CN113289911A

CN113289911A - Defect detection method and system for polyhedral material

Info

Publication number: CN113289911A
Application number: CN202110622663.7A
Authority: CN
Inventors: 诸庆; 胡远; 朱祺
Original assignee: Ningbo Sunny Instruments Co Ltd
Current assignee: Ningbo Sunny Instruments Co Ltd
Priority date: 2021-06-04
Filing date: 2021-06-04
Publication date: 2021-08-24
Anticipated expiration: 2041-06-04
Also published as: CN113289911B

Abstract

The invention relates to a method and a system for detecting defects of polyhedral materials, wherein the method comprises the following steps: a. respectively carrying out defect detection on a cutting surface, an incident and emergent surface and a reflecting surface of the polyhedral material; b. after the defect detection is finished, carrying out center positioning detection on the polyhedral material; c. and outputting the polyhedral material according to the results of the defect detection and the center positioning detection. The invention can carry out high-efficiency and high-precision defect detection on polyhedral materials such as prisms and the like.

Description

Defect detection method and system for polyhedral material

Technical Field

The invention relates to a method and a system for detecting defects of polyhedral materials.

Background

With the rapid updating and iteration of smart phones, people have higher and higher requirements on the camera quality of the mobile phones. The realization of optical zooming through a prism is one of effective ways for improving the image quality, and the application of a periscopic lens is the most reasonable scheme for realizing zooming of mainstream mobile phones in the current market. Prism processing belongs to optical cold processing, and has higher difficulty, and the final inspection of prism products in the prior art is mainly detected through an artificial microscope, and the artificial inspection has the following defects at present:

1. the detection efficiency is low. Taking a triangular prism as an example, the product has 3 optical surfaces and 2 rough surfaces, and many detection items are needed, so the manual efficiency is very low, and the detection efficiency is about 200-350 psc/h.

2. The detection precision is low. Because production line staff observes through the microscope, therefore staff's subjective judgement is more, and different staff's standard discrepancy is great, consequently also diverse to the judgement standard of product, and the precision of detection is lower.

3. The detection difficulty is high. Because the imaging surface is more, it is difficult to image each smooth and clean face completely under the microscope simultaneously, therefore some local details are difficult to image clearly, the defect on the product is difficult to monitor through the human eye singly, and the detection difficulty is higher for manual detection.

Disclosure of Invention

The invention aims to provide a method and a system for detecting defects of polyhedral materials.

In order to achieve the above object, the present invention provides a method and a system for detecting defects of polyhedral materials, wherein the method comprises the following steps:

a. respectively carrying out defect detection on a cutting surface, an incident and emergent surface and a reflecting surface of the polyhedral material;

b. after the defect detection is finished, carrying out center positioning detection on the polyhedral material;

c. and outputting the polyhedral material according to the results of the defect detection and the center positioning detection.

According to one aspect of the invention, in the step (a), when detecting the cut surface, collecting the image of the cut surface of the polyhedral material, and positioning the user-defined effective detection area in the image;

performing area affine on the user-defined effective detection area through a positioning center, performing ink coating area interception and angle measurement on the area after affine, and performing defect detection on the ink coating area;

when the defect detection is carried out, detecting ink leakage and broken edges by using a threshold segmentation method;

during angle measurement, the outlines of edges of the polyhedral material are fitted, and an included angle between adjacent outlines is calculated.

According to one aspect of the present invention, in the step (a), when the detection of the entrance and exit surfaces is performed, a positioning image and a detection image of the surface to be detected are collected;

carrying out region positioning on the positioning image, respectively carrying out region affine transformation and threshold segmentation on the positioning image subjected to the region positioning, and carrying out size measurement on the region subjected to the affine transformation of the positioning image;

performing area affine on the detection image, and intercepting a non-silk-screen area of the detection image by using an affine area of the detection image and a segmented area of the positioning image;

intercepting the silk-screen area of the detection image by utilizing the intercepted non-silk-screen area and the affine back area of the detection image;

and respectively carrying out defect detection on the silk-screen area and the non-silk-screen area of the detected image.

According to one aspect of the invention, during size measurement, point sets of the edge outline of the polyhedral material are calculated and fitted into straight lines, then the distance between the straight lines is calculated, and the distance function during calculation is selected through interface configuration.

According to one aspect of the invention, in the step (a), when the defect detection of the reflecting surface is performed, a positioning image and a detection image of the reflecting surface are collected, and the positioning image is matched and positioned;

carrying out regional affine transformation on the matched and positioned image, and carrying out collapse defect detection on the region after affine of the positioned image;

and carrying out region affine transformation on the detected image by combining the matched and positioned positioning image, and carrying out defect detection on the region of the detected image after affine.

According to one aspect of the invention, different detection surfaces are imaged to obtain scanned images of the reflecting surface, the incident and emergent surfaces and the edge;

taking the centers of the positioning images of the reflecting surface, the incident and emergent surface and the edge as positioning references, and carrying out affine transformation on the detection images of the reflecting surface, the incident and emergent surface and the edge;

carrying out defect detection on the affine region of the detected image by utilizing a gray level enhancement method, a threshold segmentation method and a blob analysis method;

and detecting the defects of the affine region of the positioning image by using detection region extraction, threshold segmentation and blob analysis methods.

According to one aspect of the invention, the positioning image is collected using a point light source and the detection image is collected using an annular light source.

According to one aspect of the invention, the polyhedral material is a prism.

The defect detection system comprises a circulation module, a feeding module, a cutting surface detection module, a reflecting surface detection module, an incoming and outgoing surface detection module and a discharging module, wherein the feeding module, the cutting surface detection module, the reflecting surface detection module, the incoming and outgoing surface detection module and the discharging module are arranged around the circulation module, and a discharging positioning module is further arranged between the discharging module and the circulation module.

According to an aspect of the present invention, the optical module further includes a reflecting surface compensation unit disposed in the reflecting surface detection module and/or the entrance and exit surface detection module.

According to one aspect of the invention, the circulation module comprises a turntable, a material rotating motor and a loading suction nozzle, the loading suction nozzle and the material rotating motor are positioned on two sides of the turntable, an output end of the material rotating motor is connected with the loading suction nozzle, and the loading module comprises a loading suction nozzle.

According to one aspect of the present invention, the cutting surface detection module includes a first lens, a first camera located on an image side of the first lens, a first annular light source located on an object side of the first lens, and a sliding table for adjusting the first lens and the first camera to complete focusing;

the cutting surface detection modules are arranged in even number and are respectively positioned on two sides of the circulation module.

According to an aspect of the present invention, the reflective surface detection module includes a second lens, a second camera located on an image side of the second lens, a second annular light source located on an object side of the second lens, a second point light source coaxially disposed in the second lens, two first linear light sources located on an object side of the second annular light source, two third point light sources respectively aligned with the incident surface and the exit surface for illumination, and a first servo motor.

According to an aspect of the present invention, the light incident and exiting surface detection module includes a third lens, a third camera located on an image side of the third lens, a third annular light source located on an object side of the third lens, and a fourth point light source coaxially disposed in the third lens;

the two incoming and outgoing surface detection modules are arranged vertically and are respectively used for detecting an incoming surface and an outgoing surface;

the system further comprises a second bar-shaped light source and a second servo motor, wherein the second bar-shaped light source is located between the two incoming and outgoing surface detection modules, and the second servo motor is used for driving the incoming and outgoing surface detection modules to complete focusing.

According to one scheme of the invention, aiming at the problem of low manual detection efficiency, the detection efficiency (uph) of the provided prism defect detection system can reach 1600psc/h, which is improved by at least 5 times compared with the detection efficiency of manual detection of 300psc/h, thereby effectively improving the detection efficiency.

According to one scheme of the invention, aiming at the problem of low manual detection precision, the provided defect detection method for the universal prism product has the repeated detection precision of 10um, and the accuracy of the detection result is effectively ensured.

According to one scheme of the invention, aiming at the problem of high difficulty in manual detection, the provided prism defect detection system can be used for carrying out individual detection on each optical surface, so that the defect details of each optical surface can be clearly imaged. A targeted defect detection algorithm is designed according to each optical surface, and each defect can be accurately detected.

According to one scheme of the invention, aiming at the problem of low detection efficiency of the artificial microscope, the provided prism defect detection system can automatically and accurately extract the detection area in a self-defined roi (effective detection area) mode before detection, so that the validity of the area is ensured, and the omission risk is reduced. Through the gray level enhancement algorithm, the gray level of the defects such as shallow dirt, shallow scratch and the like can be effectively improved, and the detection accuracy is improved.

Drawings

FIG. 1 schematically illustrates a block diagram of a prism to which the method and system of one embodiment of the present invention are directed;

FIG. 2 schematically shows an overall flow diagram of a defect detection method according to an embodiment of the invention;

FIG. 3 schematically shows a cut surface inspection flowchart in a defect inspection method according to an embodiment of the present invention;

fig. 4 is a view schematically showing respective results in the cut surface detection in the defect detection method according to the embodiment of the present invention;

FIG. 5 is a flow chart of the detection of the light incident and emergent surfaces in the defect detection method according to an embodiment of the present invention;

fig. 6 and 7 are diagrams schematically showing respective results of the detection of the entrance/exit surface in the defect detection method according to the embodiment of the present invention;

FIG. 8 is a flow chart schematically illustrating the detection of a reflective surface in a defect detection method according to an embodiment of the present invention;

fig. 9 is a view schematically showing results of the reflection surface detection in the defect detection method according to the embodiment of the present invention;

FIG. 10 schematically illustrates a component diagram of a defect detection system in accordance with an embodiment of the present invention;

FIG. 11 is a schematic diagram of a loading module and a cutting plane detection module in a defect detection system according to an embodiment of the invention;

FIG. 12 is a schematic diagram of a reflective surface detection module in a defect detection system according to an embodiment of the present invention;

fig. 13 is a schematic diagram showing a configuration of an entrance/exit surface detection module in the defect detection system according to the embodiment of the present invention.

Detailed Description

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.

In describing embodiments of the present invention, the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like are used in an orientation or positional relationship that is based on the orientation or positional relationship shown in the associated drawings, which is for convenience and simplicity of description only, and does not indicate or imply that the referenced device or element must have a particular orientation, be constructed and operated in a particular orientation, and thus, the above-described terms should not be construed as limiting the present invention.

The present invention is described in detail below with reference to the drawings and the specific embodiments, which are not repeated herein, but the embodiments of the present invention are not limited to the following embodiments.

The defect detection method can be used for detecting the defects of the prism. Of course, the method and system of the present invention will be described in detail below with reference to triangular prisms as an example, but the present invention is also applicable to other polyhedral materials, such as pyramid, prism, etc. shaped materials, i.e. including all prism products. Referring to fig. 1, taking a triangular prism as an example, the shape of the triangular prism can be regarded as a triangular prism, and the main detection area includes two triangular cutting surfaces G and H; three optical surfaces as prism side surfaces are respectively an incident surface A, an emergent surface B and a reflecting surface C; A. b, C three edges E intersecting each other. The incident surface and the exit surface are not distinguished from each other, and the present invention is denoted by A, B for convenience of explanation.

Referring to fig. 2, during detection, the defect detection of the cutting surface, the entrance and exit surface and the reflection surface of the polyhedral material is firstly completed. The order of completion is not limited too much, and for the triangular prism, the detection of the cutting surfaces G and H may be performed first, then the detection of the incident surface a and the exit surface B may be performed, and finally the detection of the reflecting surface C may be performed. After the detection is completed, the discharging operation can be carried out, but before discharging, the prism needs to be subjected to center positioning detection firstly, so that the discharging position is determined, and the center positioning detection of the reflecting surface C can be realized through specific access. And finally, respectively finishing the unloading of OK products and NG products according to the defect detection result and the center positioning detection result. Therefore, the method is suitable for the fields of machine vision, automatic optical detection systems and the like and is used for detecting the defects of the polyhedron.

Referring to fig. 3, in the present invention, the G/H surface (cut surface) of the prism is not a smooth surface, and thus the defect on the surface is not highly required. However, missing ink defect detection is also essential for ink-coated products. Therefore, when the detection of the cutting surface is carried out, firstly, the image of the cutting surface of the polyhedral material is collected (the user-defined effective detection area is used as an area for collection), and the user-defined effective detection area is positioned. And performing area affine on the user-defined effective detection area through a positioning center, performing ink coating area interception and angle measurement on an area after the affine of the cutting surface, and then performing defect detection on the ink coating area of the cutting surface. When the defect detection is carried out, the ink leakage and the edge breakage are detected by using a threshold segmentation method. During angle measurement, the outlines of edges of the polyhedral material are fitted, and an included angle between adjacent outlines is calculated. In summary, the invention provides a method for customizing a roi area for a G/H surface of a prism, which can accurately intercept the G/H surface of the prism, wherein the customized roi is shown in fig. 4a, and a mask generated by the customized roi is shown in fig. 4 b. Aiming at a G/H surface ink coating product, the invention provides an efficient ink leakage detection algorithm, namely, a G/H surface detection area is firstly positioned, and then the user-defined roi area is subjected to affine transformation through a positioning center. After affine transformation, the roi corresponding to the G/H surface can be intercepted, at this time, an effective detection area only including the inking area can be obtained, defects such as ink leakage and edge breakage can be accurately detected by designing various threshold segmentation algorithms, the ink defect detection effect (i.e., the detected ink defect) is shown in fig. 4c, and a partial enlarged view of the defect detection effect is shown in fig. 4 d. Aiming at the G/H surface, the invention also provides a measurement algorithm for measuring the included angle of the edge, the included angle between every two outlines is calculated by fitting the outlines of the three edges, and the detection effects are respectively shown in fig. 4e and 4 f.

Referring to fig. 5, in the present invention, when detecting the entrance and exit surfaces, the positioning image and the detection image of the surface to be detected are collected. And carrying out region positioning on the positioning image, respectively carrying out region affine transformation and threshold segmentation on the positioning image, and carrying out size measurement on the region after affine transformation of the positioning image. Performing area affine on the detection image, and intercepting the non-silk-screen area (namely the effective area 1) of the detection image by using the area after affine of the detection image and the area after segmentation of the positioning image. And intercepting the silk-screen area (namely the effective area 2) of the detection image by utilizing the intercepted non-silk-screen area and the affine back area of the detection image. And finally, respectively carrying out defect detection on the silk-screen area and the non-silk-screen area of the detected image. Therefore, the A/B surface effective area is interfered by the silk screen printing area in the imaging area, if the whole effective area is directly used as the detection area, serious over detection can be caused in the detection process, and therefore the effective area is divided into two effective detection areas in a self-defined mode under the background. Aiming at the detection of the silk-screen area, the method provides a user-defined area intercepting process, and the effective area and the interference area can be accurately separated. The above-mentioned acquired detection image is shown in fig. 6a, the positioning image is shown in fig. 6b, and the positioning effect is shown in fig. 6 d. First, the positioning image is threshold-divided to separate the white area shown in fig. 6b (this area can be referred to as a1), and a mask diagram shown in fig. 6c is formed. Then, the affine area of the detected image shown in fig. 6a is subtracted from the mask image shown in fig. 6c to obtain a silk-screen area as shown in fig. 6e, at this time, defect detection can be performed on the silk-screen area, and the detection effect image is shown in fig. 6 f. And aiming at the detection of the effective area, intercepting the non-silk-screen printing area by subtracting the mask image shown in fig. 6c from the silk-screen printing area image, and then carrying out defect detection on the effective detection area.

Aiming at shallow dirt and shallow scratch, the method of the invention provides a gray level enhancement algorithm which can obviously amplify the contrast of the defects of the original image. Firstly, traversing the collected positioning image and detection image (original image) of the surface to be detected by using a rectangle core with the size of 5-by-5, and performing nonlinear stretching on each traversed area through the following formula:

s＝clog(1+r)；

wherein S is the gray value corresponding to the enhanced pixel point, r is the gray value of the original pixel point, and c is the stretching coefficient.

The gray scale difference between the defect gray scale and the background gray scale can be enlarged by the above method, and the gray scale distribution of the digital image shows that the gray scale distribution of the image pixels is between 0 and 255, and the maximum gray scale value of the image exceeds 255 after the nonlinear stretching, so that the stretched image needs to be normalized by the following formula:

wherein dst (i, j) is the normalized gray level, src (x, y) is the gray level of the pixel point of the original image, min is the lower limit of the required gray level, and max is the upper limit of the gray level required to be normalized.

After normalization, the gray scale of the image is limited to 0-255. Finally, the average gray scale of the traversal region is obtained, if the average gray scale of the region is larger than a set threshold value, the region is a defect region, and the gray scale of the region is set to be 255; otherwise, if the gray scale of the region is smaller than the set gray scale, the region is the background, and the gray scale of the region is set to 0. Through the steps, the gray scale amplification can be carried out on the shallow dirt. Fig. 7a shows the original image with point defects, and the gray scale distribution of the point defects is shown in fig. 7 b. After the gray level of the image is enhanced, the defect map shown in fig. 7c can be obtained, the gray level of the defect after the gray level enhancement is distributed as shown in fig. 7d, the gray level difference is directly increased to 255 from 15, the background and the foreground are effectively separated, and the subsequent defect detection is facilitated.

Aiming at the requirement of measuring the overall dimension of a product, the method provided by the invention provides a general measurement algorithm, and the dimension of the product can be accurately measured. Firstly, point sets of the edge (namely, the edge) outline of the prism are calculated, the point sets are fitted into straight lines, and then the distance between the straight lines is calculated, wherein the distance can be ensured to be optimal by selecting different distance functions through interface configuration. The effect of the dimension measurement according to the line number dimension measuring method of the present invention, taking the a-plane (i.e., the incident plane) as an example, is shown in fig. 7e and 7 f.

Referring to fig. 8, in the present invention, the C-plane (i.e. reflective surface) of the prism is also a difficulty of the detection method of the present invention. Due to the interference of shadow imaging of the A/B surface on imaging, if only the C surface is detected, partial defects cannot be imaged, and the defects are missed to be detected, so that the compensation surface of the C surface needs to be additionally detected. In the invention, when the defect detection is performed on the reflecting surface, the positioning image and the detection image of the reflecting surface are firstly collected, and the positioning image is matched and positioned, wherein the positioning effect is shown in fig. 9 a. And carrying out regional affine transformation on the matched and positioned image, and carrying out collapse defect detection on the region after the affine of the positioned image. And performing area affine transformation on the detected image by combining the matched and positioned positioning image, and performing effective area defect detection on the affine area of the detected image. The invention images different detection surfaces to obtain the scanning images of the reflecting surface, the incident and emergent surfaces and the seamed edge. And performing affine transformation on the detection images of the reflecting surface, the incident and emergent surface and the edge by taking the centers of the positioning images of the reflecting surface, the incident and emergent surface and the edge as the positioning reference. And detecting the defect of the affine region of the detected image by utilizing a gray level enhancement method, a threshold segmentation method and a blob analysis method. And detecting the defects of the affine region of the positioning image by using detection region extraction, threshold segmentation and blob analysis methods. Therefore, the invention provides a tomography method, which can clearly image the defects of the C surface, the A surface, the B surface and the edge by imaging different detection surfaces through a camera under the irradiation of an annular light source, wherein each image is a group, and five groups of images are scanned in total. The center of the C surface of the point light imaging is used as a positioning reference, after affine transformation is carried out on the image shot by the annular light source, the detection area can be accurately cut, and the stability and reliability of the detection area are guaranteed. Furthermore, a teaching mode of roi that can manually draw and automatically divide is also proposed, and the obtained roi is shown in fig. 9b, thereby ensuring the accuracy of the obtained roi. Aiming at the defect detection of the effective area, the invention also provides a universal detection algorithm which mainly comprises the methods of gray level enhancement, threshold segmentation, blob analysis and the like. Aiming at the defects of light dirt, light scratch and the like with small gray difference between the foreground and the background, the defects can be effectively extracted by adopting a gray enhancement mode, and the universality of the detection algorithm is ensured. Aiming at the edge and corner breakage defects of a product, a special edge and corner breakage detection algorithm is provided, which mainly comprises detection area extraction, threshold segmentation, blob analysis and other methods, the edge and corner breakage can be accurately detected through the algorithm, the breakage detection area defined by the extracted breakage defect detection standard is shown as an inner frame in fig. 9e, and the breakage defect detection effect is shown as a frame in fig. 9 f. By the method, the center of the C surface can be accurately positioned, the area shown in FIG. 9 can be accurately intercepted according to the center coordinate, the background gray difference of the defects such as dots, scratches and the like is effectively pulled away through an image enhancement algorithm, the defects are accurately identified through a segmentation algorithm, the scratch detection effect is shown in FIG. 9C, and the dot defect detection effect is shown in FIG. 9 d.

In the above detection, the detection image is collected by using the annular light source, the positioning image is collected by using the point light source, and the positioning image and the detection image can be respectively collected by switching the light source during the collection, which can be referred to the following description of the system.

Referring to fig. 10, the system for implementing the defect detection method of the present invention includes a circulation module 1, a feeding module 2, a cutting surface detection module 3, a reflecting surface detection module 4, an incident and emergent surface detection module 5, and a blanking module 6, which are disposed around the circulation module 1 (or called product circulation mechanical structure), and a blanking positioning module 7 is further disposed between the blanking module 6 and the circulation module 1. Certainly, in order to realize the detection of the C-plane compensation surface and improve the efficiency, the invention is further provided with a reflection surface compensation unit which is arranged in either the reflection surface detection module 4 or the entrance and exit surface detection module 5, or both the reflection surface compensation unit and the entrance and exit surface detection module, so that a user can selectively start the detection according to actual needs. The modules are main components of the defect detection system, and the modules are arranged as shown in fig. 10 to form detection stations of all items, namely a G-plane detection station, an H-plane detection station, an AB detection station and a C-plane detection station. Therefore, products are circulated on the circulation module 1, and the defect detection of all areas is completed through all the stations. After the detection is finished, the discharging positioning module 7 can position the product, and the discharging module 6 respectively puts the OK product and the NG product into the corresponding material trays. Of course, in some embodiments, the detection of the C-plane and the AB-plane may be the reverse of the above embodiments, that is, the detection of the C-plane may be performed by first transferring to the C-plane detection station, and then transferring to the AB-plane detection station.

Referring to fig. 11, in this embodiment, the circulation module 1 includes a rotary table 11, a material rotation motor 12 and a loading nozzle 13, the loading nozzle 13 and the material rotation motor 12 are located on the upper and lower sides of the rotary table 11, and an output end of the material rotation motor 12 is connected to the loading nozzle 13. In the invention, the object carrying suction nozzle 13 is used for receiving materials fed from the feeding module 1, and the material rotating motor 12 is a U-axis rotating motor, so that the angle of the materials can be adjusted by driving the object carrying suction nozzle 13 to rotate. As can be seen from fig. 11, the loading module 2 includes a loading nozzle. The cutting surface detection module 3 includes a first lens 31, a first camera 32 located on the image side of the first lens 31, a first annular light source 33 located on the object side of the first lens 31, and a sliding table 34 for adjusting the focusing of the first lens 31 and the first camera 32. The present embodiment has two cutting surface detection modules 3, i.e., two appearance defect detection stations are formed to detect two cutting surface regions respectively. Of course, in other embodiments, the number of the cutting surface detection modules 3 may be selected according to actual needs, but is preferably an even number, so as to facilitate the detection of the cutting surfaces that are relatively present. The cutting surface detection station arranged close to the feeding module 2 can also carry out feeding photographing and positioning on the product, so that the product can be accurately placed on the turntable 11; and the other cutting surface detection station can identify the angle of the product, so that the verticality of the optical axis of the camera lens can be ensured when the product is positioned at the cutting surface detection station and the AB detection station. The first camera 32 is a high-quality image sensor, the first lens 31 is a high-resolution telecentric lens, the first annular light source 33 is a high-angle light source with a specific angle, so that high-angle shooting is formed, the shooting is more uniform in a processing mode compatible with G surfaces and H surfaces of different processing conditions, such as frosting, inking and the like, the distance between the light source and a product is fixed, and the integral optical system is guaranteed to be most sensitive to defects at the distance. The angle of the light source can be determined according to the height of the light source from the product. The light source is triggered to shoot images, and a camera and a lens can focus a product (mainly a sample) through the manual adjustment sliding table 34, so that the working distance can be conveniently adjusted during equipment debugging. In addition, the station not only realizes the detection of the cutting surface, but also can provide a positioning function. Namely, the product can adjust the angle of the product at any time in the detection process through the U-axis rotating motor in the bearing area. Therefore, the detection content of the G/H surface comprises the detection of appearance defects such as light leakage, less ink and the like of the G surface and the H surface, and also comprises the angular positioning (namely the angular positioning of the H surface) and the central positioning of the G surface of the product. And the detection algorithm designed for the smooth surface G/H is combined, and the detection area can be accurately positioned in a mode of self-defining the detection area. In addition, the method of the invention also provides a local mean algorithm, thereby effectively detecting the defects and improving the detection accuracy. The angle measurement algorithm in the method can accurately measure the angle information between every two edges, and effectively monitor the quality of products. In the device, the left module in fig. 11 takes a downward shooting mode to shoot and detect the cutting surface at the bottom of the product, and after the product position is determined, the cutting surface is moved and placed on the carrying suction nozzle 13 of the turntable 11 through the feeding module 2. In the drawing 11, the right side module adopts the upward beating, the product is rotated to the corresponding position through the turntable 11 to detect a single product, the product angle is determined, and the subsequent station detection is facilitated through the corresponding rotating angle of the lower U-axis rotating motor (namely, the material rotating motor 12). That is, the two cutting surface detection modules 3 are respectively located at the upper side and the lower side of the turntable 11 to respectively complete the detection of the G surface and the H surface.

Referring to fig. 12, in the present invention, the reflective surface detection module 4 includes a second lens 41, a second camera 42 located on the image side of the second lens 41, a second annular light source 43 located on the object side of the second lens 41 and having a high angle, a second point light source 44 coaxially disposed in the second lens 41, two first bar light sources 45 located on the object side of the second annular light source 43, two third point light sources 46 respectively aligned with the incident surface and the exit surface for illumination, and a first servo motor 47. The detection station formed by the reflecting surface detection module 4 mainly detects appearance defects of the A, B, C surface and the edge E area. In the optical system of the module, the second camera 42 is a high-quality image sensor, the second lens 41 is a high-resolution telecentric lens, and the 6 light sources of 3 types are matched, and the image sensor and the telecentric lens can be adjusted by the first servo motor 47, so that the optical system can perform operations such as layer shooting (shooting along the thickness in the vertical direction) on a detection product, and the like, and realize multi-region defect detection. The annular light source in the module also adopts a high-angle light source, so that the influence of stray light caused by the fact that light rays on the A surface and the B surface are reflected and enter the lens is avoided. The second point light source 44 is located at the built-in coaxial port of the lens and is fixed with a jackscrew, thereby ensuring the vertical uniformity of the light source. The lighting method of the coaxial point light can be used for detecting defects such as broken edges and the like, and can provide positioning images. The distance between the second annular light source 43 and the surface of the product is fixed, and the first bar-shaped light sources 45 are placed on both sides of the product at a fixed angle. As can be seen from FIG. 12, the point light sources in the module are externally arranged, are respectively perpendicular to the AB surface of the product at a fixed angle, and are aligned to the AB surface in the light source direction, so that the module can be used for detecting surface shallow scratch defects and can effectively reduce the omission of equipment. The illumination scheme of the strip light source can improve the brightness of the edge. The light sources are sequentially triggered and shoot corresponding images, and in combination with the detection method designed for the smooth face C, the camera and the lens can scan layer by layer through the first servo motor 47 to shoot the product layer by layer to image the smooth face C, so that clear imaging of defects of different layers is guaranteed. In addition, the method of the invention also has a mode of self-defining the detection area, can manually select the detection area aiming at products of different models, and ensures the universality of the detection method. Aiming at different defect categories, the invention adopts different segmentation algorithms, mainly comprising methods of local threshold segmentation, self-adaptive threshold segmentation, dynamic threshold segmentation and the like, and ensures that the defects can be effectively segmented. Aiming at the defects of light dirt, light scratch and the like with small gray difference between the foreground and the background, the defects can be effectively extracted by adopting the four segmentation algorithm modes, and the effectiveness of the detection algorithm is ensured. Aiming at the edge-chipping and corner-chipping defects of products, the special edge-chipping and corner-chipping detection algorithm disclosed by the invention can be combined, and mainly comprises the methods of extracting edge-chipping and corner-chipping detection areas, segmenting threshold values, analyzing blob and the like. Meanwhile, the image sensor and the telecentric lens of the module can be adjusted differently through the first servo motor 47, so that multiple images can be shot on different surfaces in the actual shooting process, and the scratch defect detection inside the film layer in the A surface and the B surface is met. Comprehensively, the detection contents of the C station comprise dots on the C surface, scratches, dirt, edge breakage, ink overflow of silk screen printing, ink shortage of silk screen printing and light leakage of silk screen printing; scratching the surface A and the surface B; and the AB surface intersected edges have the appearance defects of ink shortage, ink overflow and light leakage.

Referring to fig. 13, in the present invention, two identical and mutually perpendicular entrance and exit surface detection modules 5 are provided for performing defect detection on an entrance surface a and an exit surface B of the same product, respectively. The light incident and emergent surface detection module 5 includes a third lens 51, a third camera 52 positioned on the image side of the third lens 51, a third annular light source 53 with a high angle positioned on the object side of the third lens 51, and a fourth point light source 54 coaxially arranged in the third lens 51. Of course, the system further includes a second bar light source 55 and a second servo motor 56, which are matched with the entrance and exit surface detection modules 5, and the second bar light source 55 is located between the two entrance and exit surface detection modules 5. And the detection station formed by the incoming and outgoing surface detection module 5 mainly detects appearance defects of the areas of the surfaces A and B and the edge E. In the optical system of the module, the third camera 52 is a high-quality image sensor, the third lens 51 is a high-resolution telecentric lens, and the above-mentioned 5 three types of light sources are matched. In some embodiments, the image sensor and the telecentric lens can also be adjusted by the second servo motor 56 to focus the inspection product. As can be seen from the above, the two incoming and outgoing surface detection modules 5 are disposed perpendicular to each other, that is, the two camera lenses are perpendicular to the a-plane and the B-plane respectively and also in a perpendicular state. The state can effectively reduce stations, so that the equipment volume is smaller. Of course, this is only a special arrangement, and the two camera lenses may be separately inspected, or may be separated into two stations similar to the cutting surface inspection module 2. The fourth point light source 54 is also arranged at the built-in coaxial port of the lens and is also fixed through a jackscrew, so that the vertical uniformity of the light source is ensured. The third annular light source 53 is at a fixed distance from the product surface and the second strip light source 55 is placed at a fixed angle between the two sets of optical systems for detecting edges (AB cross plane edges). Each light source in the module is sequentially triggered and corresponding images are shot, so that the light sources can be mutually matched, and the image shooting efficiency is improved. Specifically, the coaxial point light on one camera lens can be simultaneously used for the backlight of the other camera lens, so that the shooting of two images can be completed in the time of one image. Of course, in other embodiments, the light sources with different functions may be split and photographed separately. The illumination mode of the strip-shaped light source can also improve the brightness of the edge. Meanwhile, when the module is used for detection, the product can rotate in an angle through the U-axis rotating motor in the bearing area, so that the appearance defect detection of a part of the area (namely the C-surface compensation surface) of the C surface is compatible. In some embodiments, the two light incident and emergent surface detection modules 5 can also perform horizontal movement in the X and Y directions by the second servo motor 56 to complete focusing of the product, that is, the focusing mode can be selected according to actual requirements. By combining the detection algorithm for the A/B area of the optical surface, different areas are processed in a mode of self-defining the effective area ROI and the silk screen area ROI, and effective detection of the whole area of the product is guaranteed. The method also comprises a gray level enhancement algorithm (which can be used in the detection field), so that the defects of shallow imaging, such as dots, scratches and the like can be accurately detected, and the reliability and the accuracy of detection are improved. The method for measuring the size of the optical surface can accurately measure the size information of the prism and effectively monitor the quality of the prism product. Comprehensively, the detection contents of the AB station comprise dots, scratches, dirt, broken edges, screen printing ink overflow, screen printing ink shortage, screen printing light leakage and membrane shortage appearance defects of the A surface and the B surface; the A surface and the B surface respectively have the appearance defects of ink shortage, ink overflow and light leakage of the crossed edge of the C surface; and the appearance defects of dots, scratches and dirt in a small part of area of the C surface.

Therefore, when the system is used for defect detection, the feeding module 2 is used for sucking and transporting the material box loaded with the product to a feeding designated position, and the feeding camera moves to the designated area to photograph and identify the material tray. And then the suction nozzle sucks the product to turn over to a detection area above a cutting surface detection module 3 closest to the feeding module 2. After the G surface is photographed and detected, the center of the determined position is placed on the carrying suction nozzle 13 of the turntable 11, and then the G surface is sequentially circulated on different stations. Firstly, the H surface is detected and positioned and the ink defect is detected by an upper shooting and lower detecting camera in another cutting surface detecting module 3. And then the next station is transferred to an incoming and outgoing surface detection module 5 (namely an AB surface detection station) to detect AB surface defects, positioning of an A/B surface and detection of defects such as bits, scratches, demolding, dirt and the like are carried out, the product is rotated by a certain angle on the same station, an optical system in a module on one side of the station is used for shooting, then the turntable is rotated to the position of a C surface compensation surface, and detection of defects such as bits, scratches, dirt and the like of the C surface compensation surface is carried out. And then the light source on the station is triggered to shoot in sequence when the light source flows to the reflecting surface detection module 4, and the C-shaped smooth surface is positioned and the defects such as spot, scratch, demoulding, dirt and the like are detected. And finally, the product is transferred to a blanking station through a turntable, and the suction nozzle is used for sucking the product to rotate and then placing the product above the blanking positioning module 7 for position recognition, so that the C-plane center positioning detection is completed. And after all station results are carried out or operated, according to the final result of product judgment, the NG products and the OK products are distinguished through the blanking module, and the suction nozzles are used for respectively placing the products into the corresponding material trays.

In conclusion, aiming at the problem of low manual detection efficiency, the detection efficiency (uph) of the prism defect detection system provided by the invention can reach 1600psc/h, which is improved by at least 5 times compared with the detection efficiency of manual detection of 300psc/h, thereby effectively improving the detection efficiency. Aiming at the problem of low manual detection precision, the provided defect detection method for the universal prism product has the advantages that the repeated detection precision reaches 10um, and the accuracy of a detection result is effectively ensured. Aiming at the problem of high difficulty in manual detection, the provided prism defect detection system can be used for carrying out independent detection on each optical surface, and the defect details of each optical surface can be clearly imaged. A targeted defect detection algorithm is designed according to each optical surface, and each defect can be accurately detected. Aiming at the observation mode of a human eye microscope, the invention can accurately extract the detection area by a user-defined roi mode, thereby ensuring the validity of the area and reducing the risk of missing detection. Through the gray level enhancement algorithm, the gray level of the defects such as shallow dirt, shallow scratch and the like can be effectively improved, and the detection accuracy is improved.

The above description is only one embodiment of the present invention, and is not intended to limit the present invention, and it is apparent to those skilled in the art that various modifications and variations can be made in the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A defect detection method for polyhedral materials comprises the following steps:

2. The defect detection method of claim 1, wherein in the step (a), when detecting the cut surface, collecting an image of the cut surface of the polyhedral material, and positioning a custom effective detection area in the image;

3. The defect detection method of claim 1, wherein in the step (a), when the detection of the entrance and exit surfaces is performed, a positioning image and a detection image of the surface to be detected are acquired;

4. A defect detection method as claimed in claim 3, wherein during the dimensional measurement, the point sets of the edge profile of the polyhedral material are calculated and fitted to straight lines, and then the distance between the straight lines is calculated, the distance function during the calculation being selected by the interface configuration.

5. The defect detection method according to claim 1, wherein in the step (a), when the defect detection of the reflecting surface is performed, a positioning image and a detection image of the reflecting surface are collected, and the positioning image is matched and positioned;

6. The defect detection method of claim 5, wherein different detection surfaces are imaged to obtain scanned images of the reflecting surface, the incident and emergent surfaces and the edge;

7. A defect detection method according to claim 2, 3 or 5, characterized in that the positioning image is acquired by means of a point light source and the detection image is acquired by means of a ring light source.

8. The method of claim 1, wherein the polyhedral material is a prism.

9. A system for implementing the defect detection method of any one of claims 1 to 8, comprising a circulation module (1), and a feeding module (2), a cutting surface detection module (3), a reflecting surface detection module (4), an incident and emergent surface detection module (5) and a blanking module (6) which are arranged around the circulation module (1), wherein a blanking positioning module (7) is further arranged between the blanking module (6) and the circulation module (1).

10. The system according to claim 9, further comprising a reflecting surface compensation unit disposed in the reflecting surface detection module (4) and/or the entrance and exit surface detection module (5);

the circulation module (1) comprises a rotary table (11), a material rotating motor (12) and a carrying suction nozzle (13), the carrying suction nozzle (13) and the material rotating motor (12) are positioned on two sides of the rotary table (11), the output end of the material rotating motor (12) is connected with the carrying suction nozzle (13), and the loading module (2) comprises a loading suction nozzle;

the cutting surface detection module (3) comprises a first lens (31), a first camera (32) positioned on the image side of the first lens (31), a first annular light source (33) positioned on the object side of the first lens (31), and a sliding table (34) used for adjusting the focusing of the first lens (31) and the first camera (32);

the cutting surface detection modules (3) are arranged in even number and are respectively positioned on two sides of the circulation module (1);

the reflecting surface detection module (4) comprises a second lens (41), a second camera (42) positioned on the image side of the second lens (41), a second annular light source (43) positioned on the object side of the second lens (41), a second point light source (44) coaxially arranged in the second lens (41), two first strip light sources (45) positioned on the object side of the second annular light source (43), two third point light sources (46) respectively aligned with an incident surface and an emergent surface for irradiation, and a first servo motor (47);

the light incident and emergent surface detection module (5) comprises a third lens (51), a third camera (52) positioned on the image side of the third lens (51), a third annular light source (53) positioned on the object side of the third lens (51), and a fourth point light source (54) coaxially arranged in the third lens (51);

the two incoming and outgoing surface detection modules (5) are arranged vertically and are used for detecting an incoming surface and an outgoing surface respectively;

the system further comprises a second bar-shaped light source (55) and a second servo motor (56), wherein the second bar-shaped light source (55) is located between the two entrance and exit surface detection modules (5), and the second servo motor (56) is used for driving the entrance and exit surface detection modules (5) to complete focusing.