CN108844961A - A kind of temperature controller case vision detection system and method - Google Patents

Abstract

The invention discloses a kind of temperature controller case vision detection system and methods, including：Glass rotary disc, the first station, second station and 3rd station, the first station, second station and 3rd station are arranged along the edge of glass rotary disc respectively；The camera of first station is used to shoot the bottom outer ring image of tested temperature controller case；The camera of second station is used to shoot the bottom inner ring image of tested temperature controller case；The camera of 3rd station is used to shoot the top image of tested temperature controller case.The present invention can automatically carry out temperature controller case different zones case of surface defects detecting and acquiring image, and the defect problem of case top and bottom is found by image analysis, improves detection efficiency and accuracy.It can be applied to Machine Vision Inspecting System field.

Description

A visual inspection system and method for a thermostat housing

technical field

The invention relates to a visual inspection system and method, in particular to a visual inspection system and method for a temperature controller housing.

Background technique

The mass production of modern industry is difficult to guarantee the quality of each shell, so the company needs to measure the size and surface defects of each shell before the shell leaves the factory. The common method is to manually measure the shell by the production staff. Measurement and visual inspection, manual inspection is labor-intensive for workers, and is limited by the mental state of workers, proficiency level, lack of experience and other reasons, resulting in slow detection of the shell, low efficiency, and prone to missed inspections. Errors such as wrong detection, low degree of automation.

Contents of the invention

The present invention provides a visual inspection system and method for a thermostat housing, which can automatically detect the size and surface defects of the thermostat housing.

The solution of the present invention to solve the technical problem is: a visual inspection system for the temperature controller housing, including: a glass turntable, a first station, a second station and a third station, the first station, the second station The first station and the third station are respectively set along the edge of the glass turntable;

The first station, the second station and the third station are all equipped with a fixed frame, a camera and a machine vision light source, and the machine vision light source and the camera are all arranged in the fixed frame, wherein the first station and the second station The cameras of the third station are all set under the glass turntable and facing the glass turntable. The cameras of the third station are set above the glass turntable and facing the glass turntable. The glass turntable is used to drive the temperature controller housing under test. Pass the first station, the second station and the third station respectively;

The camera at the first station is used to take images of the outer ring of the bottom of the thermostat housing under test;

The camera at the second station is used to take images of the bottom inner ring of the temperature controller housing under test;

The camera at the third station is used to capture the top image of the temperature controller casing under test.

Further, the visual light source includes a ring light source and a dome light source, and the glass turntable is arranged in a section between the ring light source and the dome light source.

Further, the camera includes a CCD sensor camera and an FS-LM3517 industrial lens.

A method for visual inspection of a thermostat housing, comprising:

Obtain the bottom outer ring image, the bottom inner ring image and the top image of the temperature controller casing under test from the cameras of the first station, the second station and the third station respectively;

Preprocessing the bottom outer ring image, the bottom inner ring image and the top image of the temperature controller housing under test respectively, to obtain the preprocessed bottom outer ring image, bottom inner ring image and top image;

Extracting the radius value information of the outer circle from the bottom outer circle image, and comparing the radius value information with the preset standard outer circle radius value to determine whether the temperature controller housing under test is qualified;

Establish a reference template, match the bottom inner circle image and the top image according to the reference template, locate the ROI area, and obtain the designated detection area;

Rigid body affine transformation and global threshold value processing are sequentially performed on the specified detection area, a detection gray scale interval is set, and whether the temperature controller housing under test is qualified is judged according to the detection gray scale interval.

Further, the preprocessing includes: Gaussian filtering processing.

Further, the method for extracting the radius value information of the outer circle from the bottom outer circle image includes: performing multi-threshold segmentation on the bottom outer circle image to obtain a thick outer edge of the bottom outer circle, and performing least squares segmentation on the bottom outer circle. The circle is fitted by multiplication to obtain the radius value information of the outer ring at the bottom.

The beneficial effects of the present invention are: the present invention automatically detects, collects images and analyzes the surface defects in different areas of the bottom and top of the thermostat casing, accurately and efficiently finds the size errors and surface defects of the casing, and improves the temperature control system. Detection efficiency and accuracy.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly describe the drawings that need to be used in the description of the embodiments. Apparently, the described drawings are only some embodiments of the present invention, not all embodiments, and those skilled in the art can obtain other designs and drawings based on these drawings without creative work.

Fig. 1 is the structural representation of detection system of the present invention;

Fig. 2 is the structural representation of first station and second station;

Fig. 3 is the structural representation of the 3rd station;

Fig. 4 is the original image collected by the first station;

Fig. 5 is the image after Gaussian filtering of the image collected by the first station;

Fig. 6 is a structural schematic diagram of the bottom and top of the thermostat housing;

Fig. 7 is the thick outer edge obtained by multi-threshold segmentation during the first station detection;

Fig. 8 is the outer circular edge to be fitted during the first station detection;

Fig. 9 is a flow chart of the detection of the first station;

Fig. 10 is a schematic diagram of the second station detection reference template;

Fig. 11 is a diagram of the detection result of hole jam material in the second station;

Fig. 12 is a flow chart of the second station detection hole jam material;

Figure 13 is the third station detection reference module;

Fig. 14 is a diagram of the result of blind hole piercing at the third station;

Fig. 15 is a flow chart of the detection of the third station.

Detailed ways

The concept, specific structure and technical effects of the present invention will be clearly and completely described below in conjunction with the embodiments and accompanying drawings, so as to fully understand the purpose, features and effects of the present invention. Apparently, the described embodiments are only some of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative efforts belong to The protection scope of the present invention. In addition, all the connection/connection relationships mentioned in this article do not refer to the direct connection of components, but mean that a better connection structure can be formed by adding or reducing connection accessories according to specific implementation conditions. The various technical features in the invention can be combined interactively on the premise of not conflicting with each other.

Embodiment 1, referring to Fig. 1, Fig. 2, Fig. 3 and Fig. 6, wherein in Fig. 6, the upper figure is the top 9 of the thermostat housing, and the lower figure is the bottom 10 of the thermostat housing. A visual detection system for a thermostat housing, comprising: a glass turntable 4, a first station 1, a second station 2 and a third station 3, the first station 1, the second station 2 and the third station The bits 3 are respectively set along the edge of the glass turntable 4;

The first station 1, the second station 2 and the third station 3 are all provided with a fixed frame 5, a camera 8 and a machine vision light source, and the machine vision light source and the camera 8 are all arranged in the fixed frame 5, wherein the first The cameras 8 of the first station 1 and the second station 2 are all arranged below the glass turntable 4, and towards the glass turntable 4, and the camera 8 of the third station 3 is arranged above the glass turntable 4, and towards the glass turntable 4. Glass turntable4.

Because the heights of the outer ring and the inner ring of the bottom 10 of the housing under test are inconsistent, that is, the outer ring is closer to the lens, the depth of field of the camera 8 of the first station 1 is shallower than that of the camera 8 of the second station 2, so that Take a clearer image of the outer ring at the bottom of the shell; similarly, because the inner ring is far away from the lens, the depth of field of the camera 8 of the second station 2 is deeper than that of the camera 8 of the first station 1, so that the A clearer image of the inner ring at the bottom of the housing.

The glass turntable 4, the first station 1, the second station 2 and the third station 3 are driven by a PLC electrical control system whose model is FX3U-48MT.

The working principle of the visual inspection system:

The temperature controller housing to be tested is placed on the glass turntable 4, and the glass turntable 4 rotates under the drive of the PLC electrical system, passing through the first station 1, the second station 2 and the third station 3 respectively, the first station The camera 8 of 1 takes the outer circle image of the bottom 10 of the thermostat housing under test, the camera 8 of the second station 2 takes the inner circle image of the bottom 10 of the thermostat housing under test, and the camera 8 of the third station 3 Take the image of the top 9 of the temperature controller housing under test; the images of the temperature controller housing under test obtained by the first station 1, the second station 2 and the third station 3 are sent to the computer through the USB port, and the computer The acquired images are further analyzed and processed.

Further, the visual light source includes a ring light source 6 and a dome light source 7 , and the glass turntable 4 is arranged in a section between the ring light source 6 and the dome light source 7 .

The ring light source 6 is a 30-degree low-angle ring light source, which is conducive to improving detection accuracy and obtaining images with clear surfaces and prominent edge contours.

The camera 8 includes a CCD sensor camera and an FS-LM3517 industrial lens.

The present invention replaces manual detection by automatic detection at three stations, which not only improves the detection efficiency and accuracy, but also effectively reduces the detection cost. At the same time, the top 9 and The bottom 10 images are easy to operate and highly practical.

Wherein, the present invention also includes a visual inspection method for a thermostat housing, comprising: obtaining the housing of the thermostat under test from the cameras 8 of the first station 1, the second station 2, and the third station 3, respectively. Outer image of bottom 10, inner image of bottom 10, image of top 9;

Perform preprocessing on the outer ring image of the bottom 10, the inner ring image of the bottom 10, and the top image of the top 9 of the thermostat housing under test respectively, to obtain the preprocessed bottom outer ring image, bottom inner ring image, and top image;

Referring to FIG. 4 and FIG. 5 , the preprocessing includes: Gaussian filtering processing.

Carry out multi-threshold segmentation to the bottom outer circle image to obtain the thick outer edge of the bottom outer circle, and perform least squares method fitting circle on the bottom outer circle to obtain the radius value information of the bottom outer circle, and convert the radius Value information is compared with the preset standard outer ring radius value to judge whether the tested thermostat housing is qualified;

Establish a reference template, match the bottom inner circle image and the top image according to the reference template, locate the ROI area, and obtain the designated detection area;

Rigid body affine transformation and global threshold value processing are sequentially performed on the specified detection area, a detection gray scale interval is set, and whether the temperature controller housing under test is qualified is judged according to the detection gray scale interval.

Both the bottom 10 and the top 9 of the thermostat housing under test are found to be unqualified, and the various defect problems that appear at the bottom 10 and the top 9 are defined, and the types of defects detected by each testing station are reasonably arranged. Table 1 shows the division of various types of defects:

The above image processing process can be carried out in the software platform Halcon.

Referring to Fig. 6 to Fig. 9, the detection process of the first station 1 detecting the outer circle of the bottom 10 of the thermostat housing under test:

Perform multi-threshold segmentation on the preprocessed bottom outer circle image, use the core function auto_threshold dynamic threshold segmentation of the multi-threshold method to initially find the edge of the bottom outer circle, thereby initially segment the shell and the background, and use the function boundary to extract the closest target area The contour of the contour, perform a closed operation on the contour, and then perform a reduce_domain subtraction operation on the expanded area and the original image, so that a real image with only the outer circular edge of the bottom 10 of the shell can be obtained. Secondly, find the center by fitting the circle after extracting the edge, use the Sobel operator to extract the sub-pixel edge contour in the function edges_sub_pix, segment the extracted sub-pixel xld contour, and divide the contour into straight line segments, circles (or arcs) , Different segments of elliptical arcs, the function used is segment_contours_xld, the segmentation mode used by this function is 'lines_circles' (using straight line segments and circles (arcs) to divide), and then the function union_adjacent_contours_xld is used to merge adjacent arc contours, using The function fit_circle_contour_xld performs the least squares method to fit the circle, and the center coordinates and radius of the outer circle at the bottom of the measured shell can be obtained, and compared with the set standard outer diameter value to determine whether it meets the requirements.

Then determine the position to generate a circle according to the center coordinates and radius, obtain the accurate circle part after the morphological operation and area subtraction, and use the function polar_trans_image_ext to convert the circle under the Cartesian coordinate system in the image into a polar coordinate form, Smooth and filter the image in polar coordinates, and then find out the dissatisfied material and the long part of the nozzle in polar coordinates through the global threshold and connected region selection, and mark them out, and then use the function polar_trans_region_inv to convert the ring from polar coordinates to Cartesian coordinates, that is Obvious defects can be displayed in a window.

From the six sample experiments, most of the errors of the outer circle radius are kept within the tolerance range of 0.01mm, which basically meets the inspection requirements, and the qualified and unqualified conditions can basically be judged for the defect area of insufficient material and nozzle length.

Referring to Figure 10-Figure 12, the second station 2 detects the detection process of the inner hole hole defect at the bottom 10 of the thermostat housing under test:

First, preprocess the inner circle image of the bottom 10 of the thermostat under test, establish a reference template, use the function gen_rectangle2 to delineate the ROI rectangular area of the reference template, use the function area_center() to find its center coordinates, and use the function orientation_region to find out The area deflection angle, and then use the function reduce_domain() to separate the rectangular area from the image; in order to quickly and easily determine the parameters of the pyramid series NumLevels and contrast Contrast when creating the template create_shape_model, you need to use the function inspect_shape_model to create a representation of the shape model form, which can display multiple pyramid series models, each object is accessed through select_obj, and then the function create_shape_model() is used to create a template, and the function get_shape_model_contours() is used to obtain the outline of the template in order to find the template more quickly and accurately .

The second step is to generate the ROI area for detecting the card material hole. The function union2 combines the two ROI circle areas as a detection area. The find_shape_model() function finds the area that best matches the reference template in the image and returns the area of the area. The coordinates of the center point and the rotation angle, using the center point and angle combined with the function vector_angle_to_rigid(), perform rigid affine transformation on the detection area, so that each image can find the position of the corresponding detection area, and then perform global threshold processing on the detection area, if the hole If the grayscale value in the area is less than 40 and the threshold area is greater than 0.1 mm, there will be a hole jam defect 11 .

The second station 2 detects pitting and material shrinkage on the inner circle of the bottom 10 of the temperature controller housing to be tested. When one of the three defects of hole jam, pitting and material shrinkage occurs in the inner circle of the bottom 10, it is determined that the temperature controller housing under test is unqualified.

Referring to Figures 13 to 15, the third station 3 detects the detection process of the blind hole defect 12 on the top 9 of the thermostat housing under test:

The detection principle of the third station 3 is similar to the detection principle of the second station 2. First, the image of the top of the temperature controller under test is preprocessed, and the rectangular module where the two circles are located is used as a reference template, and this position is selected as the reference template. The template is because there are no defects in this location and there are obvious features such as two circles and letters. Then generate the blind hole detection area, find the area that best matches the reference template in the image, locate the positions of the two circles, use edge segmentation and edge fitting to obtain a complete and accurate small circle, and obtain the radius of the circle from it. The average value of the radius measured by the two circles is compared with the standard radius, and the error must be within the tolerance range of 0.01mm to be considered qualified.

Then use the coordinate values of the detection area in the obtained reference template to affine transform the detection area, accurately locate the detection area in the image, and perform global threshold processing on the detection area, wherein the threshold area with a gray value greater than 200 is greater than 0.01mm ² , then there will be defects of blind hole lack of material 12.

The third station 3 detects that the detection principle of the I-beam lack of material and material shrinkage on the top 9 of the temperature controller housing under test is similar to the detection principle of the blind hole lack of material on the top 9 . When one of the three defects of blind hole lack of material, I-shaped rib lack of material and material shrinkage occurs on the top 9, it is determined that the tested temperature controller housing is unqualified.

The present invention measures and collects images of surface defects in different areas of the bottom 10 and top 9 of the thermostat housing through three different automatic detection devices, and compares and analyzes the images through software algorithms to find the housing accurately and efficiently. The problem of dimensional errors and surface defects improves the detection efficiency, reduces human detection errors, and forms an effective and standardized machine vision detection system.

The preferred embodiments of the present invention have been described in detail above, but the invention is not limited to the described embodiments, and those skilled in the art can also make various equivalent modifications or replacements without violating the spirit of the present invention. These equivalent modifications or replacements are all within the scope defined by the claims of the present application.

Claims (6)

1. a kind of temperature controller case vision detection system, which is characterized in that including：Glass rotary disc, the first station, second station and 3rd station, the first station, second station and 3rd station are arranged along the edge of glass rotary disc respectively；

First station, second station and 3rd station are equipped with fixed frame, camera and machine vision light source, the machine vision Light source and camera are arranged in fixed frame, wherein the first station and the camera of second station are arranged at glass rotary disc Lower section, and towards the glass rotary disc, the camera of 3rd station is arranged in the top of glass rotary disc, and towards the glass Turntable, glass rotary disc is for driving tested temperature controller case to pass through the first station, second station and 3rd station respectively；

The camera of first station is used to shoot the bottom outer ring image of tested temperature controller case；

The camera of second station is used to shoot the bottom inner ring image of tested temperature controller case；

The camera of 3rd station is used to shoot the top image of tested temperature controller case.

2. a kind of temperature controller case vision detection system according to claim 1, which is characterized in that the vision light source packet Annular light source and dome light source are included, the glass rotary disc is set in the section between the annular light source and dome light source.

3. a kind of temperature controller case vision detection system according to claim 1, which is characterized in that the camera includes Ccd sensor camera and FS-LM3517 industrial lens.

4. a kind of temperature controller case visible detection method, which is characterized in that including：

The loop graph outside the bottom for obtaining tested temperature controller case in the camera of the first station, second station and 3rd station respectively Picture, bottom inner ring image and top image；

The bottom outer ring image of tested temperature controller case, bottom inner ring image and top image are pre-processed respectively, obtained Pretreated bottom outer ring image, bottom inner ring image and top image；

The radius value information of outer ring is extracted from the image of the bottom outer ring, and will be outside the radius value information and preset standard Radius value comparison is enclosed, judges whether tested temperature controller case is qualified；

Reference template is established, is matched according to reference template with the bottom inner ring image, top image, orients the area ROI Domain obtains specified detection zone；

Rigid body affine transformation, global threshold processing are successively carried out to the specified detection zone, setting detects gray scale interval, according to The detection gray scale interval judges whether tested temperature controller case is qualified.

5. a kind of temperature controller case visible detection method according to claim 4, which is characterized in that the pretreatment packet It includes：Gaussian filtering process.

6. a kind of temperature controller case visible detection method according to claim 4, which is characterized in that from the bottom outer ring The method of the radius value information of extraction outer ring includes in image：Multi-threshold segmentation is carried out to bottom outer ring image, is obtained on earth The thick outer edge of portion outer ring, the bottom outer ring carry out least square method fitting circle, obtain the radius value letter of the bottom outer ring Breath.