CN116786435B - Intelligent bullet sorting system - Google Patents

Intelligent bullet sorting system

Info

Publication number
CN116786435B
CN116786435B CN202310859853.XA CN202310859853A CN116786435B CN 116786435 B CN116786435 B CN 116786435B CN 202310859853 A CN202310859853 A CN 202310859853A CN 116786435 B CN116786435 B CN 116786435B
Authority
CN
China
Prior art keywords
image
rank
tray
bullet
bullets
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.)
Active
Application number
CN202310859853.XA
Other languages
Chinese (zh)
Other versions
CN116786435A (en
Inventor
叶振飞
王英利
刘大亮
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.)
BEIJING ZODNGOC AUTOMATIC TECHNOLOGY CO LTD
Original Assignee
BEIJING ZODNGOC AUTOMATIC TECHNOLOGY CO LTD
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 BEIJING ZODNGOC AUTOMATIC TECHNOLOGY CO LTD filed Critical BEIJING ZODNGOC AUTOMATIC TECHNOLOGY CO LTD
Priority to CN202310859853.XA priority Critical patent/CN116786435B/en
Publication of CN116786435A publication Critical patent/CN116786435A/en
Application granted granted Critical
Publication of CN116786435B publication Critical patent/CN116786435B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
    • B07C5/02—Measures preceding sorting, e.g. arranging articles in a stream orientating
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
    • B07C5/36—Sorting apparatus characterised by the means used for distribution
    • B07C5/361—Processing or control devices therefor, e.g. escort memory
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G06T5/30—Erosion or dilatation, e.g. thinning
    • 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
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
    • H04N1/024—Details of scanning heads ; Means for illuminating the original
    • H04N1/028—Details of scanning heads ; Means for illuminating the original for picture information pick-up
    • H04N1/03—Details of scanning heads ; Means for illuminating the original for picture information pick-up with photodetectors arranged in a substantially linear array

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

本发明涉及一种智能子弹分拣系统，包括送料传输部分、图像采集部分、智能识别检测部分、通信部分、控制部分、分拣执行部分，送料传输部分包括传送带、托盘和运动控制系统，图像采集部分包括线扫相机、镜头、光源系统、图像采集卡、旋转编码器，智能识别检测部分包括检测算法模块，检测算法模块用于对整个托盘视野范围内所有不同类型的子弹进行检测；通信部分包括数据库和通信模块，分拣执行部分包括机械手，通过控制机械手将不同类型的子弹抓取到相应位置来实现子弹的分拣。本发明能实现对子弹的在线、快速、准确实时检测分拣，可以自动对新检测到的子弹型号进行模板新建，并进行数据统计和分析。

This invention relates to an intelligent bullet sorting system, comprising a feeding and conveying section, an image acquisition section, an intelligent recognition and detection section, a communication section, a control section, and a sorting execution section. The feeding and conveying section includes a conveyor belt, a tray, and a motion control system. The image acquisition section includes a line scan camera, a lens, a light source system, an image acquisition card, and a rotary encoder. The intelligent recognition and detection section includes a detection algorithm module used to detect all different types of bullets within the entire field of view of the tray. The communication section includes a database and a communication module. The sorting execution section includes a robotic arm, which is controlled to grasp different types of bullets and place them in corresponding positions to achieve bullet sorting. This invention enables online, rapid, accurate, and real-time detection and sorting of bullets. It can automatically create templates for newly detected bullet models and perform data statistics and analysis.

Description

Intelligent bullet sorting system
Technical Field
The invention relates to an intelligent bullet sorting system, and belongs to the technical field of bullet sorting.
Background
In recent years, machine vision inspection technology has begun to play an important role in the inspection of many products. An important link of modern construction of army is recycling of bullets, however, due to the fact that types of bullets are different due to the fact that weapons are numerous, current bullet recycling classification basically uses a manual visual inspection method to conduct simple classification. The method has high labor intensity, causes visual fatigue to have wrong classification and unstable detection results, has the detection results related to the experience of workers and is very difficult to carry out process control and statistical analysis, restricts the automatic recovery classification of bullets and does not meet the requirements of industrialization and automation. In modern bullet recycling, the military production unit is in urgent need of an efficient and intelligent system to replace manpower.
Based on this, the present invention has been proposed.
Disclosure of Invention
Aiming at the defects existing in the prior art, the invention provides an intelligent bullet sorting system, which has the following specific technical scheme:
An intelligent bullet sorting system comprising:
A feed transport section including a conveyor belt, a tray, and a motion control system, in which different types of cartridges are placed in the tray, the motion of the conveyor belt being controlled by the motion control system;
The image acquisition part comprises a line scanning camera, a lens, a light source system, an image acquisition card and a rotary encoder, the tray is transmitted to the image acquisition part through a conveyor belt, and the image acquisition part acquires an original image of the tray filled with bullets;
the intelligent recognition detection part comprises a detection algorithm module, wherein the detection algorithm module is used for detecting all different types of bullets in the whole visual field range of the tray;
a communication part including a database for storing bullet types, position information, and detection results in real time and a communication module;
a control part for communicating the bullet model, the position information and the detection result to the control part through a communication module;
and the sorting execution part comprises a manipulator, the control part receives the detection result and makes corresponding instructions to the manipulator, and sorting of the bullets is realized by controlling the manipulator to grab bullets of different types to corresponding positions.
In a further development, the detection algorithm module comprises the following steps:
let the detected size be S, the diameter of the bullet be R, the distance from the top of the bullet to the line scanning camera be L, the focal length of the line scanning camera be f, the number of pixels be N, then
The running speed of the conveyor belt and the line frequency of the line scanning camera are matched according to a matching formula, wherein the matching formula is as follows:
L w is the length of the tray along the running direction of the conveyor belt, L h is the length of the tray perpendicular to the running direction of the conveyor belt, P w is the number of single-sided pixels of the line scanning camera, V is the running speed of the conveyor belt, t is the time required for the tray to enter the field of view of the line scanning camera and leave the field of view of the line scanning camera, R is the radius of the rotary encoder, B is the number of pulses which can be sent out by one circle of the rotary encoder, and lambda is the frequency doubling number which needs to be set by manual adjustment;
the algorithm in the detection algorithm module comprises the following steps:
firstly, placing an empty tray on a conveyor belt to run, and adjusting parameters of lambda to enable an image acquired by a line scanning camera to be a complete unstretched image;
Step two, respectively acquiring an Image1 of an empty tray and an Image2 of a tray with random bullets, and performing Image processing on the Image1 and the Image 2:
Firstly, performing spatial nonlinear filtering on an Image1 and an Image2, using a matrix A with a Rank m×Rankn size to slide through each pixel point of an acquired Image with a Rank p×Rankq from left to right from top to bottom, obtaining brightness values of all the pixel points, converting the brightness values into column vectors by calling a function fun1, generating a Rank m·Rankn -dimensional vector by the matrix A every time the pixel points pass through, generating Rank p·Rankq such vectors by the Image with a Rank p×Rankq size, further obtaining a Rank m·Rankn·Rankp·Rankq matrix, substituting each column of the matrix into the function fun2 to operate, generating a Rank p·Rankq -dimensional vector, and then re-reducing the Rank p·Rankq -dimensional vector into a Rank p×Rankq matrix by inverse operation of the function fun1 to complete operation;
Image enhancement is performed on Image Iamge and Image2,
I is the component value of the pixel of the original image, I * is the component value corresponding to the changed image, and n is the minimum gray value of the bullet;
an improved rapid self-adaptive image binarization method is introduced to carry out binarization processing on the image:
G (x, y) is the gray value of the current pixel point (x, y), S (x, y) is the sum of the gray values of all pixels at the upper left of the current pixel point (x, y), x represents the coordinate of the current pixel point in the x direction in the coordinate system, and y represents the coordinate of the current pixel point in the y direction in the coordinate system;
Traversing all pixels by S (x, y) =G (x, y) +S (x-1, y) +S (x, y-1) -S (x-1, y-1), calculating the average gray value of a rectangle by taking the pixels as the center, comparing with the current pixels, and further performing Image segmentation to obtain a segmented Image11 and an Image21;
Step three, carrying out contour extraction on the Image11 through Hough circle detection, extracting circular areas in which 6*8 bullets are placed on a tray, carrying out corrosion expansion operation on each circular area to obtain a detection area ROI i, taking the upper left corner of the tray as an origin position, wherein the positions of the circular areas are Cir i (x, y), i=1, 2, 48, i is an integer;
Fourthly, translational correction is carried out on the Image21 by taking Cir i (x, y) as a template, and then the Image in the detection area is obtained For imagesModel classification is carried out to obtainJ=1, 2, & gt.t, j is an integer, t is a category, and then the initial image of the j-category bullet is obtained by performing feature extraction operation
For a pair ofDecomposing to obtain each scale coefficientK=1, 2,3,4,5, pairThreshold processing is carried out according to the function F to obtain F (-), then all F (-) of the category are reconstructed and weighted and fused to obtain an initial image of j category bullets
Step five, after the initial images of the bullets of various types are collected, detecting and collecting the bullets on the new tray in real time, repeating the operations of the step two to the step four to obtain a current real-time image I current, and collecting the initial images of the bullets of class jThe data of each pixel point is subjected to a function with the current real-time image I current The method comprises the steps of obtaining expected information E j, judging the expected information E j according to the similarity, judging that the result P (j) is 1 and is represented as the existing model, directly returning coordinate information of the existing model, automatically creating a template of the new model and prompting a user;
col is the column of the image, row is the high of the image;
Gamma is a set similarity threshold set to 0.8.
In a further improvement, when the maximum diameter of the bullet to be detected is 20mm, R max is 20mm, the maximum accuracy S max required for detection is 0.03mm, the maximum pixel number N max = 666.7, the resolution of the line scan camera is at least 667×667, the minimum resolution of the line scan camera is 1334×1334, and the minimum resolution of the narrow side of the line scan camera is 8004.
The light source system comprises an open-pore backlight source positioned right above the tray, parallel line light sources which are axisymmetric and positioned at the left side and the right side above the tray, wherein the left side and the right side of the parallel line light sources are arranged in a correlation way, the included angles between the two parallel line light sources and the plane of the tray are 45 degrees, and the light source system is cooled in a stroboscopic and water-cooling dual mode.
Further improvement, the image acquisition card is used for image acquisition and transmission; the rotary encoder is used for providing acquisition signals for the line scanning camera.
The invention has the beneficial effects that:
1. The invention belongs to the application of machine vision in the field of military product detection, and realizes the online, rapid and accurate real-time detection and sorting of bullets by carrying out image acquisition, original image processing and feature extraction and discrimination on the bullets.
2. The system can automatically carry out template new construction on the newly detected bullet model, and carry out data statistics and analysis. In the algorithm, a large amount of operations are performed on the matrix, so that the algorithm is suitable for parallel acceleration operation, and the problem of real-time mass data can be solved.
3. The invention can replace manual work to recycle and sort the bullets, liberate manpower, improve efficiency and improve safety.
4. The intelligent type matching function is realized, bullet images of different types are collected, processed and stored, and then the bullet images of corresponding types can be automatically called without matching again.
5. The method has robustness, and can be used for completing the first installation and debugging for different transmission lines at different speeds without multiple debugging.
Drawings
FIG. 1 is a block diagram of an intelligent bullet sorting system according to the present invention;
FIG. 2 is a schematic view of a pallet according to the present invention;
FIG. 3 is a schematic block diagram of a light source system according to the present invention;
FIG. 4 is a flow chart of a detection algorithm module;
FIG. 5 is an original view of a bullet;
FIG. 6 is a contour diagram of a bullet after being processed using a global threshold segmentation algorithm;
FIG. 7 is a contour diagram of a bullet after being processed by the Canny algorithm;
fig. 8 is a profile view of a process employing the algorithm of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Example 1
As shown in fig. 1, the intelligent bullet sorting system includes:
And a feeding and conveying part including a conveyor belt, a tray, and a motion control system, wherein different types of bullets are placed in the tray, and the motion of the conveyor belt is controlled by the motion control system, and the feeding and conveying part can place the different types of bullets in the tray of 600 x 800mm shown in fig. 2, and convey the tray to the image acquisition part through the conveyor belt.
The image acquisition part comprises a line scanning camera, a lens, a light source system, an image acquisition card and a rotary encoder, the tray is transmitted to the image acquisition part through a conveyor belt, and the image acquisition part can acquire high-quality and high-contrast original images of the tray filled with bullets.
Because the tray size is 600 x 800mm, and there are the bullet of multiple model probably to have on the same tray, so select the line to sweep the camera and gather the image, this line sweep the camera and can avoid because the bullet highly inconsistent production that leads to the covering phenomenon in the image formation of image, and utilize it to gather the image of fixed line at every turn and splice realization to the collection of whole picture.
The image acquisition card is used for image acquisition and transmission, the data volume of the line scanning camera is too large, the common gigabit network cannot meet the real-time transmission requirement, and the data can be acquired and transmitted without loss through the image acquisition card.
The rotary encoder is used for providing acquisition signals for the line scanning camera. In the same period, the rotary encoder gives up how many pulses, and the line scan camera will acquire how many 8192 x1 images.
The light source system comprises an open-pore backlight source positioned right above the tray, parallel line light sources which are axially symmetrical and positioned at the left side and the right side above the tray, wherein the left side and the right side of the parallel line light sources are arranged in a correlation way, the included angles between the two parallel line light sources and the plane of the tray are 45 degrees, and the light source system is cooled in a stroboscopic and water-cooling dual mode.
Because the light source is angled to illuminate the surface of the tray, the light source emits parallel light, and the distance from the nearest side to the farthest side of the light source is different, so that the illumination intensity per unit area is uneven. The invention adopts the two parallel line light sources on the left and right to be arranged in opposite directions, and the illumination intensity of the middle area can be ensured to be consistent by adopting a specific angle.
The intelligent recognition detection part comprises a detection algorithm module, wherein the detection algorithm module is used for detecting all different types of bullets in the whole visual field range of the tray.
A communication part including a database for storing bullet types, position information, and detection results in real time and a communication module;
a control part for communicating the bullet model, the position information and the detection result to the control part through a communication module;
and the sorting execution part comprises a manipulator, the control part receives the detection result and makes corresponding instructions to the manipulator, and sorting of the bullets is realized by controlling the manipulator to grab bullets of different types to corresponding positions.
Example 2
The detection algorithm module comprises the following processes:
let the detected size be S, the diameter of the bullet be R, the distance from the top of the bullet to the line scanning camera be L, the focal length of the line scanning camera be f, the number of pixels be N, then If the maximum diameter of the bullet to be detected is 20mm, R max is 20mm, the required precision S max is 0.03mm, N max = 666.7, the resolution of the camera is at least 667 x 667 calculated according to the precision, in order to keep complete original information, more than 2 pixels are needed to be imaged, the minimum resolution of the camera is 1334 x 1334 calculated according to 2 pixels, and in view of the fact that the bullets arranged in a 6*8 mode are arranged, the minimum resolution of the narrow side of the camera is 8004, a Haikang 8k line scanning camera is selected, the resolution of which is 8192 x M (M is the number of spliced lines), and the requirement of the resolution of bullet detection can be met.
The accuracy of the detection result is determined by the quality of the image acquired by the camera to a great extent, the model size of the bullet is easy to judge as the acquired image contrast is higher, and the metal reflection phenomenon occurs on the surface of the bullet, so that the surface characteristics of the bullet are highlighted in order to realize large-area uniform lighting of the whole tray, a light source system as shown in fig. 3 is designed, an open pore backlight source is used for integrally improving illumination brightness, left and right opposite parallel line light sources are used for reducing reflection interference, and a stroboscopic and water cooling dual mode is adopted for reducing the temperature of the light source on site so as to ensure that the temperature of the light source on site is not too high.
Through field verification, the number of lines in practical application is 8192×11000, and a 600×800 tray can be completely collected. One image of the line scanning camera is 8192 x 1, and stitching 11000 images together is 8192 x 11000, namely the number of signals to be acquired.
The specific implementation steps of the algorithm are as follows:
Since the line scan camera needs to have relative motion with the object to present a complete image, and in order for the image not to be stretched or compressed, it is necessary that the speed and line frequency of the camera be matched. The running speed of the conveyor belt and the line frequency of the line scanning camera are matched according to a matching formula, wherein the matching formula is as follows:
L w is the length of the tray along the running direction of the conveyor belt, L h is the length of the tray perpendicular to the running direction of the conveyor belt, P w is the number of single-sided pixels of the line scanning camera, V is the running speed of the conveyor belt, t is the time required for the tray to enter the view of the line scanning camera and leave the view of the line scanning camera, R is the radius of the rotary encoder, B is the number of pulses which can be sent out by one circle of the rotary encoder, and lambda is the frequency doubling number which needs to be adjusted manually.
As shown in fig. 4, the algorithm in the detection algorithm module includes the following steps:
Step one, placing an empty tray on a conveyor belt to run, and adjusting parameters of lambda to enable an image acquired by a line scanning camera to be a complete unstretched image.
Step two, respectively acquiring an Image1 of an empty tray and an Image2 of a tray with random bullets, and performing Image processing on the Image1 and the Image 2:
firstly, performing spatial nonlinear filtering on an Image1 and an Image2, sliding each pixel point of an acquired Image with a size of Rank p×Rankq from top to bottom by using a matrix A with a size of Rank m×Rankn from left to right, obtaining brightness values of all the pixel points, converting the brightness values into column vectors by calling a function fun1, generating a Rank m·Rankn -dimensional vector by the matrix A every time the matrix A passes through one pixel point, generating Rank p·Rankq such vectors by the Image with the size of Rank p×Rankq, further obtaining a Rank m·Rankn·Rankp·Rankq matrix, substituting each column of the matrix into the function fun2, calculating to generate a Rank p·Rankq -dimensional vector, and then re-reducing the Rank p·Rankq -dimensional vector into a Rank p×Rankq matrix by inverse operation of the function fun1, thereby completing the operation. Wherein Rank m、Rankn is a row and a column of the matrix, respectively, and Rank p、Rankq is a row and a column of the matrix, respectively.
Since the gray dynamic range is narrow and the change is not obvious from the acquired bullet picture, and the bullet picture belongs to a low-contrast Image, the Image Iamge and the Image2 are subjected to Image enhancement:
i is the component value of the pixel of the original image, I * is the component value corresponding to the changed image, and n is the minimum gray value of the bullet;
an improved rapid self-adaptive image binarization method is introduced to carry out binarization processing on the image:
G (x, y) is the gray value of the current pixel point (x, y), S (x, y) is the sum of the gray values of all pixels at the upper left of the current pixel point (x, y), x represents the coordinate of the current pixel point in the x direction in the coordinate system, and y represents the coordinate of the current pixel point in the y direction in the coordinate system;
all pixels are traversed through S (x, y) =G (x, y) +S (x-1, y) +S (x, y-1) -S (x-1, y-1), then the average gray value of a rectangle is calculated by taking the pixels as the center, the average gray value is compared with the current pixels, and then Image segmentation is carried out, so that an Image11 and an Image21 after segmentation processing are obtained.
Step three, performing contour extraction on the Image11 through Hough circle detection, extracting circular areas in which 6*8 bullets are placed on the tray, performing corrosion expansion operation on each circular area to obtain a detection area ROI i, taking the upper left corner of the tray as an origin position, wherein the positions of the circular areas are Cir i (x, y), i=1, 2, 48 and i are integers.
Fourthly, translational correction is carried out on the Image21 by taking Cir i (x, y) as a template, and then the Image in the detection area is obtainedFor imagesModel classification is carried out to obtainJ=1, 2, & gt.t, j is an integer, t is a category, and then the initial image of the j-category bullet is obtained by performing feature extraction operation
For a pair ofDecomposing to obtain each scale coefficientK=1, 2,3,4,5, pairThreshold processing is carried out according to the function F to obtain F (-), then all F (-) of the category are reconstructed and weighted and fused to obtain an initial image of j category bullets
Step five, after the initial images of the bullets of various types are collected, detecting and collecting the bullets on the new tray in real time, repeating the operations of the step two to the step four to obtain a current real-time image I current, and collecting the initial images of the bullets of class jThe data of each pixel point is subjected to a function with the current real-time image I current The method comprises the steps of obtaining expected information E j, judging the expected information E j according to the similarity, judging that the result P (j) is 1 and is represented as the existing model, directly returning coordinate information of the existing model, automatically creating a template of the new model and prompting a user;
col is the column of the image, row is the high of the image;
Gamma is a set similarity threshold set to 0.8.
In the invention, the Image Iamge and the Image2 are subjected to Image enhancement by adopting a specific algorithm, because the acquired bullet Image has a narrow gray dynamic range and insignificant change, and belongs to a low-contrast Image, the gray value range of the bullet Image is expanded by setting the minimum gray value in the bullet Image as 0 and setting the part symmetrical to the minimum gray value 255 as 255, so that the gray contrast of the bullet Image can be improved, and the subsequent contour extraction is convenient.
Other image enhancement algorithms, such as:
1. image enhancement based on gamma transformation is low for image contrast and the image enhancement effect is obvious in the case of high overall brightness value (overexposure for camera).
2. The image enhancement based on the object Log transformation can expand the low gray value part of the image, display more details of the low gray value part, compress the high gray value part of the image, and reduce the details of the high gray value part, so that the aim of emphasizing the low gray value part of the image is fulfilled.
3. The gray level image of the original image is uniformly distributed in the whole gray level space from a gray level interval in a comparison set based on the image enhancement of histogram equalization, so that the nonlinear stretching of the image is realized, and the pixel values of the image are redistributed.
None of these can distinguish the grey scale of the bullet pictures significantly.
In the invention, the fourth and fifth steps can improve the bullet classifying effect. If in step four, the method of least square fitting the circle to the image ROI i * finds the largest circle, the existing model directly returns coordinate information, the new model automatically creates the template and prompts the user, then the method simply classifies the circles at the outermost periphery of the bullets, and there is a phenomenon that bullets with the same outer diameter and different inner diameters are classified into one category. The invention can accurately classify the bullets by extracting the characteristics in multiple dimensions through the present step four.
Gamma is set to a value that is measured over a large number of actual measurements and is continuously adjusted. If the gamma setting is too small, for example 0.5, it may lead to a partial bullet picture classification error, such as the fact that a-type bullets are classified into b-type bullets, and the occurrence probability of this situation is geometrically increased. If the gamma setting is too large, such as 0.9, a part of bullet pictures cannot find the built category, and a new category is automatically generated, so that the effect of actual accurate classification cannot be achieved.
In step four, pairDecomposing to obtain each scale coefficientK=1, 2,3,4,5, k maximum is 5, when k is too small, the extracted features are insufficient, such as k=3, only edge information can be extracted, when k is too large, calculation is exponentially increased, calculation time is too long, such as when k=7, detection time is 20 times of that when k=5, and detection performance is affected.
Fig. 5 is an original view of a bullet, and fig. 6 is a contour diagram of a bullet after being processed by a global threshold segmentation algorithm (a segmentation threshold is set according to a gray histogram of an image, then gray values of pixel points with gray values lower than the segmentation threshold in the image are set to 0, and the rest are set to 255). Fig. 7 is a contour map of a bullet after contour extraction by Canny algorithm (firstly, gaussian blur is performed to remove noise in an image, then gradient amplitude and direction of the image are calculated, and finally, edge contour is obtained by non-maximum suppression). Fig. 8 is a profile view of a process employing the algorithm of the present invention. As can be seen from FIGS. 5-8, conventional threshold segmentation cannot well filter out interference points in an image, conventional edge extraction can lose some profile information, and the method can remove the interference points in the image to the greatest extent possible on the basis of retaining the profile.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.

Claims (4)

1.一种智能子弹分拣系统，其特征在于，包括：1. An intelligent bullet sorting system, characterized in that it comprises: 送料传输部分，其包括传送带、托盘和运动控制系统，将不同类型的子弹放置在托盘当中，通过运动控制系统来控制传送带的运动；The feeding and conveying section includes a conveyor belt, a tray, and a motion control system. Different types of bullets are placed in the tray, and the movement of the conveyor belt is controlled by the motion control system. 图像采集部分，其包括线扫相机、镜头、光源系统、图像采集卡、旋转编码器，通过传送带将托盘传输到图像采集部分，图像采集部分采集出装有子弹的托盘原始图像；The image acquisition section includes a line scan camera, lens, light source system, image acquisition card, and rotary encoder. The tray is transported to the image acquisition section via a conveyor belt, and the image acquisition section acquires the original image of the tray containing bullets. 智能识别检测部分，其包括检测算法模块，所述检测算法模块用于对整个托盘视野范围内所有不同类型的子弹进行检测；The intelligent identification and detection section includes a detection algorithm module, which is used to detect all different types of bullets within the entire field of view of the tray; 通信部分，其包括数据库和通信模块，所述数据库用于实时存储子弹型号、位置信息和检测结果；The communication section includes a database and a communication module, wherein the database is used to store bullet type, location information and detection results in real time; 控制部分，子弹型号、位置信息和检测结果通过通信模块通信到控制部分；In the control section, bullet type, location information, and detection results are communicated to the control section via a communication module; 分拣执行部分，其包括机械手，控制部分接收检测结果并做出相应指令给机械手，通过控制机械手将不同类型的子弹抓取到相应位置来实现子弹的分拣；The sorting execution part includes a robotic arm. The control part receives the detection results and issues corresponding instructions to the robotic arm. By controlling the robotic arm to grab different types of bullets to the corresponding positions, the bullets are sorted. 所述检测算法模块包括以下过程：The detection algorithm module includes the following processes: 设检测的尺寸为S，子弹的直径为R，子弹顶部到线扫相机的距离为L，线扫相机的焦距为f，像素数为N，则 Let the size of the detection be S, the diameter of the bullet be R, the distance from the top of the bullet to the line scan camera be L, the focal length of the line scan camera be f, and the number of pixels be N, then 传送带运行的速度和线扫相机的行频按照匹配公式进行匹配，匹配公式为：The speed of the conveyor belt and the line frequency of the line scan camera are matched according to a matching formula, which is: Lw为托盘沿传送带运行方向的长度，Lh为托盘垂直于传送带运行方向的长度，Pw为线扫相机单边的像素数，V为传送带运行的速度，t为托盘从进入线扫相机视野到离开线扫相机视野所需的时间，R为旋转编码器的半径，B为旋转编码器一圈发出的脉冲数，λ为需要人工调节设置的倍频数； Lw is the length of the tray along the direction of the conveyor belt, Lh is the length of the tray perpendicular to the direction of the conveyor belt, Pw is the number of pixels on one side of the line scan camera, V is the speed of the conveyor belt, t is the time required for the tray to go from entering the field of view of the line scan camera to leaving the field of view of the line scan camera, R is the radius of the rotary encoder, B is the number of pulses emitted by the rotary encoder in one revolution, and λ is the frequency multiplier that needs to be manually adjusted. 所述检测算法模块中的算法包括以下步骤：The algorithm in the detection algorithm module includes the following steps: 步骤一、将空的托盘放置传送带上运行，调节λ的参数使得线扫相机采集出来的图像为完整未拉伸的图像；Step 1: Place the empty tray on the conveyor belt and run it. Adjust the parameter of λ so that the image captured by the line scan camera is a complete and unstretched image. 步骤二、分别采集空的托盘的图像Image1和放有随机装填子弹的托盘的图像Image2，对图像Image1和图像Image2进行图像处理：Step 2: Acquire images Image1 (empty tray) and Image2 (tray containing randomly loaded bullets), and perform image processing on Images Image1 and Image2 respectively. 首先，对图像Image1和图像Image2进行空域非线性滤波，使用Rankm×Rankn大小的矩阵A从上到下、从左至右一次滑过采集到的大小为Rankp×Rankq的图像的每一个像素点，将所有像素点的亮度值获得出来；通过调用函数fun1，将之转换为列向量，从而矩阵A每经过一个像素点就会产生一个Rankm·Rankn维度的向量，大小为Rankp×Rankq的图像就会产生Rankp·Rankq个这样的向量，进而得到一个Rankm·Rankn●Rankp●Rankq的矩阵；对该矩阵每一列代入函数fun2中运算，产生一个Rankp●Rankq维的向量，再通过函数fun1的逆运算将Rankp●Rankq维的向量重新还原为Rankp×Rankq的矩阵，完成运算；First, spatial nonlinear filtering is performed on images Image1 and Image2. A matrix A of size Rank m × Rank n is used to slide across each pixel of the acquired image of size Rank p × Rank q from top to bottom and from left to right, obtaining the brightness value of all pixels. By calling function fun1, it is converted into a column vector. Thus, matrix A generates a vector of dimension Rank m · Rank n for each pixel it passes through. The image of size Rank p × Rank q will generate Rank p · Rank q such vectors, resulting in a matrix of size Rank m · Rank n ● Rank p ● Rank q . Substituting each column of this matrix into function fun2, a vector of dimension Rank p ● Rank q is generated. Then, the inverse operation of function fun1 is used to restore the vector of dimension Rank p ● Rank q back into a matrix of dimension Rank p × Rank q , completing the operation. 对图像Iamge1和图像Image2进行图像增强，Image enhancement is performed on images Image1 and Image2. I为原图像的像素的分量值，I*为改变后的图像对应的分量值，n为子弹最小的灰度值；I represents the pixel component value of the original image, I * represents the corresponding component value of the modified image, and n represents the minimum grayscale value of the bullet. 引入改进的快速自适应图像二值化方法对图像进行二值化处理：An improved fast adaptive image binarization method is introduced to perform image binarization processing: G(x,y)为当前像素点(x,y)的灰度值，S(x,y)为当前像素点(x,y)左上方所有像素的灰度值和，x表示当前像素点在坐标系中x方向的坐标，y表示当前像素点在坐标系中y方向的坐标；G(x,y) is the gray value of the current pixel (x,y), S(x,y) is the sum of the gray values of all pixels to the left and above the current pixel (x,y), x represents the coordinate of the current pixel in the x-direction in the coordinate system, and y represents the coordinate of the current pixel in the y-direction in the coordinate system. 通过S(x,y)＝G(x,y)+S(x-1,y)+S(x,y-1)-S(x-1,y-1)对所有像素进行遍历，然后以这些像素为中心，计算矩形的平均灰度值，同当前像素比较，进而进行图像分割，得到分割处理后的图像Image11，图像Image21；By iterating through all pixels using S(x,y)＝G(x,y)+S(x-1,y)+S(x,y-1)-S(x-1,y-1), and then using these pixels as the center, the average gray value of the rectangle is calculated and compared with the current pixel to perform image segmentation, resulting in the segmented images Image11 and Image21. 步骤三、对图像Image11通过霍夫圆检测进行轮廓提取，将托盘上的6*8个子弹放置的圆形区域提取出来，对每个圆形区域进行腐蚀膨胀运算得出检测区域ROIi，以托盘左上角为原点位置，各个圆形区域位置为Ciri(x,y)，i＝1,2,...48，i为整数；Step 3: Extract the contour of Image11 using Hough circle detection to extract the circular area where the 6*8 bullets on the tray are placed. Perform erosion and dilation operations on each circular area to obtain the detection region ROI i . With the upper left corner of the tray as the origin, the position of each circular area is Cir i (x,y), i＝1,2,...48, where i is an integer. 步骤四、以Ciri(x,y)为模板，对图像Image21进行平移矫正后，得到检测区域中的图像对图像进行型号分类得到j＝1,2,…t，j为整数，t为类别；接着进行特征提取运算得到j类子弹的初始图像 Step 4: Using Cir i (x,y) as a template, perform translation correction on image Image21 to obtain the image in the detection region. For images Model classification j = 1, 2, ..., t, where j is an integer and t is the class; then feature extraction is performed to obtain the initial image of bullet class j. 对进行分解得到各尺度系数k＝1,2,3,4,5，对依据函数F进行阈值处理得到F(·)，接着对该类别的所有F(·)进行重构并加以加权融合得到j类子弹的初始图像 right Decomposition yields the scale coefficients. k = 1, 2, 3, 4, 5, for Thresholding is performed based on function F to obtain F(·). Then, all F(·) of this category are reconstructed and weighted to obtain the initial image of bullet type j. 步骤五、采集完各种类的子弹的初始图像之后，对新来的托盘上的子弹进行实时检测采集，重复步骤二至步骤四的操作，得到当前实时图像Icurrent，将j类子弹的初始图像各像素点的数据同当前实时图像Icurrent进行函数的操作，得到期望信息Ej，对期望信息Ej依据相似度进行判定，判定结果P(j)为1表示为已有型号，已有型号的直接返回坐标信息，新型号的自动创建模板并提示用户；Step 5: After acquiring the initial images of various types of bullets, perform real-time detection and acquisition on the newly arrived bullets on the tray. Repeat steps 2 to 4 to obtain the current real-time image I_current . Then, acquire the initial image of bullet type j. The data of each pixel is used in conjunction with the current real-time image I_current. The operation obtains the expected information Ej . The expected information Ej is judged based on similarity. The judgment result P(j) is 1, which means that it is an existing model. For existing models, the coordinate information is returned directly. For new models, a template is automatically created and the user is prompted. col为图像的列，row为图像的高；col is the column of the image, and row is the height of the image; γ为设定的相似度阈值，设置为0.8。γ is the set similarity threshold, which is set to 0.8. 2.根据权利要求1所述的一种智能子弹分拣系统，其特征在于：当所要检测的子弹最大直径为20mm，则Rmax为20mm，检测所需最大精度Smax为0.03mm，则最大像素数Nmax＝666.7，线扫相机的分辨率至少为667*667，线扫相机的最小分辨率为1334*1334，线扫相机窄边的最小分辨率为8004。2. The intelligent bullet sorting system according to claim 1, characterized in that: when the maximum diameter of the bullet to be detected is 20mm, then Rmax is 20mm, the maximum required detection accuracy Smax is 0.03mm, then the maximum number of pixels Nmax = 666.7, the resolution of the line scan camera is at least 667*667, the minimum resolution of the line scan camera is 1334*1334, and the minimum resolution of the narrow edge of the line scan camera is 8004. 3.根据权利要求1所述的一种智能子弹分拣系统，其特征在于：所述光源系统包括位于托盘正上方的开孔背光源、位于托盘上方左右两侧呈轴对称的平行线光源，左右两个平行线光源呈对射设置且两个平行线光源与托盘平面的夹角均为45°；采用频闪加水冷双重方式对光源系统进行降温。3. The intelligent bullet sorting system according to claim 1, characterized in that: the light source system includes an open backlight located directly above the tray, and parallel line light sources located on the left and right sides above the tray in an axisymmetric manner, the two parallel line light sources are arranged facing each other and the angle between the two parallel line light sources and the tray plane is 45°; the light source system is cooled by a combination of strobe and water cooling. 4.根据权利要求1所述的一种智能子弹分拣系统，其特征在于：所述图像采集卡用于图像采集、传输；所述旋转编码器用于给线扫相机提供采集信号。4. The intelligent bullet sorting system according to claim 1, characterized in that: the image acquisition card is used for image acquisition and transmission; the rotary encoder is used to provide acquisition signals to the line scan camera.
CN202310859853.XA 2023-07-13 2023-07-13 Intelligent bullet sorting system Active CN116786435B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310859853.XA CN116786435B (en) 2023-07-13 2023-07-13 Intelligent bullet sorting system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310859853.XA CN116786435B (en) 2023-07-13 2023-07-13 Intelligent bullet sorting system

Publications (2)

Publication Number Publication Date
CN116786435A CN116786435A (en) 2023-09-22
CN116786435B true CN116786435B (en) 2025-11-21

Family

ID=88038510

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310859853.XA Active CN116786435B (en) 2023-07-13 2023-07-13 Intelligent bullet sorting system

Country Status (1)

Country Link
CN (1) CN116786435B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111833385A (en) * 2020-07-20 2020-10-27 山东科技大学 Imaging system and imaging method of intelligent detection device for vision technology
WO2023102952A1 (en) * 2021-12-06 2023-06-15 浙江大学台州研究院 Coiled material detection device, system and method capable of achieving real-time detection

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9424634B2 (en) * 2004-03-04 2016-08-23 Cybernet Systems Corporation Machine vision system for identifying and sorting projectiles and other objects
CN104122271A (en) * 2014-07-09 2014-10-29 宁波摩视光电科技有限公司 Automated optical inspection (AOI)-based bullet apparent defect detection method
CN209125117U (en) * 2018-09-26 2019-07-19 上海集沙成塔科技股份有限公司 Bullet mark sorting device
CN210146468U (en) * 2019-04-19 2020-03-17 云南安宁化工厂有限公司 Multifunctional bullet quality inspection equipment
CN113643280B (en) * 2021-08-30 2023-09-22 燕山大学 A plate sorting system and method based on computer vision

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111833385A (en) * 2020-07-20 2020-10-27 山东科技大学 Imaging system and imaging method of intelligent detection device for vision technology
WO2023102952A1 (en) * 2021-12-06 2023-06-15 浙江大学台州研究院 Coiled material detection device, system and method capable of achieving real-time detection

Also Published As

Publication number Publication date
CN116786435A (en) 2023-09-22

Similar Documents

Publication Publication Date Title
CN110163853B (en) Edge defect detection method
CN109460754B (en) A kind of water surface foreign matter detecting method, device, equipment and storage medium
CN107230203B (en) Casting defect identification method based on human eye visual attention mechanism
CN113034399A (en) Binocular vision based autonomous underwater robot recovery and guide pseudo light source removing method
WO2018232518A1 (en) Determining positions and orientations of objects
CN111951238A (en) Product defect detection method
CN108647706B (en) Item recognition classification and defect detection method based on machine vision
CN115456955B (en) Method for detecting burrs inside ball cage dust cover
CN108629790B (en) A threshold segmentation method for light strip images based on deep residual network
CN110763700A (en) Method and equipment for detecting defects of semiconductor component
CN114926407A (en) Steel surface defect detection system based on deep learning
CN117274209B (en) Bottle body defect detection method, system, medium and electronic equipment
CN111968082A (en) Product packaging defect detection and identification method based on machine vision
CN115452328B (en) A VCSEL semiconductor laser array near-field light-emitting hole detection device and method
CN106067031A (en) Cooperate with the degree of depth learning network Machine Vision Recognition system based on artificial mechanism for correcting errors
CN119048837B (en) A PCB board component defect detection method, system, equipment and medium
CN114029943A (en) Target grabbing and positioning method and system based on image data processing
CN116934734A (en) Image-based part defect multipath parallel detection method, device and related medium
CN115272284A (en) Power transmission line defect identification method based on image quality evaluation
CN114120318B (en) A method for accurate extraction of target points in dark field images based on integrated decision tree
CN116786435B (en) Intelligent bullet sorting system
CN113705672B (en) Threshold selection method, system, device and storage medium for image target detection
CN119672426B (en) RFID chip image anomaly detection method and system based on intelligent learning
CN116863156A (en) Planar object edge extraction method based on machine vision and depth segmentation network
CN104182758B (en) Image-recognizing method for the pendulum measurement of rocket wind

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
GR01 Patent grant
GR01 Patent grant