TWI765442B - Method for defect level determination and computer readable storage medium thereof - Google Patents

Abstract

The present invention provides a method for detect level determination comprising: obtaining at least an image of an object to be detected, extracting target defect area from the image, obtaining a first feature parameter from the target defect area, obtaining a weighting value according to the first feature parameter based on a predetermined rule, obtaining a second feature parameter from the target defect area, calculating a defect grade based on the second feature parameter and the weighting value, determining the defect level corresponding to the target defect area based on the defect grade, and outputting the defect level. The present invention also provides a computer readable medium. The present invention can be used to increase the accuracy and efficiency of defect level determination.

Description

Method and storage medium for determining defect level

The invention relates to the technical field of image recognition, in particular to a method and a storage medium for determining a defect level.

The current product manufacturing industry is developing in the direction of high precision and high quality. Precision metal parts are prone to various defects such as bumps, crushes, and abrasions during processing, and the size of the defects reaches the micron level. Therefore, product testing is more important and difficult. In order to meet the requirements of appearance quality, automatic optical inspection (Automated Optical Inspection, AOI) equipment needs to accurately classify and grade the detected defects.

For the classification of defects, due to the randomness of the type of defects and the limitation of AOI, the traditional defect classification method adopts the method of machine vision appearance defect detection. However, the types of flaws are very variable, and there is no algorithm in the prior art that can cover all types of flaws to accurately detect flaws. Therefore, the existing image processing technology is not very accurate for such a variety of detection targets.

In view of the above problems, the present invention proposes a method and a computer-readable storage medium for determining a defect level, so as to improve the efficiency and accuracy of determining the level of a target defect area in an image.

A first aspect of the present application provides a method for determining a defect level, which is executed by a processor of an electronic device, the method comprising: acquiring at least one image of an object to be measured; extracting the target defect area from the image; obtaining the first characteristic parameter of the target defect area; obtaining a weighted value according to the first characteristic parameter according to a preset rule; obtaining the second characteristic parameter of the target defect area; Calculate a defect specification score according to the second characteristic parameter and the weighted value; determine a defect level corresponding to the target defect area according to the defect specification score; and output the defect level.

Preferably, the extracting the target defect region from the image comprises: extracting a plurality of target defect sub-regions in the image; and processing the plurality of target defect sub-regions to obtain the target defect region.

Preferably, the processing of the plurality of target defective sub-regions to obtain the target defective sub-regions comprises: judging the types of the plurality of target defective sub-regions; determining the positions of the plurality of target defective sub-regions; Two or several of the target defect sub-regions located adjacent to each other generate the target defect region.

Preferably, the judging the types of the multiple target defect sub-regions includes: judging the types of the multiple target defect sub-regions through a convolutional neural network model.

Preferably, the first characteristic parameters of the target defect area include any one or more of the following: position, grayscale difference, aspect ratio, grayscale variance, grayscale mean difference, contrast, small gradient dominance, large gradient dominance, grayscale Degree distribution inhomogeneity, gradient distribution inhomogeneity, energy, gray mean, gradient mean, gray mean square error, gradient mean square error, correlation, gray entropy, gradient entropy, hybrid entropy, difference moment, and inverse difference moment.

Preferably, the method further includes: It is judged whether the position of the target defect area is within a preset position range, and it is judged whether the grayscale difference of the target defect area is less than a preset value.

Preferably, the method further includes: when the position of the target defect area is not within a preset position range, or the grayscale difference of the target defect area is greater than or equal to the preset value, acquiring the target defect area the second characteristic parameter; using the second characteristic parameter as the defect specification score; and determining the defect level corresponding to the target defect area according to the defect specification score.

Preferably, the second characteristic parameter includes the size of the target defect region.

Preferably, the calculating the weighted value according to the preset rule according to the first feature parameter includes: judging whether the first feature parameter complies with a plurality of condition combinations in the preset rule; when the first feature parameter satisfies When any one of the preset rules is combined, the weight of the first characteristic parameter under the condition combination is set to 1; when the first characteristic parameter does not conform to any one of the preset rules When conditions are combined, set the weight value of the first characteristic parameter under the condition combination to 0; add up and calculate the weights of the first characteristic parameter conforming to multiple condition combinations in the preset rule to obtain a weighted value value.

Preferably, calculating the defect specification score according to the second characteristic parameter and the weighted value includes: calculating the defect specification score by using a first formula, and the first formula is: defect specification score=second characteristic parameter*(1 +Total weighted value*m), where m is a threshold.

Preferably, the determining the defect level corresponding to the target defect area according to the defect specification score includes: comparing the defect specification score with one or more preset defect level thresholds to obtain a comparison result; The defect level corresponding to the target defect area is determined according to the comparison result.

Preferably, the method further includes: according to the defect specification score of the target defect area, confirming whether to perform defect level judgment on the image again.

Preferably, the method further comprises: setting a preset threshold range according to the score range corresponding to the pre-stored defect level; comparing the defect specification score of the target defect area with the preset threshold range; and when the defect specification score When it is within the preset threshold range, the image is again subjected to flaw level determination.

Preferably, the image is again subjected to flaw level determination through a deep learning algorithm.

A second aspect of the present invention provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the aforementioned method for determining a defect level is implemented.

The method and computer-readable storage medium for determining the defect level of the present invention extract the target defect area in the image to be identified, and then calculate the target defect area according to the first characteristic parameter and the second characteristic parameter of the target defect area. Spec Score. Finally, the defect grade is obtained according to the specification score. Further, by setting the qualified threshold range, and then according to the specification score of the target defect area, it is determined whether the defect will be judged by the deep learning technology after the image processing technology is judged. The determination of deep learning technology greatly reduces the detection time, and also reduces the computing requirements of the computing host, thereby saving the construction cost of hardware.

S1~S8: Steps

2: Image

20: First scratch

21: The first strip

22: The second strip

23: The third strip

24: Fourth strip

30: First scratch

301: First area

302: Second area

303: The third area

304: Fourth area

31: Second Scratch

311: Fifth Region

312: The sixth area

200: Defect level determination system

201: Get Mods

202: Extract the module

203: Computing Modules

204: Defective Level Preliminary Judgment Module

205: Output module

206: Defect level re-judgment module

10: Electronics

11: Memory

12: Processor

13: Computer Programs

FIG. 1 is a schematic flowchart of a method for determining a defect level according to an embodiment of the present invention.

FIG. 2 is a schematic diagram of a first scratch in an image 2 provided by an embodiment of the present invention.

FIG. 3 is a schematic diagram of a first scratch in image 2 provided by an embodiment of the present invention.

4 is a schematic diagram of calculating the weights of the characteristic parameters conforming to a plurality of condition combinations in the preset rule according to the summation of the characteristic parameters of the target defect area according to an embodiment of the present invention.

FIG. 5 is a schematic diagram of a defect level determination system provided by an embodiment of the present invention.

FIG. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments may be combined with each other in the case of no conflict.

In the following description, many specific details are set forth in order to facilitate a full understanding of the present invention, and the described embodiments are only some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terms used herein in the description of the present invention are for the purpose of describing specific embodiments only, and are not intended to limit the present invention.

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of a method for determining a defect level according to an embodiment of the present invention. According to different requirements, the order of the steps in the flowchart can be changed, and some steps can be omitted. For the convenience of description, only the parts related to the embodiments of the present invention are shown.

As shown in FIG. 1 , the defect level determination method includes the following steps.

Step S1, acquiring at least one image of the object to be measured.

In one embodiment, at least one image can be obtained by photographing the object to be measured by a camera. The camera may be a line scan camera, and the object to be measured is a device such as a mobile phone or a tablet computer. In another embodiment, at least one image of the object to be measured transmitted by the server can be received. In other embodiments, at least one image of the object to be measured can be acquired from a local database. In this embodiment, Fig. The image can include a complete or partial image of the object to be measured. The image can be of any resolution, and can also be sampled high or low, depending on actual needs.

Step S2, extracting the target defect area from the image.

In one embodiment, multiple target defect regions may be included in the image. Each target defect area in the image is selected with a rectangular frame, and an image including multiple target defect areas can be obtained. The target defect area includes a plurality of target defect sub-areas.

It should be noted that the multiple target defect sub-regions in the image may be different types of defects. For example, scratch type, scratch type, bump type, and stain type. The number of defects of each type in the image can also be multiple. In order to facilitate processing, the image can be first divided into multiple target defect sub-regions, then the multiple target defect sub-regions can be classified, and the same and adjacent multiple target defect sub-regions can be aggregated into target defect sub-regions. Each target defect area in each type of target defect area is graded by the defect grade determination method described in this application.

In one embodiment, the step of extracting the target defect region from the image includes:

(1) Filter the image;

In one embodiment, the image may be filtered using image processing techniques to remove noise in the image. The filtering method may be, but not limited to, any one of median filtering, mean filtering, Gaussian filtering, and bilateral filtering. By filtering the image, the influence of noise can be eliminated. In a preferred embodiment, a median filtering or mean filtering technique is used to filter the image.

(2) Extracting multiple target defect sub-regions in the image;

In one embodiment, multiple target defect sub-regions in the image can be extracted by using image processing technology, and the image processing technology can be Region of Interest (ROI) algorithm, semantic segmentation algorithm, binarization and the like.

(3) Processing a plurality of target defect sub-regions to obtain a target defect region.

In one embodiment, different processing methods may be used according to the type and position of the defect in the image to obtain the target defect area. Specifically, the types of multiple target defect sub-regions may be first judged, the positions of the multiple target defect sub-regions may be determined, and two or more target defect sub-regions of the same type and adjacent positions may be aggregated to generate the target defect areas.

In one embodiment, the types of multiple target defect sub-regions can be determined through a convolutional neural network model.

For example, as shown in FIG. 2 , the plurality of target defect sub-regions in the image include a first strip 21 , a second strip 22 , a third strip 23 and a fourth strip 24 . It is confirmed that the type of the plurality of target defect regions is a scratch type, and the positions of the plurality of target defect sub-regions in the image are confirmed. From this, it can be known that the target defect area in the image may be a scratch type area, and the target defect area includes a plurality of continuous and closely adjacent small scratches. That is, as shown in FIG. 2 , the first scratch 20 includes a first elongated strip 21 , a second elongated strip 22 , a third elongated strip 23 and a fourth elongated strip 24 . In order to avoid mistaking the first long strip 21 , the second long strip 22 , the third long strip 23 and the fourth long strip 24 as four scratches, the first long strip 21 , the second long strip 22 , the third long strip 22 and the third The long strip 23 and the fourth long strip 24 are fitted to obtain the target defect area, that is, the first scratch 20 .

For another example, as shown in FIG. 3 , the multiple target defect sub-regions in the image include a first region 301 , a second region 302 , a third region 303 , a fourth region 304 , a fifth region 311 and a sixth region 312 . The type of the plurality of target defect regions is confirmed as a scratch type, and the positions of the plurality of target defect sub-regions in the image are confirmed. It can be seen from this that the target defect regions in the image may be two scratch-type regions, and the two target defect regions respectively include a plurality of continuous and closely adjacent small scratches. That is, as shown in FIG. 3 , the first scratch 30 includes a first area 301 , a second area 302 , a third area 303 and a fourth area 304 . The second scratch 31 includes a fifth area 311 and a sixth area 312 . In order to avoid mistaking the first area 301 , the second area 302 , the third area 303 , the fourth area 304 , the fifth area 311 and the sixth area 312 as six scratches. The first area 301, the second area 302, the third area 303 and the fourth area 304 need to be The clustering results in the target defect area, the first scratch 30 . The fifth area 311 and the sixth area 312 are clustered to obtain the target defect area, that is, the second scratch 31 .

Preferably, the step of extracting the target defect area from the image may further include: filtering out defect areas whose size is smaller than a preset size in the target defect area.

Since there may be dust on the object to be measured, short wool and other substances that are usually invisible to the naked eye of the user. When taking images of the object to be measured, dust/short hairs may be mistaken for defects. Therefore, it is necessary to remove small defects in the target defect area. In one embodiment, by comparing the size of the target defect area with the preset size, the target defect area whose size is smaller than the preset size is filtered out. For example, the area of the target defect area is compared with the preset area, and the target defect area with an area smaller than the preset area is filtered out.

Step S3, acquiring the first characteristic parameter of the target defect area.

In one embodiment, the image processing technology can be used to extract the first characteristic parameter of the target defect area. The first characteristic parameters of the target defect area include any one or more of the following: position, grayscale difference, aspect ratio, grayscale variance, grayscale mean difference, contrast, small gradient dominance, large gradient dominance, and grayscale distribution inhomogeneity , gradient distribution inhomogeneity, energy, gray mean, gradient mean, gray mean square error, gradient mean square error, correlation, gray entropy, gradient entropy, hybrid entropy, difference moment, and inverse difference moment.

Position is the position of the target defect area in the image. The grayscale difference is the grayscale difference between the target defect area and the image background. The grayscale variance is the grayscale variance between the target defect area and the image background. The gray mean difference is the mean gray difference between the target defect area and the image background. Contrast is the contrast between the target defect area and the background of the image.

Step S4, calculating a weighted value according to a preset rule according to the first characteristic parameter.

In one embodiment, it can be determined whether the first characteristic parameter complies with a plurality of condition combinations in the preset rule. When the first characteristic parameter conforms to any condition combination in the preset rule, the weight of the first characteristic parameter under the condition combination is set to 1. When the first characteristic parameter does not meet the preset rules For any combination of conditions, set the weight of the first feature parameter to 0 under the combination of conditions. A weighted value is obtained by adding up and calculating the weights of the first characteristic parameters that meet the multiple condition combinations in the preset rule.

For example, as shown in FIG. 4 , the multiple condition combinations in the preset rule include a first condition combination, a second condition combination... an Nth condition combination. Each condition combination includes multiple characteristic parameter conditions. For example, the first combination of conditions includes that the grayscale difference is greater than x1, the aspect ratio is greater than y1, and the grayscale variance is greater than z1. The second combination of conditions includes that the grayscale difference is greater than x2, the aspect ratio is greater than y2, and the grayscale variance is greater than z2. The Nth condition combination includes that the grayscale difference is greater than xn, the aspect ratio is greater than yn, and the grayscale variance is greater than zn.

It should be noted that, each condition combination in the multiple condition combinations may also include other characteristic parameter conditions. For example, the gray mean difference between the target defect area and the image background is greater than w1, the contrast between the target defect area and the image background is greater than v1, and so on.

When the first characteristic parameter meets all the conditions in the first condition combination, a weight value of 1 is obtained. When the first characteristic parameter does not meet any condition in the first condition combination, the obtained weight is 0. For example, when the grayscale difference in the first feature parameter is greater than x1, the aspect ratio is greater than y1, and the grayscale variance is greater than z1, the weight value is 1. When the grayscale difference in the first feature parameter is less than or equal to x1, or the aspect ratio is less than or equal to y1, or the grayscale variance is less than or equal to z1, the weight value is 0. By analogy, the weights of the first characteristic parameters meeting the second condition and the weights of the first characteristic parameters meeting the Nth condition can be obtained respectively. Then, the weights of the first condition, the weights of the second condition...the weights of the Nth condition are summed to obtain the weighted value.

Step S5, acquiring the second characteristic parameter of the target defect area.

In one embodiment, the second characteristic parameter includes the size of the target defect region. In other embodiments, the second feature parameter may further include other features of the target defect region, such as aspect ratio, distribution area, average gray value, average gradient, and the like.

In one embodiment, the size of the target defect area may vary depending on the type of defect. For example, when the type of target defect area is scratch type, the size of the target defect area can be is the length of the flaw. When the type of the target defect area is a scratch type, the size of the target defect area may be the area of the defect.

Step S6: Calculate the defect specification score according to the second characteristic parameter and the weighted value.

In one embodiment, calculating the defect specification score according to the second characteristic parameter and the weighted value includes: calculating the defect specification score by using a first formula, where the first formula is: defect specification score=second characteristic parameter*(1+ Weighted value*m). Wherein, the m is a preset threshold.

Preferably, the defect level determination method may further include: judging whether the position of the target defect area is within a preset position range, and judging whether the grayscale difference of the target defect area is less than a preset value.

In this embodiment, when the position of the target defect area is not within the preset position range, and the grayscale difference of the target defect area is greater than or equal to the preset value, it is confirmed that the target defect area is an easily identifiable defect. When calculating the defect specification score of the target defect area, the second feature parameter is directly used as the defect specification score. Then, the defect grade corresponding to the target defect area is determined according to the defect specification score. That is, when the position of the target defect area is not within the preset position range, and the grayscale difference of the target defect area is greater than or equal to the preset value, the defect specification score need not be calculated by considering the first characteristic parameter. That is, the weighted value calculated according to the preset rule according to the first feature parameter is zero, and the second feature parameter is directly used as the defect specification score. Then, the defect grade corresponding to the target defect area is determined according to the defect specification score.

When the position of the target defect area is within the preset position range, or the grayscale difference of the target defect area is less than the preset value, it is confirmed that the target defect area is a defect that is not easy to identify. In order to avoid missing targeted defect areas, it is not possible to accurately identify severe scratches in the image. When calculating the defect specification score of the target defect area, the first characteristic parameter needs to be considered to calculate the defect specification score, that is, the defect specification score is calculated according to the second characteristic parameter and the total weighted value. Then, the defect grade corresponding to the target defect area is determined according to the defect specification score.

In one embodiment, whether the target defect area is located in a preset position range can be determined by the coordinates of the target defect area. The preset position range can refer to the bending position in the object to be measured. area, or other areas that are prone to image distortion when taking images. For example, when the object to be measured is a mobile phone, the mobile phone case has four end points, and each end point can be in the shape of a circular arc. Since the defects at the four arc-shaped endpoints are prone to misjudgment when the object to be measured is photographed, the four arc-shaped endpoints are used as the preset position range.

In one embodiment, it can be determined whether the target defect area is located in the preset position range by judging whether the coordinates (X1, Y1) of the target defect area match the coordinates (X2, Y2) of the preset position range. When the coordinates (X1, Y1) of the target defect area match the coordinates (X2, Y2) of the preset position range, it is confirmed that the target defect area is located in the preset position range. When the coordinates (X1, Y1) of the target defect area do not match the coordinates (X2, Y2) of the preset position range, it is confirmed that the target defect area is not located in the preset position range.

Specifically, a Cartesian coordinate system (XOY) is established in the image with the point at the upper left corner of the image as the origin, wherein the X direction of the Cartesian coordinate system represents the width of the image, and the Y direction of the Cartesian coordinate system represents the width of the image. The height of the image. The coordinates (X1, Y1) of the target defect area correspond to the primitive points in the image. The coordinates (X2, Y2) of the special area in the image are determined.

It can be understood that the coordinates of the preset position range are a range of coordinates. In one embodiment, the matching of the coordinates (X1, Y1) of the target defect area with the coordinates (X2, Y2) of the preset position range may mean that when "X1" is located in the interval [X2-M, X2+M], and When "Y1" is located in the interval [Y2-N, Y2+N], it can be determined that the coordinates (X1, Y1) of the target defect area match the coordinates (X2, Y2) of the preset position range. The values of "M" and "N" can be preset.

In one embodiment, an image processing technology can also be used to obtain the grayscale difference of the target defect area as a weighting condition. By judging whether the grayscale difference of the target defect area is smaller than the preset value, when the grayscale difference is smaller than the preset value, it is confirmed that the target defect area is a defect that is not easy to identify. For example, when the object to be tested is severely scratched, the severe scratches in the image captured by the line scan camera should be displayed as black. However, due to the process of shooting the object to be measured by the line scan camera, light reflection occurs. lead to beat Severe scratches are shown in light grey in the captured image. In order to avoid missing the light gray defects and unable to accurately find serious scratches in the image, it is defined that when the grayscale difference between the target defect area and the image background is less than the preset value, the target defect area is not easy to identify. flaw.

Step S7: Determine the defect level corresponding to the target defect area according to the defect specification score.

In one embodiment, a comparison result is obtained by comparing the defect specification score with one or more preset defect level thresholds, and the defect level corresponding to the target defect area is determined according to the comparison result. The one or more preset defect level thresholds may be pre-stored in the electronic device in the form of a defect level table.

In one embodiment, the defect level table describes the relationship between the defect level and the corresponding range of defect specification scores. For example, defect grades include A grades, B grades, C grades, D grades, and E grades. The range of defect specification scores corresponding to grade A is 20~100, the range of defect specification score corresponding to grade B is 100~300, the range of defect specification score corresponding to grade C is 300~500, and the range of defect specification score corresponding to grade D is 500~ 900, the corresponding defect specification score range of E grade is 900~1000.

Compare the defect specification score of the target defect area with the defect specification score range to confirm the defect level corresponding to the target defect area. For example, when the defect specification score of the target defect area is 80, the grade of the target defect area is confirmed to be A grade.

In one embodiment, different defect level tables are pre-stored for different types of defects. For example, when the defect type is scratch type, the defect specification score in the defect grade table is the score calculated according to the length of the scratch; when the defect type is scratch type, the defect specification score in the defect grade table is based on scratches. The fraction of the area calculated.

Step S8, output the defect level.

In one embodiment, the defect level can be output to a display screen or other user interface for user reference. Defect levels can also be stored in memory or sent to a remote server.

Preferably, the defect level determination method further includes step S9 : confirming whether to perform the defect level determination on the image again according to the defect specification score of the target defect area. It can be understood that, in other embodiments, step S9 may also be omitted.

Specifically, according to the defect specification score of the target defect area, the step of confirming whether to perform the defect level judgment on the image again includes:

(1) Set the preset threshold range according to the score range corresponding to the pre-stored defect level. For example, a score of 300 in the range of scores corresponding to the defect level is set as the pass line. Then set a preset threshold range (such as 200~400) according to the qualified line

The preset threshold range is a range that can be adjusted according to requirements.

(2) Compare the defect specification score of the target defect area with the preset threshold value range, and confirm whether to perform the defect level judgment on the image again.

In one embodiment, the defect specification score of the target defect area is compared with a preset threshold range, and the correct rate of the defect level of the target defect area is confirmed. When the defect specification score of the target defect area is greater than the maximum value of the preset threshold range, or smaller than the minimum value of the preset threshold value range, the correct rate of determining the defect level of the target defect area is high, and there is no need to re-process the image. Determination of defect level; when the defect specification score of the target defect area is within the preset threshold range, the correct rate of determining the defect level of the target defect area is low.

(3) When the defect specification score of the target defect area is within the preset threshold range, perform defect level determination on the image again.

For example, when the defect specification score is greater than or equal to 400, there is a high probability that there is a real defect, so the defect level judgment accuracy of this defect specification score is high, and the defect judgment result is the original defect level. Or when the defect specification score is less than or equal to 200, there is a high probability that there is no defect, so the defect level judgment accuracy of this defect specification score is high, and the defect judgment result is the original defect level. When the defect specification score of the target defect area is greater than 200 and less than 400, the defect specification score within this range The accuracy rate of the judgment of the defect grade of the number of defects is low, and it is impossible to determine whether the defect grade of the corresponding target defect area is qualified or unqualified.

In one embodiment, the image is again subjected to defect level determination through a deep learning algorithm.

In one embodiment, a preset threshold range can be set to determine whether the judgment result of the image processing technology based on the defect specification score needs to be further judged by the deep learning technology, and it is not necessary that all the defects be judged by the deep learning technology. , which greatly reduces the detection time, and also reduces the computing requirements of the computing host, thereby saving the construction cost of hardware.

Therefore, in order to further improve the accuracy and efficiency of judging the defect level in the image, in one embodiment, after determining the preliminary level of the target defect area in the image, the method for judging the defect level of the present application may further include step S10: When the defect specification score of the target defect area is within the preset threshold range, the image is again subjected to defect level judgment. It can be understood that, in other embodiments, the step S10 may also be omitted.

In this application, for defects located in special areas (for example, where the object to be tested is curved, the captured image of defects will be distorted) and defects of special types (the types of defects that are easy to be identified by image processing technology), image-based The weighted calculation is added to the judgment of the processing technology, and a preset threshold range is set to judge the calculated value, so as to decide whether to carry out a rejudgment.

After the defect is judged by the image processing technology, if the accuracy is found to be low after calculation, it means that its characteristics are less in line with the setting of the detection algorithm, but because the image processing algorithm cannot cover all types of defects, the depth is used. Learning technology is used to assist judgment. Deep learning technology uses a large amount of real defect data to learn specific defect characteristics, so it can draw more accurate judgments.

FIG. 1 to FIG. 4 describe in detail the method for determining the defect level of the present invention, through which the efficiency and accuracy of determining the defect level can be improved. Below in conjunction with Fig. 5 and Fig. 6, to realize the described The functional modules of the software system and the hardware device architecture of the defect level determination method are introduced. It should be understood that the embodiments are only used for illustration, and are not limited by this structure in the scope of the patent application.

FIG. 5 is a structural diagram of a system for determining a defect level according to an embodiment of the present invention.

In some embodiments, the defect level determination system 200 may include a plurality of functional modules composed of program code segments. The program codes of each program segment in the defect level determination system 200 may be stored in the memory of the computer device, and executed by at least one processor in the computer device, so as to realize the function of defect level determination.

Referring to FIG. 5 , in this embodiment, the defect level determination system 200 can be divided into a plurality of functional modules according to the functions it performs, and each functional module is used to perform each step in the corresponding embodiment of FIG. 1 to achieve Defect level judgment function. In this embodiment, the flaw level determination system 200 includes an acquisition module 201 , an extraction module 202 , a calculation module 203 , a flaw level preliminary determination module 204 , an output module 205 and a flaw level re-determination module 206 . The functions of each functional module will be described in detail in the following embodiments.

The acquisition module 201 is used for acquiring at least one image of the object to be measured. The obtaining module 201 is further configured to obtain the first characteristic parameter and the second characteristic parameter of the target defect area. The first characteristic parameter includes any one or more of the following: position, grayscale difference, aspect ratio, grayscale variance, grayscale mean difference, contrast, small gradient dominance, large gradient dominance, grayscale distribution inhomogeneity, gradient distribution inhomogeneous Uniformity, Energy, Gray Average, Gradient Average, Gray Mean Square Error, Gradient Mean Square Error, Correlation, Gray Entropy, Gradient Entropy, Hybrid Entropy, Difference Moments, and Inverse Difference Moments. The second characteristic parameter includes the size, aspect ratio, distribution area, average gray value, average gradient, and the like of the target defect region.

The extraction module 202 is used for extracting target defect regions from the image.

Specifically, the extraction module 202 is used for filtering the image, extracting multiple target defect sub-regions in the image, judging the types of the multiple target defect sub-regions and determining the types of the multiple target defect sub-regions. position, aggregating two or more target defect sub-regions of the same type and adjacent in position to generate a target defect area; and filtering out defect areas whose size is smaller than a preset size in the target defect area.

The calculation module 203 is configured to perform weighted calculation according to a preset rule according to the first characteristic parameter to obtain a weighted value.

Specifically, the calculation module 203 can be used to determine whether the first characteristic parameter complies with a plurality of condition combinations in the preset rule. When the first characteristic parameter conforms to any condition combination in the preset rule, the weight of the first characteristic parameter under the condition combination is set to 1. When the first characteristic parameter does not meet any condition combination in the preset rule, the weight of the first characteristic parameter under the condition combination is set to 0. A weighted value is obtained by adding up and calculating the weights of the first characteristic parameters that meet the multiple condition combinations in the preset rule.

The calculation module 203 is further configured to calculate the defect specification score according to the second characteristic parameter and the weighted value.

The defect grade preliminary judgment module 204 is used to judge the defect grade corresponding to the target defect area according to the defect specification score.

Specifically, the defect grade preliminary judgment module 204 can compare the defect specification score with one or more preset defect grade thresholds to obtain a comparison result, and then determine the preliminary judgment defect grade corresponding to the target defect area according to the comparison result .

The output module 205 is used for outputting the defect level.

The defect grade re-judgment module 206 is used for confirming whether to perform the defect grade judgment on the image again according to the defect specification score of the target defect area. Specifically, the defect grade re-judgment module 206 sets a preset threshold range according to the score range corresponding to the pre-stored defect grade, and then compares the defect specification score with the preset threshold range. When the defect specification score is greater than the maximum value in the preset threshold range, or smaller than the minimum value in the preset threshold value range, there is no need to perform defect level judgment on the image again; when the defect specification score is within the preset threshold value range , and judge the image flaw level again.

FIG. 6 is a schematic diagram of a functional module of an electronic device according to an embodiment of the present invention. The electronic device 10 includes a memory 11 , a processor 12 , and a computer program 13 stored in the memory 11 and executable on the processor 12 , such as a program for determining defect levels.

In one embodiment, the electronic device 10 may be, but not limited to, a smart phone, a tablet computer, a computer device, and the like.

The processor 12 can execute the computer program 13 to implement the steps of the method for determining the defect level in the above method embodiment, for determining the defect level of the target defect area in the image 2 . Alternatively, the processor 12 can execute the computer program 13 to realize the functions of the modules/units in the above-mentioned system embodiments.

Exemplarily, the computer program 13 may be divided into one or more modules/units, and one or more modules/units are stored in the memory 11 and executed by the processor 12 . The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 13 in the electronic device 10 . For example, computer program 13 may be divided into modules 201-206 in FIG. 6 .

Those skilled in the art can understand that FIG. 6 is only an example of the electronic device 10, and does not constitute a limitation to the electronic device 10. The electronic device 10 may include more or less components than the one shown, or combine certain components, or Various components such as the electronic device 10 may also include input and output devices and the like.

The so-called processor 12 may be a central processing unit (Central Processing Unit, CPU), and may also include other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC) , Off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor 12 is the control center of the electronic device 10 , and uses various interfaces and lines to connect various parts of the entire electronic device 10 .

The memory 11 can be used to store computer programs 13 and/or modules/units. The processor 12 executes or executes the computer programs and/or modules/units stored in the memory 11 and calls the data to realize various functions of the electronic device 10 . The memory 11 may include an external storage medium or a memory. In addition, the memory 11 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (Secure Digital, SD) card, flash memory card (Flash Card), at least one disk memory device, flash memory device, or other volatile solid state memory device.

If the modules/units integrated in the electronic device 10 are implemented in the form of software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the processes in the methods of the above embodiments, and can also be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, so When the computer program is executed by the processor, the steps of the above method embodiments can be implemented. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be Modifications or equivalent substitutions can be made without departing from the spirit and scope of the technical solutions of the present invention.

S1~S8: Steps

Claims (14)

A defect level determination method, executed by a processor of an electronic device, is improved in that the method includes: acquiring at least one image of an object to be measured; extracting a target defect area from the image; acquiring the target defect obtain the weighted value according to the preset rule according to the first characteristic parameter; obtain the second characteristic parameter of the target defect area; calculate the defect specification score according to the second characteristic parameter and the weighted value ; determine the defect level corresponding to the target defect area according to the defect specification score; output the defect level; and confirm whether to perform the defect level determination on the image again according to the defect specification score of the target defect area. The defect level determination method according to claim 1, wherein the extracting the target defect region from the image comprises: extracting a plurality of target defect sub-regions in the image; processing the plurality of target defect sub-regions area to obtain the target defect area. The defect level determination method according to claim 2, wherein the processing the multiple target defect sub-regions to obtain the target defect sub-regions comprises: judging the types of the multiple target defect sub-regions; determining the multiple target defect sub-regions The position of the target defect sub-region; the target defect region is generated by aggregating two or more target defect sub-regions with the same type and adjacent positions. The method for determining a defect level according to claim 3, wherein the determining the types of the multiple target defect sub-regions comprises: judging the types of the multiple target defect sub-regions through a convolutional neural network model. The defect level determination method according to claim 1, wherein the first characteristic parameters of the target defect area include any one or more of the following: position, grayscale difference, aspect ratio, grayscale variance, grayscale mean difference, Contrast, small gradient dominance, large gradient dominance, gray distribution inhomogeneity, gradient distribution inhomogeneity, energy, gray average, gradient average, gray mean square error, gradient mean square error, correlation, gray entropy, gradient entropy , mixed entropy, difference moments, and inverse difference moments. The defect level determination method according to claim 5, wherein the method further comprises: judging whether the position of the target defect area is within a preset position range, and judging whether the grayscale difference of the target defect area is smaller than a preset position value. The defect level determination method according to claim 6, wherein the method further comprises: when the position of the target defect area is not within a preset position range, or the grayscale difference of the target defect area is greater than or equal to the predetermined position When setting the value, the second characteristic parameter of the target defect area is obtained; the second characteristic parameter is used as the defect specification score; the defect level corresponding to the target defect area is determined according to the defect specification score. The method for determining a defect level according to claim 7, wherein the second characteristic parameter includes the size of the target defect area. The defect level determination method according to claim 1, wherein calculating the weighted value according to the first characteristic parameter according to a preset rule comprises: judging whether the first characteristic parameter complies with a plurality of the preset rules Condition combination; when the first characteristic parameter meets any condition combination in the preset rules, set the weight of the first characteristic parameter under the condition combination to 1; when the first characteristic parameter When it does not meet any combination of conditions in the preset rules, set the weight of the first characteristic parameter under the combination of conditions to 0; add up and calculate that the first characteristic parameter meets the preset rules. The weight of the combination of multiple conditions to obtain the weighted value. The method for determining a defect level according to claim 1, wherein calculating the defect specification score according to the second characteristic parameter and the weighted value includes: The defect specification score is calculated by a first formula, where the first formula is: defect specification score=second characteristic parameter*(1+total weighted value*m), where m is a threshold. The method for determining a defect level according to claim 1, wherein the determining the defect level corresponding to the target defect area according to the defect specification score comprises: comparing the defect specification score with one or more preset defect levels The threshold is compared to obtain a comparison result; the defect level corresponding to the target defect area is determined according to the comparison result. The method for determining a defect level according to claim 1, wherein the method further comprises: setting a preset threshold range according to a score range corresponding to a pre-stored defect level; comparing the defect specification score of the target defect area with the pre-stored defect level setting a threshold range; and when the defect specification score is within the preset threshold range, perform a defect level determination on the image again. The method for determining a defect level according to claim 1, wherein the method further comprises: performing a defect level determination on the image again through a deep learning algorithm. A computer-readable storage medium is improved in that a computer program is stored thereon, and when the computer program is executed by a processor, the method for determining a defect level according to any one of claims 1 to 13 is implemented.

Images