CN115931898A

CN115931898A - Visual detection method and device for surface defects of ceramic substrate and storage medium

Info

Publication number: CN115931898A
Application number: CN202211650082.5A
Authority: CN
Inventors: 蔡春明; 陈贵
Original assignee: Hangzhou Hanzhentu New Technology Co ltd
Current assignee: Hangzhou Hanzhentu New Technology Co ltd
Priority date: 2022-12-21
Filing date: 2022-12-21
Publication date: 2023-04-07

Abstract

The embodiment of the application discloses a method and a device for visually detecting the surface defects of a ceramic substrate and a storage medium, wherein the method for visually detecting the surface defects of the ceramic substrate comprises the following steps: the method comprises the following steps that a ceramic substrate to be detected is driven to move at a constant speed through a conveying device, and through holes are formed in the lower portion of the ceramic substrate; determining a stroboscopic time sequence of a first stroboscopic light source and a second stroboscopic light source which are connected through a light source controller based on a line frequency of a linear array camera arranged above a through hole, wherein the first stroboscopic light source is arranged below the through hole, and the second stroboscopic light source is arranged obliquely above the through hole; acquiring images of the ceramic substrate by using a linear array camera according to a stroboscopic time sequence to obtain a first image and a second image of the ceramic substrate under the irradiation of a first stroboscopic light source and a second stroboscopic light source respectively; and selecting the corresponding first image and/or second image for analysis according to the surface characteristics of the images with different surface defects to obtain the detection result of the surface defects.

Description

Visual detection method and device for surface defects of ceramic substrate and storage medium

Technical Field

The application relates to the technical field of surface detection, in particular to a visual detection method and device for surface defects of a ceramic substrate and a storage medium.

Background

The surface defect detection of the ceramic substrate mostly adopts an area array CCD visual detection scheme in the current market, adopts a mode of combining a backlight and 4 strip light sources, needs 4 stations in total because the ceramic substrate detects the front and back sides, and each station adopts a group of cameras and lenses, and needs 4 groups in total. The movement of the wafer between the stations is carried by a movement platform with a suction nozzle, and a wafer turning mechanism is needed in the middle and is used for turning the ceramic substrate from the front side to the back side. The mechanism is shown in figure 1. The solution in fig. 1 has several disadvantages:

1. at least 4 detection stations are needed, the mechanical structure is complex, 4 groups of area-array cameras and lenses with large target surfaces are needed, and the cost is high.

2. An object to be detected needs to be conveyed to a station 1 from a material loading position by using a moving platform 1, conveyed to a station 2 and conveyed to a front end belt, the belt conveys the object into a sheet turning mechanism, the sheet turning mechanism finishes sheet turning action, the belt flows into a rear end belt, then the object is conveyed to a station 3 by the moving platform 2, then the object is conveyed to a station 4, and the object needs to be conveyed to a material unloading frame from the station 4 after detection. The whole movement process is complex.

3. The detection efficiency is low due to the complex motion process, and the detection efficiency is only about 40 pieces per minute.

4. The object is carried for many times in the moving process, so that the object is rigidly collided with the objective table and the sheet turning mechanism for many times, and the ceramic substrate is easy to crack.

Disclosure of Invention

An object of the embodiments of the present application is to provide a method and an apparatus for visually detecting surface defects of a ceramic substrate, and a storage medium, so as to solve the problems in the prior art that the hardware cost of the surface defect detection scheme of the ceramic substrate is high, the mechanical structure and the motion process of the detection apparatus are complex, the detection efficiency is low, and secondary damage is easily caused to the product.

In order to achieve the above object, an embodiment of the present application provides a method for visually detecting surface defects of a ceramic substrate, including the steps of: driving a ceramic substrate to be detected to move at a constant speed through a conveying device, wherein a through hole is formed below the ceramic substrate;

determining a stroboscopic time sequence of a first stroboscopic light source and a second stroboscopic light source which are connected through a light source controller based on a line frequency of the linear array camera arranged above the through hole, wherein the first stroboscopic light source is arranged below the through hole, and the second stroboscopic light source is arranged obliquely above the through hole;

acquiring images of the ceramic substrate by using the linear array camera according to the stroboscopic time sequence to obtain a first image and a second image of the ceramic substrate under the irradiation of the first stroboscopic light source and the second stroboscopic light source respectively;

and selecting the corresponding first image and/or the second image for analysis processing according to the surface characteristics of the images with different surface defects to obtain the detection result of the surface defects.

Optionally, the method for selecting the corresponding first image and/or second image for analysis processing according to the image surface features of the different surface defects includes:

carrying out filtering smoothing processing on the first image to obtain a smooth background image;

subtracting the smooth background image from the first image before filtering and smoothing to obtain a suspected highlight area related to a first defect;

and judging and screening the first defect based on the area size and/or the shape characteristic of the suspected highlight area.

extracting a main body region of the ceramic substrate from the first image;

respectively extracting edges from corner areas of the main body area, and acquiring corresponding right-angle areas according to straight lines where the edges of the corner areas are located;

and subtracting the corresponding corner area from the right-angle area to obtain a blank area, and judging whether a second defect exists according to the area size and/or the shape characteristic of the blank area.

Optionally, the method for selecting the corresponding first image and/or second image for analysis processing according to the image surface features of the different surface defects further includes:

extracting a main body region of the ceramic substrate from the first image;

and extracting edge points from the edge region of the main body region, obtaining a fitting straight line by using the edge points, and judging whether a second defect exists according to the distance from the discrete points in the edge points to the fitting straight line.

filtering and smoothing the second image to obtain a smooth background image, and extracting a bright area and a dark area from the smooth background image;

and judging whether a third defect exists by connecting the bright area and the dark area to judge whether a connected domain is formed in pair.

constructing a segmentation model which is subjected to fine tuning based on an FCN model by using a deep learning method;

and detecting the second image by using the segmentation model, and judging whether a third defect exists.

Optionally, before detecting the second image using the segmentation model, the method further includes:

constructing a classification model based on a ResNet50 model by using a deep learning method;

and classifying the second image through the classification model, and classifying the second image into a third defect image and a non-third defect image of different classes.

Optionally, the second stroboscopic light source is a three-color light source, and can stroboscopic to emit red, green and blue light;

the second image comprises a red light image, a green light image and/or a blue light image;

the line frequency of the linear array camera is four times of the stroboscopic frequency of the first stroboscopic light source and the second stroboscopic light source.

To achieve the above object, the present application also provides a ceramic substrate surface defect visual inspection device, comprising: the conveying device is used for driving the ceramic substrate to be detected to move at a constant speed, the conveying device is provided with a through hole, and the ceramic substrate is positioned above the through hole;

the linear array camera is arranged above the through hole and used for collecting images of the ceramic substrate;

the first stroboscopic light source is arranged below the through hole, and the second stroboscopic light source is arranged obliquely above the through hole;

the light source controller is respectively connected with the first stroboscopic light source and the second stroboscopic light source and is used for determining stroboscopic time sequences of the first stroboscopic light source and the second stroboscopic light source based on the line frequency of the linear array camera;

the sheet turning mechanism is used for turning the ceramic substrate;

and the analysis system is used for acquiring the images acquired by the linear array camera, and analyzing and processing the images according to the surface characteristics of the images with different surface defects to obtain the detection result of the surface defects.

To achieve the above object, the present application also provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a machine, implements the steps of the method as described above.

The embodiment of the application has the following advantages:

the embodiment of the application provides a visual inspection method for surface defects of a ceramic substrate, which comprises the following steps: driving a ceramic substrate to be detected to move at a constant speed through a conveying device, wherein a through hole is formed below the ceramic substrate; determining a stroboscopic time sequence of a first stroboscopic light source and a second stroboscopic light source which are connected through a light source controller based on a line frequency of the linear array camera arranged above the through hole, wherein the first stroboscopic light source is arranged below the through hole, and the second stroboscopic light source is arranged obliquely above the through hole; acquiring images of the ceramic substrate by using the linear array camera according to the stroboscopic time sequence to obtain a first image and a second image of the ceramic substrate under the irradiation of the first stroboscopic light source and the second stroboscopic light source respectively; and selecting the corresponding first image and/or second image for analysis according to the surface characteristics of the images with different surface defects to obtain the detection result of the surface defects.

By the method, the linear array camera is matched with the transmission device and the combination of the multi-component time-sharing stroboscopic light source, so that the hardware cost is reduced, the mechanical structure and the motion process are simplified, the detection efficiency is improved, and secondary damage to products, such as collision and fragmentation, pollution and the like in the detection process can be avoided to a great extent after simplification.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It should be apparent that the drawings in the following description are merely exemplary, and that other embodiments can be derived from the drawings provided by those of ordinary skill in the art without inventive effort.

FIG. 1 is a schematic diagram of a prior art scheme for surface defect detection of ceramic substrates;

FIG. 2 is a flowchart of a method for visual inspection of surface defects of a ceramic substrate according to an embodiment of the present disclosure;

fig. 3 is a schematic front view of a ceramic substrate to be tested according to the visual inspection method for detecting surface defects of the ceramic substrate provided in the embodiment of the present application;

fig. 4 is a schematic diagram of a light source stroboscopic timing sequence of a method for visually inspecting a surface defect of a ceramic substrate according to an embodiment of the present disclosure;

FIG. 5 is a schematic diagram illustrating an image representation of a void defect in a ceramic substrate surface defect visual inspection method according to an embodiment of the present disclosure;

FIG. 6 is a schematic diagram illustrating an image representation of a crack defect of a ceramic substrate surface defect visual inspection method according to an embodiment of the present disclosure;

FIG. 7 is a schematic diagram illustrating an image representation of defect defects of a ceramic substrate surface defect visual inspection method according to an embodiment of the present application;

FIG. 8 is a schematic diagram of a blank area of a method for visually inspecting a surface defect of a ceramic substrate according to an embodiment of the present disclosure;

FIG. 9 is a schematic representation of an image of an adhesion and scratch defect of a visual inspection method for a surface defect of a ceramic substrate according to an embodiment of the present disclosure;

fig. 10 is a schematic structural diagram of a ceramic substrate surface defect visual inspection apparatus according to an embodiment of the present application.

Detailed Description

The present disclosure is not intended to be limited to the particular embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

In addition, the technical features mentioned in the different embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

An example of the present application provides a method for visual inspection of surface defects of ceramic substrate, and referring to fig. 2, fig. 2 is a flow chart of a method for visual inspection of surface defects of ceramic substrate provided in an embodiment of the present application, it should be understood that the method may further include additional frames not shown and/or may omit the illustrated frames, and the scope of the present application is not limited in this respect.

The characteristics of the ceramic substrate to be tested, which are exemplified in the embodiments of the present application, include:

1. the shape of the ceramic substrate to be measured is similar to a chamfer rectangle, the size is relatively single, the length is 70 mm, the width is 60 mm, the thickness is between 0.3 and 0.5 mm, and products with different specifications have different thicknesses.

2. And chamfering four corners of the ceramic substrate to be detected, wherein one chamfer has a larger size, the other three chamfers are smaller, and the positions of the larger chamfers of products with different specifications are possibly different.

3. One of the four edges of the ceramic substrate to be measured is provided with 1 to 2 semicircular gaps, and the number and the positions of the gaps of products with different specifications are possibly different.

4. Regular rectangular nicks are formed on the front surface of the ceramic substrate to be tested, and the distribution and the depth of the nicks of products with different specifications are possibly different.

The schematic diagram of the front side of the ceramic substrate to be detected is shown in fig. 3, and the schematic diagram is not shown because the back side and the front side of the ceramic substrate to be detected are basically consistent and only have no red score line.

It should be understood that the scheme of the present application is also applicable to the ceramic substrates to be tested having similar characteristics, and the present application does not limit the specific forms and characteristics of these ceramic substrates to be tested as long as the ceramic substrates to be tested can perform visual inspection of surface defects based on the principle of the scheme of the present application.

The embodiment provides a method for detecting surface defects by using a composite detection device which is used for detecting the surface defects of an object by matching a single linear array camera with a plurality of light sources and alternately acquiring images in a time-sharing manner.

In step 101, a ceramic substrate to be detected is driven to move at a constant speed by a conveying device, and a through hole is arranged below the ceramic substrate.

Specifically, in some embodiments, the conveying device is a flexible conveying belt, and the conveying belt drives the ceramic substrate to be detected to move at a constant speed. The conveyor belt can be divided into two sections, a through hole is formed between the two sections of conveyor belts, in the embodiment, the through hole is in a slit shape, and the ceramic substrate to be detected moves at a constant speed on the two sections of conveyor belts and passes through the slit from the upper side, so that the lower surface is exposed through the slit. The conveying device can be in other forms, the through hole can be in other shapes, only the lower surface of the ceramic substrate to be detected can be exposed, and the shape of the through hole can be determined according to actual needs.

At step 102, determining, by a light source controller, a strobe timing sequence of a first strobe light source and a second strobe light source connected based on a line frequency of the line camera disposed above the through hole, the first strobe light source being disposed below the through hole, the second strobe light source being disposed obliquely above the through hole.

In some embodiments, the second stroboscopic light source is a three-color light source, and can strobe red, green and blue lights; the second image comprises a red light image, a green light image and/or a blue light image; the linear array camera is a CCD camera.

Specifically, a linear array CCD camera is arranged right above a gap and is responsible for acquiring images of a ceramic substrate, a backlight light source (a first stroboscopic light source) is arranged right below the gap between two sections of conveying belts, light irradiates the ceramic substrate vertically, a low-angle three-color light source (a second stroboscopic light source) is arranged right above the gap between the two sections of conveying belts 1, the ceramic substrate is irradiated by the light in a low-angle inclined mode, the two light sources are connected with a light source controller simultaneously, the whole system is connected with a PC (analysis system), and finally the PC is responsible for carrying out operation analysis on the acquired images to give detection results.

At step 103, acquiring images of the ceramic substrate by using the line scan camera according to the strobe timing sequence, and obtaining a first image and a second image of the ceramic substrate under the irradiation of the first strobe light source and the second strobe light source, respectively.

Specifically, the low-angle three-color light source alternately strobes red, green and blue light and cooperates with a backlight light source (which may be white light), which is equivalent to 4 kinds of (color) light sources alternately strobing according to a certain time sequence, and the time sequence is determined by the light source controller cooperating with the line frequency of the linear array CCD camera. The light source strobe timing diagram refers to fig. 4.

In some embodiments, the line frequency of the line CCD camera used is 4 times of the stroboscopic frequency of the light source, that is, the 1,2,3,4 line images collected by the line CCD camera respectively correspond to the backlight of the first stroboscopic light source, and the information collected when the three-color light (red), the three-color light (green), and the three-color light (blue) of the second stroboscopic light source are alternately stroboscopically turned on. Similarly, the 5,6,7,8 row images acquired by the linear array CCD camera respectively correspond to the light source lighting timing sequence of the next cycle, and so on until the complete images of the complete ceramic substrate are acquired, and finally the images are decomposed into complete images under the condition of 4 different light sources according to rows, wherein the number of the first images is 1, and the number of the second images is 3. Subsequently, software and an algorithm can select images collected under the condition of proper light source illumination according to the expression forms of different defect types under different light sources for operation processing, and finally, a detection result is given.

At step 104, according to the image surface features of different surface defects, the corresponding first image and/or second image is/are selected for analysis processing, and the detection result of the surface defect is obtained.

The following examples illustrate the analytical detection of surface defects in ceramic substrates. The main defects of the surface of the ceramic substrate comprise air holes, edge defects, cracks, adhesion, scars, smudges and the like, and each defect is provided with a special processing algorithm aiming at each defect according to different image surface characteristics. The following will be presented for different defect classifications. It should be understood that the analysis and detection method described in the following embodiments can be applied to other kinds of ceramic substrates to be detected with the same or similar surface defects, and the present application is not limited thereto as long as the types of defects of the various kinds of ceramic substrates can be analyzed and detected based on the principle of the method.

In some embodiments, the method for selecting the corresponding first image and/or second image for analysis processing according to the image surface characteristics of different surface defects comprises:

subtracting the smooth background image from the first image before filtering and smoothing to obtain a suspected highlight area about the first defect;

Specifically, the first defect may be a void or a crack.

Air holes:

the air holes are through holes which completely penetrate or partially penetrate through the surface of the ceramic substrate and appear as bright spots on a dark background on a backlight image, as shown in fig. 5, the left image with the score lines in fig. 5 is a front image of the ceramic substrate, the right image is a back image of the ceramic substrate, and the two screenshots are not two sides of the same ceramic substrate but belong to two different ceramic substrates respectively. In some embodiments, the device further comprises a sheet turning mechanism, and the ceramic substrate can be turned over by the sheet turning mechanism, so that images of the front surface and the back surface can be acquired.

The defect of the air hole needs to find out a bright spot with a higher gray value in an image background, and then the interference of other non-defect bright spots such as a cutting line is eliminated according to the morphological characteristics of the air hole, which can be roughly divided into the following three steps.

1. Filtering and smoothing the first image, generally adopting a gaussian filter to perform offline convolution processing on the image to obtain a smooth background image, generally adopting one-dimensional gaussian filtering operation in two directions of width and height of the image respectively for improving the efficiency during image filtering operation, rather than adopting two-dimensional gaussian filtering operation in two directions simultaneously, and adopting the following gaussian basic formula

2. After the image is smoothed with the background, the first image original image is subtracted from the smoothed background image with the background smoothed, and a brightness threshold is set, so that a suspected highlight area of the pore can be obtained.

3. And on the basis of the suspected area obtained in the step, screening a real pore area through the area size and/or shape characteristics. Since the pore defects are approximately circular, roundness is used as a main shape feature for shape screening. The roundness characteristics of a region are calculated as follows:

C＝mi n(1,C0)

where a represents the area of the pseudo-highlight region, dmax represents the maximum distance from the center of the pseudo-highlight region to the edge of the region, C0 is the initial circularity, and C is the circularity obtained finally.

Cracking:

the surface of the ceramic substrate has transverse and longitudinal nicks, so that internal cracks and complete cracking cracks can be generated in the production and transportation processes, the internal cracks refer to that the ceramic substrate does not generate cracks penetrating through the surface of the ceramic substrate, but cracks are generated in the ceramic substrate, and a black linear defect is represented on a first image of a backlight; a full crack refers to a through-surface crack in the ceramic substrate, which appears as a white crack in the first image. As shown in fig. 6, the left graph of fig. 6 shows an internal crack as a black line, and the right graph shows a complete crack as a white split. In order to eliminate notch interference of the ceramic substrate, crack detection of the ceramic substrate is carried out on the back surface of the ceramic substrate. The detection method is similar to that of the air hole, firstly, the gray background is smoothed, then, the crack area is extracted by utilizing the gray difference, and the real crack area is extracted according to the linear shape characteristics of the crack, wherein the gray linear area is an internal crack, and the highlight linear area is a through opening crack. The specific implementation method refers to the detection method of the air holes, and is not described herein again.

extracting a main body region of the ceramic substrate from the first image;

Specifically, the second defect may be an edge defect, where the edge defect refers to an edge gap or a loss of four corners at the edge or the four corners of the ceramic substrate due to a collision or the like, as shown in fig. 7, the left diagram shows a loss of the upper right corner of the ceramic substrate, and the right diagram shows a loss of the left edge of the ceramic substrate. It should be noted that, ceramic substrates of different models have 1 to 2 notches at the fixed position of one edge thereof, and are used for identifying the ceramic substrates of different models, and the ceramic substrates need to be removed when detecting defects. The notch at the lower part of the left edge of the right drawing in fig. 7 is the model mark notch, and only the smaller notch at the upper part is a real defect in the drawing. The general steps for detecting defect defects are as follows:

1. a main body region of the ceramic substrate is extracted from the first image, and the ceramic substrate is divided into 3 major regions, a corner region, an edge region and a model identification region according to different specifications and models of the ceramic substrate.

2. And respectively extracting edges from corner regions of the four corners and calculating intersection points of edge straight lines. As shown in fig. 1, the four corners of the ceramic substrate have chamfers, and the chamfers are different in size, so that the theoretical blank area WD is obtained by subtracting the real corner area of the ceramic substrate main body from the calculated right-angle area, as shown in fig. 8, the white triangular area in fig. 8 is the calculated WD area, and then whether the four corners of the ceramic substrate have defects can be determined according to the area and shape of the WD area.

In some embodiments, the method for selecting the corresponding first image and/or second image for analysis processing according to the image surface features of different surface defects further includes:

extracting a main body region of the ceramic substrate from the first image;

extracting edge points from the edge region of the main body region, obtaining a fitting straight line by using the edge points, and judging whether a second defect exists according to the distance from discrete points in the edge points to the fitting straight line.

Specifically, based on the foregoing embodiment, when calculating the edge region defects of the four sides, the edge points of the ceramic substrate are extracted from the first image, the edge points are used to fit a straight line by the least square method, then the distance between the discrete points on the edge and the fitted straight line is calculated, and whether the edge defect exists around the point is determined according to the threshold value of the distance. As for the notch area of the ceramic substrate marking the model, the notch area can be eliminated through affine transformation according to the four corners of the image and the fixed position of the wafer where the image is located. The least square method is calculated as follows:

setting a least square linear target equation as follows:

f(x)＝kx+b

according to the least-squares principle, the square of the error is

Where y (x) is the set of extracted true edge contour points. Respectively calculating the partial derivatives of k and b according to the criterion of the minimum sum of squares of the errors and making the partial derivatives zero

Solving to obtain:

/>

after the linear equation is calculated, the distance from no point on the edge to the straight line is calculated, and defect judgment is made according to the calculated distance result.

and judging whether a third defect exists or not by communicating the bright area and the dark area to judge whether a communicated domain is formed in pair.

In particular, the third defect may be adhesion and scar. The adhesion defect of the ceramic substrate refers to fine particles sintered on the surface of the ceramic substrate due to the fact that particle impurities attached to the surface are not processed in time in the production process of the ceramic substrate, and the fine particles are represented as small bulges on the smooth surface of the ceramic substrate. In contrast to adhesion, scratches are scratches and scratches on the surface of a ceramic substrate during the manufacturing process, which are expressed as small pits or ravines on the smooth surface of the ceramic substrate. When the image is collected by adhesion and scars, a low-angle surface light source is adopted, when the low-angle light source irradiates adhesion (protrusion) on a plane, one side close to the light source is high in brightness, and the other side far away from the light source is dark in gray; in contrast, a flaw defect appears to be dark on the side near the light source and bright on the side away from the light source. Similar to the effect of shining sunset afterglow on mountains (adhesion) and rivers or canyons (scars), as shown in fig. 9, the left drawing in fig. 9 is the adhesion defect on the back surface of the ceramic substrate, and the right drawing is the scar defect on the front surface of the ceramic substrate, so the adhesion and scars are illustrated together because their appearance on the image is similar, and the processing method is also similar. The specific analysis and detection steps comprise:

traditional vision algorithms deal with adhesions and scars:

according to the gray features of 'light and shade adjacent' of the adhesion and the scars on the image, bright and dark areas under a smooth background can be extracted according to a method for detecting similarity of air holes and cracks, except that after the bright and dark areas are extracted, neighborhood judgment needs to be carried out on the bright and dark areas, the bright and dark areas are communicated through gray morphological operation, the areas of the communicated areas can be formed in a light and shade pair mode, and the adhesion and the scars are preliminarily judged.

In some embodiments, the method for analyzing and processing the second image further includes:

Specifically, the model is segmented by a deep learning algorithm to process adhesion and scars:

most of adhesion and scar areas can be processed through the traditional vision by the aid of regional shape characteristics of a light and shade connected domain, but adhesion and scar type defect samples encountered by continuous operation of equipment on a production line are increased continuously, the morphology is different, and the requirements are difficult to meet by simple traditional vision algorithms and parameter adjustment, so that a deep learning algorithm is introduced, a segmentation model which is subjected to fine tuning based on an FCN (fuzzy c-means) model in deep learning is used, pixel-level classification is realized, model parameters are continuously fine-tuned along with the increase of training samples, the detection capability of the segmentation model is stronger and stronger, the segmentation model is matched with the traditional vision algorithm, and the detection requirement of the equipment is finally met.

In some embodiments, before detecting the second image using the segmentation model, further comprising:

Specifically, classification of adhesion defects, flaw defects, and false detection areas:

the adhesion and the scar areas detected by the traditional algorithm are only distinguished by the light source polishing direction at first, but the defects are different in shape, such as mountains and canyons on the ground in a map, the walking directions are various, and the defects cannot be distinguished by simply depending on the 'positive surface' and the 'negative surface', so that the classification of the adhesion and the scar defects is disordered. In addition, the traditional algorithm and the deep learning segmentation model can introduce partial false detection, so that the OK area, the adhesion area and the scar area of the false detection need to be finely classified. In order to finish classification work, the most adept classification method in the field of deep learning is introduced, a classification model aiming at three types of images including adhesion, scars and over-inspection OK products is trained on the basis of a classical ResNet50 model, and the defects of the adhesion and the scars are detected and classified in two steps, so that the false inspection rate is greatly reduced, and the classification accuracy of the defects of the adhesion and the scars is remarkably improved.

The advantages of the present application further include:

1. the light source combination mode is flexible, different light sources can be flexibly matched according to the surface defect characteristics of different types of ceramic substrate products, and multiple images under different light source illumination conditions can be acquired by one camera in a time-sharing mode by combining the inherent image acquisition characteristics of the linear array camera, so that the defects of different types can be detected by selecting proper images.

2. The detection speed is high, and the linear array camera is matched with various light sources to acquire a plurality of images in a time-sharing manner, so that the detection speed of the equipment can reach more than 65 images per minute.

3. The invention can realize the detection of all the defects of the front and the back of the product only by matching two cameras with one film turning mechanism, thus only one station is needed to be turned over. If an area-array camera is used with different light sources, multiple stations are required.

4. The equipment is matched with a high-performance GPU, and a deep learning algorithm is adopted, so that the detection of difficult defects such as irregular scars, irregular adhesion and the like which are difficult to solve by the traditional visual algorithm is solved.

Fig. 10 is a block diagram of a ceramic substrate surface defect visual inspection apparatus according to an embodiment of the present disclosure. The device comprises:

the device comprises a conveying device 1, a detection device and a control device, wherein the conveying device 1 is used for driving a ceramic substrate 4 to be detected to move at a constant speed, a through hole is formed in the conveying device, and the ceramic substrate 4 is positioned above the through hole;

the linear array camera 5 is arranged above the through hole and used for collecting images of the ceramic substrate 4;

the first stroboscopic light source 3 and the second stroboscopic light source 2 are arranged, the first stroboscopic light source 3 is arranged below the through hole, and the second stroboscopic light source 2 is arranged obliquely above the through hole;

the light source controller 6 is connected with the first stroboscopic light source 3 and the second stroboscopic light source 2 respectively, and is used for determining stroboscopic time sequences of the first stroboscopic light source 3 and the second stroboscopic light source 2 based on the line frequency of the linear array camera 5;

the sheet turning mechanism is used for turning the ceramic substrate 4;

and the analysis system 7 is used for acquiring the images acquired by the linear array camera 5, and analyzing and processing the images according to the surface characteristics of the images with different surface defects to obtain the detection result of the surface defects.

For the specific implementation method, reference is made to the foregoing method embodiments, which are not described herein again.

The present application may be methods, apparatus, systems, and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied thereon for carrying out various aspects of the present application.

The computer readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic memory device, a magnetic memory device, an optical memory device, an electromagnetic memory device, a semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or electrical signals transmitted through electrical wires.

The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device over a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in the respective computing/processing device.

Computer program instructions for carrying out operations of the present application may be assembler instructions, instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, the electronic circuitry can execute computer-readable program instructions to implement aspects of the present application by utilizing state information of the computer-readable program instructions to personalize the electronic circuitry, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA).

Various aspects of the present application are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions.

These computer-readable program instructions may be provided to a processing unit of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processing unit of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer-readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

It is noted that, unless expressly stated otherwise, all features disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose. Thus, unless expressly stated otherwise, each feature disclosed is one example only of a generic series of equivalent or similar features. Where used, further, preferably, still further and more preferably is a brief introduction to the description of the other embodiment based on the foregoing embodiment, the combination of the contents of the further, preferably, still further or more preferably back strap with the foregoing embodiment being a complete construction of the other embodiment. Several further, preferred, still further or more preferred arrangements of the back tape of the same embodiment may be combined in any combination to form a further embodiment.

Although the present application has been described in detail with respect to the general description and the specific embodiments, it will be apparent to those skilled in the art that some modifications or improvements may be made based on the present application. Accordingly, such modifications and improvements are intended to be within the scope of this invention as claimed.

Claims

1. A visual inspection method for surface defects of a ceramic substrate is characterized by comprising the following steps:

driving a ceramic substrate to be detected to move at a constant speed through a conveying device, wherein a through hole is formed below the ceramic substrate;

and selecting the corresponding first image and/or second image for analysis according to the surface characteristics of the images with different surface defects to obtain the detection result of the surface defects.

2. The method according to claim 1, wherein the step of selecting the corresponding first image and/or second image for analysis based on the image surface characteristics of different surface defects comprises:

3. The method according to claim 1, wherein selecting the corresponding first image and/or second image for analysis based on the image surface characteristics of the different surface defects comprises:

extracting a main body region of the ceramic substrate from the first image;

4. The method according to claim 1 or 3, wherein the method of selecting the corresponding first image and/or second image for analysis processing based on the image surface characteristics of the different surface defects further comprises:

extracting a main body region of the ceramic substrate from the first image;

5. The method according to claim 1, wherein selecting the corresponding first image and/or second image for analysis based on the image surface characteristics of the different surface defects comprises:

carrying out filtering smoothing processing on the second image to obtain a smooth background image, and extracting a bright area and a dark area from the smooth background image;

6. The method according to claim 1 or 5, wherein the method of selecting the corresponding first image and/or second image for analysis processing based on the image surface characteristics of the different surface defects further comprises:

7. The ceramic substrate surface defect visual inspection method of claim 6, further comprising, prior to inspecting said second image using said segmentation model:

8. The ceramic substrate surface defect visual inspection method of claim 1,

the conveying device is a flexible conveying belt;

the second stroboscopic light source is a three-color light source and can stroboscopic to emit red, green and blue light;

the second image comprises a red light image, a green light image, and/or a blue light image;

9. A ceramic substrate surface defect visual inspection device, comprising:

the conveying device is used for driving the ceramic substrate to be detected to move at a constant speed, the conveying device is provided with a through hole, and the ceramic substrate is positioned above the through hole;

the sheet turning mechanism is used for turning the ceramic substrate;

10. A computer storage medium on which a computer program is stored, the computer program, when executed by a machine, implementing the steps of a method according to any one of claims 1 to 8.