CN111325728A - Product defect detection method, device, equipment and storage medium - Google Patents

Abstract

The invention discloses a product defect detection method, device, equipment and storage medium, and relates to the technical field of automatic detection, wherein a product defect detection method includes: acquiring an image of a product to be detected; , obtain the first image segmentation result; use the Gaussian mixture model algorithm to process the image of the product to be detected to obtain the second image segmentation result; use the post-processing algorithm to process the first image segmentation result and the second image segmentation result to obtain the detection result. The invention can greatly improve the detection speed, is suitable for surface defect detection of different types of products, and can also reduce the influence of external factors such as different light sources, detection parts placement angles and shadow conditions on the detection results, and realize the detection of product defects in the production line. Accurate detection.

Description

Product defect detection method, device, equipment and storage medium

technical field

The present invention relates to the technical field of automatic detection, in particular to a product defect detection method, device, equipment and storage medium.

Background technique

In the process of industrial production, due to the external environment such as mechanical vibration, sound and light, as well as complex production processes, the appearance of the produced products may be poor, so that the produced products become defective products and reduce production efficiency. In order to ensure that the appearance of the product meets the corresponding production requirements, it is necessary to carry out a series of inspections on the products in the industrial production process.

At present, the detection methods of product surface defects mainly include manual detection and machine vision detection. The manual detection method has the problems of low detection efficiency and high production cost. The machine vision inspection method can realize the automation of surface defect detection, but the surface defect detection models in the existing machine vision inspection only detect a specific product surface or a specific type of defect. However, in the actual production process, a production line may produce a variety of products, and the types of surface defects of different types of products may also be different. At this time, when detecting surface defects of different products, it is necessary to design and develop different surface defect detection models. The whole process cycle is long, and the time and labor costs are also high.

SUMMARY OF THE INVENTION

The purpose of the present invention is to solve one of the technical problems existing in the prior art at least to a certain extent. Therefore, the present invention proposes a product defect detection method, which can greatly improve the detection speed and is suitable for surface defect detection of different types of products.

The invention also provides a product defect detection device.

The invention also provides a product defect detection device.

The present invention also provides a computer-readable storage medium.

In a first aspect, an embodiment of the present invention provides a product defect detection method, including:

Obtain the image of the product to be detected;

The image of the product to be detected is processed by using the Otsu algorithm to obtain the first image segmentation result;

The Gaussian mixture model algorithm is used to process the image of the product to be detected, and the second image segmentation result is obtained;

A post-processing algorithm is used to process the first image segmentation result and the second image segmentation result to obtain a detection result.

A product defect detection method according to the embodiment of the present invention has at least the following beneficial effects:

1. Can greatly improve the detection speed;

2. Suitable for surface defect detection of different types of products;

3. It can reduce the influence of external factors such as different light sources, inspection parts placement angles and shadows on the inspection results, and achieve accurate detection of product defects in the production line.

According to a product defect detection method according to other embodiments of the present invention, the Otsu algorithm is used to process the image of the product to be detected to obtain a first image segmentation result, including:

Perform grayscale processing on the image of the product to be detected to obtain a grayscale image;

Perform gamma correction on the grayscale image to obtain a gamma corrected image;

Perform Otsu segmentation on the gamma correction map to obtain the Otsu segmentation result;

Perform morphological processing on Otsu segmentation results to obtain morphological correction results;

Perform median filtering on the morphological correction result to obtain the first image segmentation result.

A product defect detection method according to an embodiment of the present invention improves the traditional Otsu algorithm. The traditional Otsu algorithm includes grayscale processing and Otsu segmentation, while the improved Otsu algorithm also includes gamma correction, morphological processing and median value. filter. Compared with the traditional Otsu algorithm, the improved Otsu algorithm is simpler to calculate, so it can detect product images quickly and stably without spending a lot of computing time, and also improves the detection accuracy.

According to a product defect detection method according to other embodiments of the present invention, morphological processing is performed on the Otsu segmentation result to obtain a morphological correction result, including:

The low-pass filtering method is used to perform grayscale smoothing and edge smoothing on the Otsu segmentation results, and the morphological correction results are obtained.

According to a product defect detection method according to other embodiments of the present invention, a Gaussian mixture model algorithm is used to process the image of the product to be detected to obtain a second image segmentation result, including:

Extract the hue saturation value (HSV) space color of the image of the product to be detected to obtain the HSV color map;

The expected maximum (EM) algorithm segmentation is performed on the HSV color image to obtain the second image segmentation result.

In a product defect detection method according to an embodiment of the present invention, a Gaussian mixture model is used to simulate the pixel value of an input image of a product to be detected, and different single Gaussian models are used to determine the type of each pixel, so as to detect the product image. , to obtain the second image segmentation result.

According to a product defect detection method according to other embodiments of the present invention, EM algorithm segmentation is performed on the HSV color image to obtain a second image segmentation result, including:

Build a Gaussian mixture model;

The Gaussian mixture model is used to process the HSV color map to obtain the model number of the Gaussian mixture model;

According to the number of models of the Gaussian mixture model, the optimal model of the Gaussian mixture model is obtained;

The EM algorithm is used to estimate the parameters of the optimized model, and the second image segmentation result is obtained.

According to a product defect detection method according to other embodiments of the present invention, a post-processing algorithm is used to process the first image segmentation result and the second image segmentation result to obtain the detection result, including:

processing the first image segmentation result and the second image segmentation result to obtain a third image segmentation result;

The morphological reconstruction method is used to detect and repair the vulnerability edge of the third image segmentation result, and the detection result is obtained.

A product defect detection method according to an embodiment of the present invention can overcome the problem that illumination makes it difficult for the EM algorithm to classify correctly through vulnerability edge detection and vulnerability repair, thereby improving the detection accuracy of the EM algorithm.

In a second aspect, an embodiment of the present invention provides a product defect detection device, including:

The image acquisition module is used to acquire the image of the product to be detected;

a first image segmentation module, used to process the image of the product to be detected by using the Otsu algorithm to obtain a first image segmentation result;

The second image segmentation module is used to process the image of the product to be detected by using a Gaussian mixture model algorithm to obtain a second image segmentation result;

The post-processing module is used for processing the first image segmentation result and the second image segmentation result by using a post-processing algorithm to obtain a detection result.

A product defect detection device according to an embodiment of the present invention has at least the following beneficial effects:

1. The segmentation results of the first image segmentation module are stable and fast, and the segmentation results of the second image segmentation module are accurate, fast in calculation, and do not require a large number of training samples;

2. The post-processing module processes the first image segmentation result and the second image segmentation result, which can not only detect the product image reliably, but also greatly improve the detection speed;

3. Suitable for surface defect detection of different types of products;

4. It can reduce the influence of external factors such as different light sources, inspection parts placement angles and shadows on the inspection results, and achieve accurate detection of product defects in the production line.

According to a product defect detection device according to other embodiments of the present invention, the post-processing algorithm includes vulnerability edge detection and vulnerability repair.

In a third aspect, an embodiment of the present invention provides a product defect detection device, including:

at least one processor, and,

a memory communicatively coupled to the at least one processor; wherein,

The memory stores instructions executable by the at least one processor to enable the at least one processor to perform a product defect detection method according to some embodiments of the present invention.

A product defect detection device according to an embodiment of the present invention has at least the following beneficial effects:

1. Can greatly improve the detection speed;

2. Suitable for surface defect detection of different types of products;

3. It can reduce the influence of external factors such as different light sources, inspection parts placement angles and shadows on the inspection results, and achieve accurate detection of product defects in the production line.

In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute one of some embodiments of the present invention Product defect detection methods.

A computer-readable storage medium according to an embodiment of the present invention has at least the following beneficial effects:

1. Can greatly improve the detection speed;

2. Suitable for surface defect detection of different types of products;

3. It can reduce the influence of external factors such as different light sources, inspection parts placement angles and shadows on the inspection results, and achieve accurate detection of product defects in the production line.

Description of drawings

1 is a schematic flowchart of a specific embodiment of a product defect detection method in an embodiment of the present invention;

2 is a schematic flowchart of another specific embodiment of a product defect detection method in an embodiment of the present invention;

3 is a schematic flowchart of another specific embodiment of a product defect detection method in an embodiment of the present invention;

4 is a schematic flowchart of another specific embodiment of a product defect detection method in an embodiment of the present invention;

FIG. 5 is a block diagram of a module of a specific embodiment of a product defect detection device in an embodiment of the present invention.

Detailed ways

The concept of the present invention and the technical effects produced will be clearly and completely described below with reference to the embodiments, so as to fully understand the purpose, characteristics and effects of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative efforts are all within the scope of The scope of protection of the present invention.

In the description of the present invention, if a feature is referred to as "arranged", "fixed", "connected", "installed" on another feature, it can be directly set, fixed, connected to the other feature, or it can be Indirectly set, fastened, connected, mounted on another feature.

In the description of the embodiments of the present invention, if it refers to "multiple", it means more than two. If it refers to "greater than", "less than", and "exceeds", it should be understood as not including this number. "Above", "below" and "within" should be understood to include the number. If it refers to "first" and "second", it should be understood to be used to distinguish technical features, but not to indicate or imply relative importance, or to imply indicate the number of indicated technical features or to imply indicate the indicated The sequence of technical features.

Example 1

Referring to FIG. 1 , a schematic flowchart of a specific embodiment of a product defect detection method in an embodiment of the present invention is shown. As shown in FIG. 1 , the specific steps of a product defect detection method according to an embodiment of the present invention include:

S1000. Obtain an image of the product to be detected.

To obtain an image of the product to be inspected, a product scanning imaging device can be used to scan the product to be inspected to obtain image samples of the product to be inspected.

S1100. Use the Otsu algorithm to process the image of the product to be detected to obtain a first image segmentation result.

The Otsu algorithm divides the image into two parts: background and target according to the grayscale characteristics of the image. The greater the inter-class variance between the background and the target, the greater the difference between the two parts that make up the image. When part of the target is mistakenly divided into the background, or part of the background is mistakenly divided into the target, the difference between the two parts will become smaller. Therefore, the segmentation that maximizes the between-class variance means that the probability of misclassification is minimized.

In other specific embodiments of the embodiments of the present invention, referring to FIG. 2 , a schematic flowchart of another specific embodiment of a product defect detection method in an embodiment of the present invention is shown. As shown in Figure 2, a product defect detection method according to an embodiment of the present invention adopts the Otsu algorithm to process the image of the product to be detected to obtain a first image segmentation result, and the specific steps include:

S1110. Perform grayscale processing on the image of the product to be detected to obtain a grayscale image.

The color value of each pixel on a grayscale image, also known as grayscale, refers to the color depth of a point in a black-and-white image, generally ranging from 0 to 255, with 255 for white and 0 for black. The so-called gray value refers to the degree of shade of color. In this embodiment, the grayscale processing of the image of the product to be detected can be performed in any of the following four ways:

(1) Component method, the brightness of the three components in the RGB color image is used as the grayscale value of the three grayscale images, and one grayscale image can be selected according to the application needs.

(2) The maximum value method, the maximum value of the brightness of the three components in the RGB color image is used as the gray value of the gray image.

(3) The average value method, the average value of the brightness of the three components in the RGB color image is used as the gray value of the gray image.

(4) Weighted average method, according to the importance and other indicators, the three components in the RGB color image are weighted and averaged with different weights.

S1120. Perform gamma correction on the grayscale image to obtain a gamma correction image.

Gamma correction can eliminate the influence of lighting imbalance on the imaging effect when the picture is captured. Gamma correction for grayscale image is a nonlinear operation on the grayscale value of the image, so that the grayscale value of the output image has an exponential relationship with the grayscale value of the input image.

S1130. Perform Otsu segmentation on the gamma correction image to obtain an Otsu segmentation result.

Perform Otsu segmentation on the gamma correction map, traverse the thresholds from 0-255 in order, find the optimal threshold, and take the optimal threshold to divide the gamma correction map into foreground color and background color. The greater the difference between the two parts, the more effectively the image can be segmented.

S1140. Perform morphological processing on the Otsu segmentation result to obtain a morphological correction result.

Morphological processing of Otsu segmentation results can eliminate edge burrs in Otsu segmentation results.

In other specific embodiments of the embodiments of the present invention, morphological processing is performed on the Otsu segmentation result to obtain a morphological correction result, including using a low-pass filtering method to perform grayscale smoothing and edge smoothing on the Otsu segmentation result to obtain a morphological correction result. result.

The low-pass filtering method can make the pixel value in the grayscale image flat, keep the points with little change (corresponding to the segmentation area), and filter out the points with sharp changes in the pixel value in the image (corresponding to edge burrs). The low-pass filtering method specifically includes the following steps:

S1141. Determine the size of the kernel function.

The linearly inseparable mode of the low-dimensional space can be linearly separable by nonlinear mapping to the high-dimensional feature space, but if this technique is directly used for classification or regression in the high-dimensional space, there is a form that determines the nonlinear mapping function. And parameters, feature space dimensions and other issues, and the biggest obstacle is the "curse of dimensionality" that exists in high-dimensional feature space operations. The use of kernel function technology can effectively solve such problems.

The kernel function can transform the inner product operation of the high-dimensional space into the kernel function calculation of the low-dimensional input space, thus ingeniously solving the problem of "curse of dimensionality" calculated in the high-dimensional feature space. The size of the kernel function can be determined according to the actual situation.

S1142. Determine the anchor point position.

Select an anchor point position in the original image, and the anchor point position can be determined according to the actual situation.

S1143. Determine the pixel value of the anchor point according to the kernel function and the anchor point position.

The pixel value of the anchor point is determined by calculating the pixel value of the area around the anchor point, and the pixel value of the anchor point is convolved with the weight of the kernel function to obtain the pixel value of the anchor point. Through the kernel function, the pixel value of the smooth area will not change, and the glitch area will be re-smoothed due to the distribution of the number of surrounding pixels.

S1150. Perform median filtering on the morphological correction result to obtain a first image segmentation result.

The median filtering method is similar to the low-pass filtering method, but the selection of the kernel function is different. In low-pass filtering, the pixel value of the anchor point is determined according to the weight of the kernel function and the pixel value of the area around the anchor point. In median filtering, the pixel value of the anchor point is the median value of the area covered by the kernel function. The median filter can eliminate the small defect areas that do not meet the actual situation in Otsu segmentation.

A product defect detection method according to an embodiment of the present invention improves the traditional Otsu algorithm. The traditional Otsu algorithm includes grayscale processing and Otsu segmentation, while the improved Otsu algorithm also includes gamma correction, morphological processing and median value. filter. Compared with the traditional Otsu algorithm, the improved Otsu algorithm is simpler to calculate, so it can detect product images quickly and stably without spending a lot of computing time, and also improves the detection accuracy.

S1200. Use a Gaussian mixture model algorithm to process the image of the product to be detected to obtain a second image segmentation result.

The Gaussian mixture model is to use the Gaussian probability density function to accurately quantify things, and decompose a thing into several models based on the Gaussian probability density function. The principle and process of establishing a Gaussian model for the image background: the image grayscale histogram reflects the frequency of a certain grayscale value in the image, and can also be regarded as an estimation of the grayscale probability density of the image. If the image contains a large difference between the target area and the background area, and there is a certain difference in grayscale between the background area and the target area, then the grayscale histogram of the image presents a double peak-valley shape, and one of the peaks corresponds to the target area. , the other peak corresponds to the central grayscale of the background. For complex images, especially industrial product images, they are generally multimodal. The image segmentation problem can be solved by considering the multimodal nature of the histogram as a superposition of multiple Gaussian distributions.

In other specific embodiments of the embodiments of the present invention, referring to FIG. 3 , a schematic flowchart of another specific embodiment of a product defect detection method in an embodiment of the present invention is shown. As shown in FIG. 3 , in a product defect detection method according to an embodiment of the present invention, a Gaussian mixture model algorithm is used to process an image of a product to be detected to obtain a second image segmentation result. The specific steps include:

S1210. Perform hue saturation value (HSV) space color extraction on the image of the product to be detected to obtain an HSV color map.

S1220. Perform expected maximum (EM) algorithm segmentation on the HSV color image to obtain a second image segmentation result.

In other specific embodiments of the embodiments of the present invention, the EM algorithm is performed on the HSV color image to obtain a second image segmentation result, and the specific steps include:

S1221. Build a Gaussian mixture model.

The mathematical expression form of the Gaussian mixture model of the embodiment of the present invention is shown in formula (1):

的条件。N(x/μ_k,Σ_k)表示高斯分布，其中μ_k和Σ_k分别表示高斯分布的均值和方差。Among them, x represents the pixel point, and p(x) represents the probability of the pixel point appearing in the picture. k represents the type of pixel (defective or non-defective), and p(k) represents the probability of different pixel types in the picture. p(x/k) represents the probability of a pixel appearing in the picture given the known pixel type. π _k is the same as p(k) and represents the probability of different pixel types in the image. Because the probability distribution of the pixel type belongs to a discrete model, it can be represented by a number less than 1. It should be noted that it must meet the

conditions of. N(x/μ _k , Σ _k ) represents a Gaussian distribution, where μ _k and Σ _k represent the mean and variance of the Gaussian distribution, respectively.

The essence of the Gaussian mixture model is to fuse several single Gaussian models to make the model more complex, resulting in more complex samples. In theory, if there are enough single Gaussian models fused by a Gaussian mixture model, and the weights between them are set reasonably enough, the mixture model can fit samples of any distribution.

S1222. Use the Gaussian mixture model to process the HSV color map to obtain the model number of the Gaussian mixture model.

The model number of the Gaussian mixture model of the embodiment of the present invention is determined according to formula (2):

以及θ表示高斯分布的参数，对应式(1)中的μ_k,Σ_k，

表示输入图片的编码长度，π_k表示图片中不同像素点类型的概率，P表示高斯分布的参数数量，K表示模型数量，N表示每张图片的像素点数量，

表示对参数数量的惩罚项。Among them, Y represents the input image,

And θ represents the parameters of the Gaussian distribution, corresponding to μ _k , Σ _k in formula (1),

Indicates the encoding length of the input image, π _k represents the probability of different pixel types in the image, P represents the number of parameters of the Gaussian distribution, K represents the number of models, N represents the number of pixels in each image,

Represents a penalty term for the number of parameters.

The HSV color map is input into the Gaussian mixture model, and the optimal number of models can be calculated by formula (2).

S1223. Obtain an optimization model of the Gaussian mixture model according to the number of models of the Gaussian mixture model.

The optimal number of models is determined, so that the Gaussian mixture model can be further optimized to obtain the optimized model.

S1224. Use the EM algorithm to estimate the parameters of the optimization model to obtain a second image segmentation result.

The EM algorithm is an iterative algorithm for parameter estimation, and the EM algorithm in the embodiment of the present invention specifically includes the following steps:

(1) E step

The first step is done by estimating the Q equation. The mathematical expression of the Q equation is shown in equation (3):

Q(μ,Σ,π,μ ² ,Σ ² ,π ² )＝E _γ [In p(y,γ/μ,Σ,π)/Y,μ ² ,Σ ² ,π ² ] (3)

Among them, Y represents the input image, μ and Σ represent the mean and variance of the Gaussian distribution, respectively, π represents the probability of different pixel types in the image,

In the iterative algorithm, the first update of the Q equation needs to automatically set the initial value of the iteration. The iterative values of the Q equation after the first time are calculated from the results obtained in the M steps.

(2) M step

The M step is the parameter estimation step, and the mathematical expression of the compact form of the three parameter estimation formulas is shown in equation (4):

μ ⁱ⁺¹ ,Σ ⁱ⁺¹ ,π ⁱ⁺¹ =arg max Q(μ,Σ,π,μ ⁱ ,Σ ⁱ ,π ⁱ ) (4)

Among them, μ and Σ represent the mean and variance of the Gaussian distribution, respectively, π represents the probability of different pixel types in the picture, and the superscript i represents the result of the ith iteration.

The specific expressions of the three parameter estimation formulas in formula (4) are related to the specific forms of modeling, and the mathematical expressions of the three parameter estimation formulas are shown in formulas (5) to (7):

Among them, μ _k and Σ _k represent the mean and variance of the Gaussian distribution, respectively, π _k represents the probability of different pixel types in the picture, and the superscript i represents the result of the ith iteration. T represents the number of pixels, N(.) represents Gaussian distribution, y represents pixels, K represents the number of models,

The E step and the M step are the main steps of the EM algorithm. The EM algorithm iterates between steps (1) and (2), and the judgment formula for iterative stop is shown in formula (8):

Among them, μ _k and Σ _k represent the mean and variance of the Gaussian distribution, respectively, N(.) represents the Gaussian distribution, x represents the pixel point, π _k represents the probability of different pixel types in the image, and K represents the number of models. Stop the iterative process when p(x/π, μ, Σ) reaches a certain value.

In a product defect detection method according to an embodiment of the present invention, a Gaussian mixture model is used to simulate the pixel value of an input image of a product to be detected, and different single Gaussian models are used to determine the type of each pixel, so as to detect the product image. , to obtain the second image segmentation result.

S1300. Use a post-processing algorithm to process the first image segmentation result and the second image segmentation result to obtain a detection result.

In other specific embodiments of the embodiments of the present invention, referring to FIG. 4 , a schematic flowchart of another specific embodiment of a product defect detection method in the embodiments of the present invention is shown. As shown in FIG. 4 , in a product defect detection method according to an embodiment of the present invention, a post-processing algorithm is used to process the first image segmentation result and the second image segmentation result to obtain the detection result. The specific steps include:

S1310. Process the first image segmentation result and the second image segmentation result to obtain a third image segmentation result.

The first image segmentation result and the second image segmentation result are processed. First, the first image segmentation result is converted into a first matrix, and the second image segmentation result is converted into a second matrix. Then, the corresponding elements of the first matrix and the second matrix are convolved to obtain a third matrix. Finally, according to the previous matrix transformation rule, the third matrix is inversely transformed, and the third image segmentation result can be obtained. The third image segmentation result is essentially the intersection of the first image segmentation result and the second image segmentation result.

In the matrix conversion rule of the embodiment of the present invention, the pixel value of the defect segmentation area is 1, and the pixel value of the normal area is 0. By convolving the corresponding elements of the first matrix and the second matrix, the defect area segmented by both the first image segmentation result and the second image segmentation result will be set to 1, and the defect area segmented by either of the two will be set to 1. Non-defective areas will be set to 0.

S1320. Use the morphological reconstruction method to detect and repair the vulnerability edge of the third image segmentation result, and obtain the detection result.

The reason for the vulnerability is that the illumination at the time of capturing the image prevents the EM algorithm from classifying correctly. The morphological reconstruction method is used to detect and repair the vulnerability edge of the third image segmentation result in order to eliminate the factors that cause the algorithm to misjudge due to the influence of illumination.

The morphological reconstruction method in the embodiment of the present invention specifically includes the following two steps:

(1) Vulnerability edge detection steps

After performing matrix transformation on the image segmentation result, for a pixel, if there are two different pixel values in the surrounding 4 pixels, it means that this pixel belongs to the edge, and the position of this pixel is saved. The origin of the coordinates (1,1) is defined as the pixel point in the upper left corner of the image, and all the pixel points with the pixel value of 1 in the image are traversed by the edge detection algorithm, and all the existing edge points in the entire image are obtained.

(2) Vulnerability patching steps

Using the characteristics of a closed envelope to connect the edge areas of the vulnerability, first, randomly select a point on the image, record the abscissa of the point, and find the point with the same abscissa in the set of edge points. Then, select the horizontal direction, all the pixels in the middle of these two edge points belong to the vulnerability, this step can be regarded as the search of the vulnerability area. Set the pixel value of the middle vulnerability pixel from 0 to 1, so as to complete the patching of part of the vulnerability. Next, re-select an edge point and repeat the patching steps until the vulnerability patching is completed.

A product defect detection method according to an embodiment of the present invention can overcome the problem that illumination makes it difficult for the EM algorithm to classify correctly through vulnerability edge detection and vulnerability repair, thereby improving the detection accuracy of the EM algorithm.

Example 2

Referring to FIG. 5 , a block diagram of a module of a specific embodiment of a product defect detection apparatus in an embodiment of the present invention is shown. As shown in FIG. 5 , a product defect detection device according to an embodiment of the present invention includes an image acquisition module, which is used to acquire an image of the product to be detected; a first image segmentation module is used to process the image of the product to be detected by using the Otsu algorithm to obtain The first image segmentation result; the second image segmentation module is used to process the image of the product to be detected by using the Gaussian mixture model algorithm to obtain the second image segmentation result; the post-processing module is used to use the post-processing algorithm to process the first image segmentation result and The second image segmentation result is processed to obtain a detection result.

In a product defect detection device according to an embodiment of the present invention, the first image segmentation module has stable and fast segmentation results, the second image segmentation module has accurate segmentation results, fast calculation speed, and does not require a large number of training samples, and the post-processing module is responsible for the first image segmentation module. The image segmentation result and the second image segmentation result are processed, which can not only detect the product image reliably, but also greatly improve the detection speed.

In other specific embodiments of the embodiments of the present invention, the post-processing algorithm includes vulnerability edge detection and vulnerability repair. Through vulnerability edge detection and vulnerability repair, the problem that illumination makes it difficult for the EM algorithm to classify correctly can be overcome, thereby improving the detection accuracy of the EM algorithm.

A product defect detection apparatus according to an embodiment of the present invention can run in computing devices such as a desktop computer, a notebook computer, a palmtop computer, and a cloud server. A product defect detection device, the operable device may include, but not limited to, a processor and a memory. Those skilled in the art can understand that the example is only an example of a product defect detection device, and does not constitute a limitation to a product defect detection device, which may include more or less components, or a combination of some Components, or different components, such as a product defect detection device, may also include input and output devices, network access devices, buses, and the like.

Example 3

An embodiment of the present invention provides a product defect detection device, based on Embodiment 1, comprising at least one processor, and a memory connected in communication with the at least one processor; wherein, the memory stores instructions executable by the at least one processor , the instructions are executed by at least one processor to enable the at least one processor to execute a product defect detection method according to some embodiments of the present invention.

The product defect detection device according to the embodiment of the present invention can greatly improve the detection speed, is suitable for the detection of surface defects of different types of products, and can also reduce the influence of external factors such as different light sources, placement angles of detection parts, and shadows on detection. The impact of the results, to achieve accurate detection of product defects in the production line.

In a product defect detection device according to an embodiment of the present invention, the processor may be a central processing unit (Central Processing Unit, CPU), or 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. Parts of an operational device for a defect detection method.

The memory can be used to store computer programs and/or modules, and the processor can implement a product defect detection method by running or executing the computer programs and/or modules stored in the memory, and calling the data stored in the memory. various functions of the device. The memory can mainly include a stored program area and a stored data area, wherein the stored program area can store an operating system, an application program (such as a sound playback function, an image playback function, etc.) required for at least one function; data (such as audio data, phone book, etc.) In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, internal memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

Example 4

An embodiment of the present invention provides a computer-readable storage medium. Based on Embodiment 1, the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute a product of some embodiments of the present invention. Defect detection method.

A computer-readable storage medium according to the embodiment of the present invention can greatly improve the detection speed, is suitable for surface defect detection of different types of products, and can also reduce the impact of external factors such as different light sources, detection parts placement angles, and shadow conditions. The influence of the detection results, to achieve accurate detection of product defects in the production line.

The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned embodiments. Within the scope of knowledge possessed by those of ordinary skill in the technical field, various modifications can be made without departing from the purpose of the present invention. kind of change. Furthermore, the embodiments of the present invention and features in the embodiments may be combined with each other without conflict.

Claims (10)

1. a product defect detection method, is characterized in that, comprises: Obtain the image of the product to be detected; The Otsu algorithm is used to process the image of the product to be detected to obtain a first image segmentation result; The Gaussian mixture model algorithm is used to process the image of the product to be detected to obtain a second image segmentation result; A post-processing algorithm is used to process the first image segmentation result and the second image segmentation result to obtain a detection result. 2. A product defect detection method according to claim 1, wherein the Otsu algorithm is used to process the image of the product to be detected to obtain a first image segmentation result, comprising: Perform grayscale processing on the image of the product to be detected to obtain a grayscale image; performing gamma correction on the grayscale image to obtain a gamma correction image; Otsu segmentation is performed on the gamma correction map to obtain an Otsu segmentation result; Morphological processing is performed on the Otsu segmentation result to obtain a morphological correction result; Perform median filtering on the morphological correction result to obtain the first image segmentation result. 3. a kind of product defect detection method according to claim 2 is characterized in that, described carrying out morphological processing to described Otsu segmentation result, obtains morphological correction result, comprising: The low-pass filtering method is used to perform grayscale smoothing and edge smoothing on the Otsu segmentation result to obtain the morphological correction result. 4. A product defect detection method according to any one of claims 1 to 3, wherein the Gaussian mixture model algorithm is used to process the image of the product to be detected to obtain a second image segmentation result, comprising: : Perform HSV space color extraction on the image of the product to be detected to obtain an HSV color map; EM algorithm segmentation is performed on the HSV color image to obtain the second image segmentation result. 5. A product defect detection method according to claim 4, characterized in that, performing EM algorithm segmentation on the HSV color image to obtain the second image segmentation result, comprising: Build a Gaussian mixture model; Using the Gaussian mixture model to process the HSV color map to obtain the model number of the Gaussian mixture model; According to the number of models of the Gaussian mixture model, the optimization model of the Gaussian mixture model is obtained; The EM algorithm is used to estimate the parameters of the optimized model to obtain the second image segmentation result. 6 . The method for detecting product defects according to claim 4 , wherein the post-processing algorithm is used to process the first image segmentation result and the second image segmentation result to obtain a detection result, comprising: 6 . : processing the first image segmentation result and the second image segmentation result to obtain a third image segmentation result; The morphological reconstruction method is used to perform edge detection and repair on the third image segmentation result to obtain the detection result. 7. A product defect detection device, characterized in that, comprising: The image acquisition module is used to acquire the image of the product to be detected; a first image segmentation module, used to process the image of the product to be detected by using the Otsu algorithm to obtain a first image segmentation result; A second image segmentation module, configured to process the image of the product to be detected by using a Gaussian mixture model algorithm to obtain a second image segmentation result; The post-processing module is configured to use a post-processing algorithm to process the first image segmentation result and the second image segmentation result to obtain a detection result. 8 . The device for detecting product defects according to claim 7 , wherein the post-processing algorithm includes vulnerability edge detection and vulnerability repair. 9 . 9. A product defect detection device, characterized in that, comprising: at least one processor, and, a memory communicatively coupled to at least one of the processors; wherein, The memory stores instructions executable by at least one of the processors, the instructions being executed by the at least one of the processors to enable the at least one of the processors to perform the execution of any one of claims 1 to 6 product defect detection method. 10. A computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute the method according to any one of claims 1 to 6. product defect detection method.

Images