading of litchis, the litchi phenotype parameter detection method based on image processing and three-dimensional point cloud in the embodiment is subjected to data analysis in an experimental mode.
The test selects Feizixiao litchi variety as a sample, and is collected from Guangzhou city increasing urban area in Guangdong province.
(1) Mechanical injury data analysis
In order to verify the accuracy of litchi point cloud loss area identification, 50 litchis subjected to mechanical damage of different degrees are prepared for the test, and the litchi mechanical damage identification accuracy under different number of view angles is detected.
As shown in table 1 below, the damage detection accuracy of litchi at a single viewing angle, 2 viewing angles, 3 viewing angles and 4 viewing angles was tested, curvature calculation was performed after point cloud synthesis, and the detection result of mechanical damage on the litchi surface was obtained. From the results in table 1, it can be seen that the detection rate of mechanical damage to litchi is lower under a single visual angle, the number of missed litchi detection is more, the detection accuracy of mechanical damage is obviously improved along with the increase of the detection visual angles, and the detection accuracy of litchi under 4 detection visual angles is 94%. According to the detection result, the method for calculating the point cloud curvature change value can be used for rapidly and effectively detecting the mechanical damage of the litchi under a certain error permission condition, and has high detection accuracy.
Viewing angle (personal) Total number of samples (number) Correct detection quantity (number) Detection accuracy (%)
1 50 30 60
2 50 38 76
3 50 42 84
4 50 47 94
Table 1 litchi mechanical injury test results
(2) Litchi coloring rate extraction data analysis
In order to verify the extraction result of the litchi coloring rate, the experiment uses the judgment of the litchi coloring rate by manpower as the verification of real data, and in addition, the experiment adopts the method for reconstructing three-dimensional point cloud of litchi based on MVS-SFM algorithm in the prior art to extract the litchi point cloud coloring rate, and the method is compared with the method;
Specifically, as shown in fig. 12, the result obtained in the step of extracting the litchi coloring rate provided by the invention and the litchi coloring rate manually judged are obviously linearly related, the correlation coefficient R2 and the root mean square Error (Root Mean Square Error, RMSE) of the litchi coloring rate in the RGB chart are 0.9574 and 0.0809%, the average absolute Error (Mean Absolute Error, MAE) is 6.33%, and the average relative Error (MEAN RELATIVE Error, MRE) is 4.17%, respectively.
Specifically, as shown in fig. 11, the process of reconstructing three-dimensional point cloud of litchi based on MVS-SFM algorithm to extract the coloring rate of litchi point cloud is as follows:
Firstly, reconstructing an acquired litchi image by using an MVS-SFM algorithm to obtain litchi point cloud, then carrying out plane segmentation on the litchi point cloud based on a random sampling consistency algorithm (Random Sample Consensus, RANSAC), obtaining complete and clean litchi point cloud through processing such as statistical filtering, and finally obtaining a red coloring rate result of the litchi surface according to the determined litchi red threshold range.
As shown in FIG. 13, the three-dimensional point cloud of the reconstructed litchi finally based on the MVS-SFM algorithm shows obvious linear distribution of the litchi point cloud coloring rate and the litchi coloring rate judged manually, wherein R2 and RMSE are respectively 0.9205 and 0.0563%, MAE is 4.37%, and MRE is 6.01%. The result shows that the litchi tinting rate result obtained through the computer vision processing has better correlation with the tinting rate result judged by manual measurement.
Finally, although the MVS-SFM algorithm can obtain a complete litchi point cloud and is simple to operate, the sparse and dense reconstruction process of a single litchi is long in time consumption, high in equipment requirement, and the average reconstruction of the point cloud of a litchi and the extraction of the coloring rate are required to be 10 minutes, and the time required for extracting the coloring rate of the litchi by adopting the litchi coloring rate provided by the invention is less than 1mm, so that the overall speed is improved by 90%.
(3) Litchi volume extraction data analysis
In order to verify that the litchi detection result is verified by adopting the litchi depth image to carry out 3D modeling on the litchi and extracting the litchi volume extraction result, the experiment utilizes a manual volume measurement mode to verify the litchi detection result. Specifically, the volume measurement of litchi is carried out by adopting a drainage method in the experiment.
As shown in FIG. 14, after multiple experiments, the invention adopts the depth image of the litchi to carry out 3D modeling on the litchi, and extracts the predicted volume result and the manual measurement volume of the litchi volume to show obvious linear distribution, wherein R2 and RMSE of the predicted volume of the litchi are 0.8901 and 2.4733cm3, MAE is 1.59cm3 and MRE is 7.94 percent respectively.
According to the detection result, the 3D model established by the litchi depth image is utilized, the result of estimating the litchi volume by the pixel value thickness based on the litchi 3D model has better correlation with the manual measurement result, and the detection precision rate reaches 89%.
In summary, the three-dimensional point cloud of the litchi is obtained by acquiring RGB color images and depth images of the litchi under 4 different visual angles through mapping, and the litchi images and the point cloud are processed by utilizing image processing and a three-dimensional point cloud technology, so that nondestructive and rapid extraction of three phenotype parameters of mechanical injury, coloring rate and volume of the litchi is realized, the economic value of the litchi is improved, and scientific basis is provided for quality classification of the litchi.
Example 2:
As shown in fig. 15, a litchi phenotype parameter detection system based on image processing and three-dimensional point cloud is used for the litchi phenotype parameter detection method based on image processing and three-dimensional point cloud, and the system comprises a litchi image acquisition platform and a cloud server 6;
As shown in fig. 14, the litchi image acquisition platform comprises a camera bellows 1, an LED light source 2, a litchi placing plate 7, a depth camera 4 and a camera bracket 3, wherein the LED light source 2 is installed at the top in the box body of the camera bellows 1, the depth camera 4 is installed below the LED light source 2 through the camera bracket 3, the litchi placing plate 7 is installed at the bottom of the box body of the camera bellows 1, and the lens of the depth camera 4 faces the litchi placing plate 7; the litchi placing plate 7 is used for placing litchi to be detected, and the depth camera 4 is in communication connection with the cloud server 6;
the working flow of the litchi image acquisition platform is as follows:
during detection, firstly, the litchi to be detected is placed under the depth camera 4, the camera lens is kept to face the center of the litchi to be detected horizontally, then the litchi to be detected is uniformly rotated in a manual mode, depth images and RGB images of the left view, the right view, the overlook view and the look-up view of the litchi to be detected are obtained, at this time, the depth camera 4 processes the obtained depth images and RGB images of the four views to obtain original three-dimensional point cloud data of the litchi to be detected, and the depth images, the RGB images and the original three-dimensional point cloud data of the four views of the litchi to be detected are transmitted to the cloud server 6.
The cloud server 6 includes a processor and a memory; the memory has stored thereon non-transitory computer instructions which, when executed by the processor, perform the litchi phenotype parameter detection method based on image processing and three-dimensional point cloud as described in embodiment 1.
The depth camera 4 employs INTEL REALSENSE D depth cameras.
Specifically, INTEL REALSENSE D's 405 depth camera is a close-range depth camera, the frame rate is 90 frames/s, the depth range is 7cm-50cm, the depth field is 87 ° (horizontal) ×58 ° (vertical), the frame rate is 90 frames/s, the resolution of a color camera, namely RGB mode, is 1280×720, and the resolution output of the depth camera mode is 1280×720 at the highest.
In the description of the present invention, it should be noted that, unless explicitly stated and agreed otherwise, the terms "disposed," "mounted," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention can be understood by those of ordinary skill in the art according to the specific circumstances.
The above examples are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above examples, and any other changes, modifications, substitutions, combinations, and simplifications that do not depart from the spirit and principle of the present invention should be made in the equivalent manner, and the embodiments are included in the protection scope of the present invention.

Claims (10)

1.一种基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，包括以下步骤：1. A method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud, characterized in that it comprises the following steps: S1、获取待测荔枝多个视角下的二维图像，所述二维图像包括深度图像和RGB图像；S1, obtaining two-dimensional images of litchi to be tested under multiple viewing angles, wherein the two-dimensional images include a depth image and an RGB image; S2、将所述待测荔枝多个视角下的二维图像进行映射得到待测荔枝的原始三维点云数据，对所述原始三维点云数据进行背景分割和降噪滤波处理，得到待测荔枝的三维点云数据，计算所述待测荔枝的三维点云数据的点云曲率和法向量，统计出用于判断荔枝表面损失区域的曲率阈值，从而对荔枝损伤区域点云进行分割，识别到所述待测荔枝的机械伤区域；S2, mapping the two-dimensional images of the litchi to be tested under multiple viewing angles to obtain original three-dimensional point cloud data of the litchi to be tested, performing background segmentation and noise reduction filtering on the original three-dimensional point cloud data to obtain three-dimensional point cloud data of the litchi to be tested, calculating the point cloud curvature and normal vector of the three-dimensional point cloud data of the litchi to be tested, and calculating the curvature threshold for judging the surface loss area of the litchi, thereby segmenting the litchi damage area point cloud and identifying the mechanical damage area of the litchi to be tested; S3、对步骤S1得到的所述待测荔枝多个不同视角下的RGB图像转换至HSI颜色模型空间，得到多张不同视角的HSI模型下的荔枝图像，通过预设的红色像素点阈值提取出所述多张不同视角的HSI模型下的荔枝图像中各自的红色像素点并取平均值，将所述平均值作为待测荔枝的红色着色率；S3, converting the RGB images of the litchi to be tested under multiple different viewing angles obtained in step S1 into an HSI color model space to obtain multiple litchi images under the HSI model at different viewing angles, extracting the red pixels of each of the litchi images under the HSI model at different viewing angles by using a preset red pixel threshold, and taking an average value, and taking the average value as the red coloring rate of the litchi to be tested; S4、根据所述待测荔枝多个不同视角下的深度图像f(x,y)和背景深度图像b(x,y)得到包含了荔枝厚度的深度图像h(x,y)，对所述包含了荔枝厚度的深度图像h(x,y)进行预处理，采用预处理后的深度图像h(x,y)建立的待测荔枝顶部和底部3D模型，根据所述待测荔枝顶部和底部3D模型计算出所述待测荔枝的体积值。S4. Obtain a depth image h(x, y) including the thickness of the litchi according to the depth images f(x, y) of the litchi to be tested at multiple different viewing angles and the background depth image b(x, y), preprocess the depth image h(x, y) including the thickness of the litchi, and establish a 3D model of the top and bottom of the litchi to be tested using the preprocessed depth image h(x, y), and calculate the volume value of the litchi to be tested according to the 3D model of the top and bottom of the litchi to be tested. 2.根据权利要求1所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，步骤S1中所述多个视角包括左视、右视、俯视和仰视四个视角。2. The method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud according to claim 1, wherein the multiple viewing angles in step S1 include four viewing angles: left viewing, right viewing, top viewing and top viewing. 3.根据权利要求1所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，步骤S2中所述待测荔枝的三维点云数据的点云曲率和法向量的推导过程如下：3. The method for detecting phenotypic parameters of litchi based on image processing and three-dimensional point cloud according to claim 1, characterized in that the derivation process of the point cloud curvature and normal vector of the three-dimensional point cloud data of the litchi to be tested in step S2 is as follows: 由于点云中的每个点都有一个邻域点云，该邻域点云使用特定的表面来接近，某一点的曲率可以用该点及其邻域点拟合的局部曲面曲率来表示；因此以圆周率为中心点，在圆周率附近均匀选取k个点，二次曲面方程可表示为：Since each point in the point cloud has a neighborhood point cloud, which is approached by a specific surface, the curvature of a point can be expressed by the local surface curvature fitted by the point and its neighborhood points; therefore, with pi as the center point, k points are uniformly selected near pi, and the quadratic surface equation can be expressed as: z(x,y)＝ax2+bxy+cy2 z(x,y)＝ax 2 +bxy+cy 2 根据最小二乘原理，zi的平方和可表示为:According to the least squares principle, the sum of squares of z i can be expressed as: 由此可求出二次曲面方程的系数a、b、c的值：From this, we can find the values of coefficients a, b, and c of the quadratic surface equation: 则曲面上的曲线就可以表示为：Then the curve on the surface can be expressed as: r＝(x(t),y(t))r＝(x(t),y(t)) 点Pi处的单位法向量可以表示为：The unit normal vector at point Pi can be expressed as: 将Pi点处的法向量作为邻域内局部曲面的法向量，则邻域内各点的协方差矩阵为：Taking the normal vector at point Pi as the normal vector of the local surface in the neighborhood, the covariance matrix of each point in the neighborhood is: C·Xj＝λj·Xj，j＝1,2,3C·X j ＝λ j ·X j , j＝1,2,3 其中，P0为邻域点的质心，k为邻域点的个数，λj和Xj分别表示C的特征值和特征向量；Where P 0 is the centroid of the neighborhood points, k is the number of neighborhood points, λ j and X j represent the eigenvalue and eigenvector of C respectively; 矩阵C的最小特征值对应的特征向量是点(Pi的法向量，构造协方差矩阵，通过特征值分解计算矩阵的特征值，当特征值满足λ0≤λ1≤λ2条件时，邻域点的曲率可表示为：The eigenvector corresponding to the minimum eigenvalue of the matrix C is the normal vector of the point (P i. The covariance matrix is constructed, and the eigenvalues of the matrix are calculated by eigenvalue decomposition. When the eigenvalue satisfies the condition λ0≤λ1≤λ2, the curvature of the neighborhood point can be expressed as: 最终，邻域曲率ki即为邻域曲面的曲率程度。Finally, the neighborhood curvature k i is the degree of curvature of the neighborhood surface. 4.根据权利要求1所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，步骤S2中所述用于判断荔枝表面损失区域的曲率阈值为1.81。4. The method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud according to claim 1, characterized in that the curvature threshold for determining the litchi surface loss area in step S2 is 1.81. 5.根据权利要求1所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，步骤S3的具体过程如下：5. The method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud according to claim 1, wherein the specific process of step S3 is as follows: 首先将步骤S3中得到的多个不同视角下的RGB图像中R分量、G分量和B分量分别进行归一化处理，在RGB颜色空间中位于空间位置(x,y)的像素点的颜色用该像素点用R分量R(x,y)、G分量G(x,y)和B分量B(x,y)3个数值表示；然后将归一化处理后的RGB图像中的荔枝区域转换到HSI颜色空间，HSI颜色空间中像素点的H分量H(x,y)、S分量S(x,y)和I分量I(x,y)的表示如下：First, the R component, G component and B component of the RGB images under multiple different viewing angles obtained in step S3 are normalized respectively, and the color of the pixel at the spatial position (x, y) in the RGB color space is represented by three numerical values of the R component R(x, y), the G component G(x, y) and the B component B(x, y) of the pixel; then, the litchi area in the normalized RGB image is converted to the HSI color space, and the H component H(x, y), S component S(x, y) and I component I(x, y) of the pixel in the HSI color space are represented as follows: 其中， in, 最终根据上述公式得到所述待测荔枝多个不同视角下的RGB图像对应的HSI模型下的荔枝图像；Finally, according to the above formula, the litchi images under the HSI model corresponding to the RGB images of the litchi to be tested under multiple different viewing angles are obtained; 然后根据预设的通过预设的红色像素点阈值提取所述多张不同视角的HSI模型下的荔枝图像中各自的红色像素值并求平均值，将所述平均值作为待测荔枝的红色着色率；所述红色着色率为整个待测荔枝表面红色像素点与待测荔枝所有像素点总面积值的比值；所述红色像素点为HSI模型下的待测荔枝图像中荔枝表面的红色H分量。Then, according to a preset red pixel threshold, the red pixel values of each of the multiple litchi images under the HSI model at different viewing angles are extracted and the average value is calculated, and the average value is used as the red coloring rate of the litchi to be tested; the red coloring rate is the ratio of the red pixel points on the entire surface of the litchi to be tested to the total area value of all pixel points of the litchi to be tested; the red pixel point is the red H component of the litchi surface in the litchi image to be tested under the HSI model. 6.根据权利要求5所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，所述预设的红色像素点阈值的取值范围为35°-40°。6. The method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud according to claim 5, characterized in that the value range of the preset red pixel threshold is 35°-40°. 7.根据权利要求1所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，步骤S4中对所述包含了荔枝厚度的深度图像h(x,y)进行预处理方式为侵蚀、膨胀和高斯平滑操作。7. The method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud according to claim 1 is characterized in that in step S4, the depth image h(x, y) containing the thickness of litchi is preprocessed by erosion, dilation and Gaussian smoothing operations. 8.根据权利要求1所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，步骤S4中所述深度图像f(x,y)和背景深度图像b(x,y)的拍摄环境和拍摄高度相同。8. The method for detecting litchi phenotypic parameters based on image processing and three-dimensional point cloud according to claim 1, characterized in that the shooting environment and shooting height of the depth image f(x, y) and the background depth image b(x, y) in step S4 are the same. 9.一种基于图像处理和三维点云的荔枝表型参数检测系统，用于实现如权利要求1-8任一项所述的基于图像处理和三维点云的荔枝表型参数检测方法，其特征在于，所述系统包括荔枝图像获取平台和云服务器；9. A litchi phenotypic parameter detection system based on image processing and three-dimensional point cloud, used to implement the litchi phenotypic parameter detection method based on image processing and three-dimensional point cloud as described in any one of claims 1 to 8, characterized in that the system comprises a litchi image acquisition platform and a cloud server; 所述荔枝图像获取平台包括暗箱、LED光源、荔枝放置板、深度相机和相机支架，所述LED光源安装在所述暗箱的箱体内的顶部，所述深度相机通过相机支架安装在所述LED光源的下方，所述荔枝放置板安装在所述暗箱的箱体的底部，且所述深度相机的镜头朝向所述荔枝放置板；所述荔枝放置板用于放置待测荔枝，所述深度相机与所述云服务器通讯连接；The litchi image acquisition platform includes a dark box, an LED light source, a litchi placement board, a depth camera and a camera bracket, wherein the LED light source is installed at the top of the dark box, the depth camera is installed below the LED light source through the camera bracket, the litchi placement board is installed at the bottom of the dark box, and the lens of the depth camera faces the litchi placement board; the litchi placement board is used to place litchis to be tested, and the depth camera is connected to the cloud server for communication; 所述云服务器包括处理器和存储器；所述存储器上存储有非暂时性计算机指令，当所述非暂时性计算机指令被处理器运行时，执行如权利要求1-8任一项所述的基于图像处理和三维点云的荔枝表型参数检测方法。The cloud server includes a processor and a memory; the memory stores non-temporary computer instructions, and when the non-temporary computer instructions are executed by the processor, the litchi phenotypic parameter detection method based on image processing and three-dimensional point cloud as described in any one of claims 1-8 is executed. 10.根据权利要求9所述的基于图像处理和三维点云的荔枝表型参数检测系统，其特征在于，所述深度相机采用Intel Realsense D405深度相机。10. The litchi phenotypic parameter detection system based on image processing and three-dimensional point cloud according to claim 9, characterized in that the depth camera adopts an Intel Realsense D405 depth camera.
CN202410505174.7A 2024-04-25 2024-04-25 Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud Pending CN118446969A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410505174.7A CN118446969A (en) 2024-04-25 2024-04-25 Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410505174.7A CN118446969A (en) 2024-04-25 2024-04-25 Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud

Publications (1)

Publication Number Publication Date
CN118446969A true CN118446969A (en) 2024-08-06

Family

ID=92328822

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410505174.7A Pending CN118446969A (en) 2024-04-25 2024-04-25 Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud

Country Status (1)

Country Link
CN (1) CN118446969A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119049041A (en) * 2024-08-19 2024-11-29 深圳职业技术大学 Litchi nondestructive sorting method and device, storage medium and computer equipment

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119049041A (en) * 2024-08-19 2024-11-29 深圳职业技术大学 Litchi nondestructive sorting method and device, storage medium and computer equipment

Similar Documents

Publication Publication Date Title
CN112308832B (en) Bearing quality detection method based on machine vision
CN109285145B (en) A smartphone-based method for measuring the height of multiple standing trees
CN106709950B (en) Binocular vision-based inspection robot obstacle crossing wire positioning method
CN110210448B (en) Intelligent face skin aging degree identification and evaluation method
CN110443836A (en) A kind of point cloud data autoegistration method and device based on plane characteristic
CN109580630A (en) A kind of visible detection method of component of machine defect
CN109410139A (en) Inside a kind of historical relic and surface disease digital assay appraisal procedure
CN107392929B (en) An intelligent target detection and size measurement method based on human visual model
CN113705564B (en) A method for identifying and reading pointer instruments
CN103593840B (en) Method for detecting phenotype of Arabidopsis
CN116912691A (en) A remote sensing image change detection method based on multi-feature evidence fusion and structural similarity
CN109255785A (en) A kind of bearing open defect detection system
CN112241964B (en) Light strip center extraction method for line structured light non-contact measurement
CN118446969A (en) Litchi phenotype parameter detection method and system based on image processing and three-dimensional point cloud
CN108629771A (en) A kind of blind evaluation method of picture quality with scale robustness
CN104392209B (en) A kind of image complexity evaluation method of target and background
CN120765660A (en) Egg oiling quality detection method based on machine vision
CN119206099B (en) A construction site measurement system based on image processing
CN120721727A (en) A machine vision-based aluminum-plastic pharmaceutical packaging detection system and method
CN119205870A (en) A method and system for estimating the mass of underwater swimming fish
CN117739823A (en) Target ranging method, system and service robot
CN109084721B (en) Method and apparatus for determining a topographical parameter of a target structure in a semiconductor device
CN116645418A (en) 2D and 3D camera-based screen button detection method, device and related medium
CN107462221B (en) A method for automatic measurement of fisheye features
CN119444704B (en) A droplet detection method and system based on machine vision

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination