CN115861320B - Intelligent detection method for automobile part machining information - Google Patents

Abstract

The invention relates to the technical field of image data processing, in particular to an intelligent detection method for automobile part processing information. The method comprises the steps of obtaining a gray image of the bearing surface after fluorescent magnetic powder injection; obtaining a suspected defect area in the gray level image through a watershed algorithm; obtaining a pixel point corresponding to the maximum gray value in each suspected defect area as a target pixel point, clustering the pixel points in each suspected defect area, and taking the area of a cluster where the target pixel point is located as a target area; acquiring the aggregation degree of fluorescent magnetic powder of each target area according to the area of the suspected defect area corresponding to the target area and the position distribution and gradient distribution of the pixel points in the target area; acquiring a gray level difference between a target area and a gray level image as a target difference; obtaining the similarity between the target area and the background area according to the aggregation degree of the fluorescent magnetic powder and the target difference; and determining a defect area in the gray image according to the similarity. The efficiency of detecting defective areas is improved.

Description

An intelligent detection method for processing information of automobile parts

technical field

The invention relates to the technical field of image data processing, in particular to an intelligent detection method for processing information of automobile parts.

Background technique

Bearing is a very critical part of auto parts, the quality of bearing is related to the driving safety of the car. During the production and processing of bearings, defects will occur due to excessive forging heating temperature or excessive holding time, and in severe cases, the grain boundaries will be oxidized or even melted. The burnt bearing is forged in this defect state, and is subjected to forging, punching and rolling with a heavy hammer, and the defect will tear and form a larger defect.

In the prior art, the defect detection on the bearing surface is usually carried out using the fluorescent magnetic particle method. The fluorescent magnetic powder will adhere to the defect area on the bearing surface to form a fluorescent area, which can be segmented by the commonly used watershed algorithm in the field of image segmentation. However, the fine fluorescent magnetic powder will also adhere to the unevenness of the bearing surface to form a fluorescent area. Therefore, when segmented by the watershed algorithm, the area of the suspected defect area retained after flooding will be larger due to the larger fluorescent area. The suspected defect area includes both some uneven areas without defects and defect areas, which makes the segmented defect areas inaccurate, interferes with the identification of defect areas, and is not conducive to the detection of defect areas.

Contents of the invention

In order to solve the technical problem of inaccurate detection of defect areas caused by the interference of the uneven area of the bearing on the defect area, the purpose of the present invention is to provide an intelligent detection method for processing information of automobile parts. The technical solution adopted is as follows:

An embodiment of the present invention provides an intelligent detection method for processing information of automobile parts, the method includes the following steps:

Obtain a grayscale image of the bearing surface after fluorescent magnetic particle spraying;

The suspected defect area in the grayscale image is obtained by the watershed algorithm; the pixel point corresponding to the maximum gray value in each of the suspected defect areas is obtained as the target pixel point, and the pixel point in each of the suspected defect area is obtained. Carrying out clustering, using the region of the cluster where the target pixel is located as the target region;

Acquiring the degree of fluorescent magnetic powder aggregation of each target area according to the area of the suspected defect area corresponding to the target area and the position distribution and gradient distribution of pixels in the target area;

Obtaining the grayscale difference between the target area and the grayscale image as the target difference; acquiring the similarity between the target area and the background area according to the aggregation degree of the fluorescent magnetic powder and the target difference;

Merging the suspected defect area corresponding to the target area with the background area according to the similarity, to determine the defect area in the grayscale image.

Further, the method for obtaining the aggregation degree of the fluorescent magnetic powder includes:

Acquire the coordinates of the pixel corresponding to the maximum gray value in the target area as the first coordinate, calculate the distance between the coordinate corresponding to each pixel in the target area and the first coordinate, and obtain the target area The average distance between the pixels in ; the average distance is negatively correlated and normalized, and the result is taken as the first result;

Obtaining the gradient mean value of the edge pixels in the target area as the average gradient;

Obtaining the number of pixels in the suspected defective region corresponding to the target region, that is, the area of the suspected defective region, as the first area;

Calculate the product of the first result, the average gradient and the first area as the aggregation degree of the fluorescent magnetic powder in the target area.

Further, the method for obtaining the target difference includes:

calculating the average gray value in the target area as the first value;

calculating the average gray value in the gray image as a second value;

The difference between the first value and the second value is used as the target difference between the target area and the grayscale image.

Further, the method for obtaining the similarity includes:

The product of the degree of fluorescent magnetic particle aggregation and the target difference is negatively correlated and normalized as the similarity between the target area and the background area.

Further, the method for determining the defect area in the grayscale image according to the similarity includes:

Set the similarity threshold. When the similarity is greater than the similarity threshold, there is no defect in the suspected defect area where the target area is located, and the suspected defect area is merged with the background area; the unmerged area is the defect area, and the defect area is verified.

Further, the method for clustering the pixels in each of the suspected defect regions includes:

Use the DBSCAN density clustering algorithm to cluster the pixels in the suspected defect area according to the position and gray value of the pixels.

The present invention has following beneficial effects:

The suspected defect area in the grayscale image of the bearing surface is obtained by the watershed algorithm, and then only the suspected defect area is analyzed to improve the efficiency of defect area detection; the pixel corresponding to the maximum gray value in each suspected defect area is obtained as The target pixel is convenient for subsequent analysis of the specific characteristics of the suspected defect area; therefore, the pixels in each suspected defect area are clustered, and the cluster area where the target pixel is located is used as the target area. The pixel point is the pixel point with a larger gray value peak in the suspected defect area, which can reflect the attachment of the fluorescent magnetic powder in the corresponding suspected defect area. Therefore, according to the area of the suspected defect area corresponding to the target area and the pixel point in the target area The position distribution and gradient distribution of each target area are obtained to obtain the degree of fluorescent magnetic particle aggregation in each target area, and the defect area is initially judged; in order to further determine the defect area, the gray difference between the target area and the gray image is obtained as the target difference, and the target area is indirectly determined. The degree of difference between the suspected defect area and the background area; then the similarity between the target area and the background area is obtained according to the degree of fluorescent magnetic particle aggregation and the target difference, and the suspected defect area corresponding to the target area is merged with the background area according to the similarity, and the bearing surface is removed. The interference to defect area detection caused by unevenness eliminates the over-segmented area caused by the watershed algorithm, determines the defect area in the grayscale image, and improves the accuracy of defect area detection.

Description of drawings

In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or in the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description The drawings are only some embodiments of the present invention, and those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a schematic flow chart of an intelligent detection method for processing information of automobile parts provided by an embodiment of the present invention;

Fig. 2 is a distribution diagram of the gray value in the suspected defect area provided by an embodiment of the present invention.

Detailed ways

In order to further explain the technical means and effects adopted by the present invention to achieve the intended purpose of the invention, the specific implementation manner, The structure, characteristics and effects thereof are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, the particular features, structures or characteristics of one or more embodiments may be combined in any suitable manner.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field of the invention.

The specific scheme of an intelligent detection method for processing information of automobile parts provided by the present invention will be described in detail below in conjunction with the accompanying drawings.

Please refer to Fig. 1, which shows a schematic flowchart of a method for intelligent detection of automobile parts processing information provided by an embodiment of the present invention, the method includes the following steps:

Step S1: Obtain a grayscale image of the bearing surface after fluorescent magnetic particle spraying.

Specifically, the purpose of the embodiment of the present invention is to detect the defects on the bearing surface by the fluorescent magnetic powder method. First, a certain amount of fluorescent magnetic powder is sprayed on the bearing surface by using a SAIF flaw detector, and then the bearing surface after spraying the fluorescent magnetic powder is collected by a high-definition camera. The obtained image is denoised. In the embodiment of the present invention, a mean filter algorithm is used to denoise the obtained image, and then the denoised image is grayscaled to obtain a grayscale image.

Wherein, the mean value filtering algorithm and the grayscale processing are both existing technologies, and will not be described in detail here.

Step S2: Obtain the suspected defect area in the grayscale image through the watershed algorithm; obtain the pixel corresponding to the maximum gray value in each of the suspected defect areas as the target pixel, and for each of the suspected defect areas The pixels are clustered, and the area of the cluster where the target pixel is located is taken as the target area.

Specifically, the bearing surface is marked by the fluorescent magnetic particle method. Different flatness of the bearing surface will form different bright and dark areas. A whole fluorescent area is formed; the defects on the surface of the bearing are oxidized and cracked at the grain boundary of the surface layer under microscopic observation. If the segregation of the internal components of the metal is serious, the grain boundary will also begin to melt, and in severe cases, a sharp hole will be formed. , the fluorescent magnetic powder will be adsorbed in the defect area, and also form a whole fluorescent area. Analyze the fluorescent area on the bearing surface. When the fluorescent area is bigger and brighter, it means that the fluorescent area is more likely to be a larger uneven area or defect area on the bearing surface. In the grayscale image, the grayscale value corresponding to the fluorescent area is larger, and the grayscale value corresponding to the other areas except the fluorescent area, that is, the background area, is smaller. The initial segmented area in the grayscale image is the suspected defect area obtained by the watershed algorithm. The watershed algorithm is to find the minimum value in the grayscale image, and then start to expand. If there are too many small areas in the grayscale image, many small catchment basins will be generated, which will cause the grayscale image to be over-segmented. Therefore, when segmented by the watershed algorithm, due to the large fluorescent area, the area of the suspected defect area retained after flooding will be large, so that the suspected defect area includes both some uneven areas without defects and defect areas, making the segmentation The defect area that comes out is not accurate. Therefore, it is necessary to process the obtained suspected defect areas, remove the suspected defect areas without defects, and then identify the remaining suspected defect areas to determine the defect areas. The embodiment of the present invention optimizes the flooding position of the watershed algorithm according to the characteristics of the defect area, obtains the best flooding position, and removes the interference caused by the fluorescent magnetic particle marking in the uneven area without defects, so that the retained suspected defects The area is a defect area.

Among them, the watershed algorithm is a well-known technology, and will not be repeated here.

The suspected defect area obtained by the watershed algorithm is directly judged according to the threshold value, and it is impossible to distinguish which suspected defect areas are uneven areas and defect areas. Therefore, it is necessary to analyze the specific characteristics of each suspected defect area to identify the suspected defects. The area is merged, and the suspected defect area belonging to the normal uneven area is merged with the background area, that is, when segmented by the watershed algorithm, it is not only a "flooding" process, but also the suspected defect area needs to be judged. Let the suspected defect area belonging to the normal uneven area also become the object of "flooding".

To analyze the specific characteristics of the suspected defect area, it is necessary to obtain the target area in each suspected defect area first. The specific operation of obtaining the target area is as follows:

Obtain the pixel point corresponding to the maximum gray value in each suspected defect area as the target pixel point, if there are at least two pixels corresponding to the maximum gray value in a suspected defect area, then arbitrarily select a pixel point corresponding to the maximum gray value pixel as the target pixel. Use the DBSCAN density clustering algorithm to cluster the pixels in the suspected defect area according to the position and gray value of the pixel. Among them, the DBSCAN density clustering algorithm is to cluster according to the position of the pixel and select the pixel by the gray value Which pixels are clustering objects. Obtain the area of the cluster where the target pixel is located as the target area. In the embodiment of the present invention, the neighborhood radius in the DBSCAN density clustering algorithm is set to 3, and the minimum number of neighborhood pixels is 3. Here, the threshold is an empirical threshold, which can be set by the implementer according to different implementation environments. Among them, the DBSCAN density clustering algorithm is a well-known technology, and will not be repeated here.

Among them, the reasons for using the area of the cluster where the target pixel is located as the target area are as follows:

The fluorescent magnetic powder attached to the defect area is more and uniform, so the overall gray value of the defect area is large and the gray level fluctuation is large; the gray value distribution of pixels in the defect area is shown in area B in Figure 2, and the area in B is The height of each peak is the gray value of the corresponding pixel in the defect area; there is a background area between the suspected defect areas formed by the attachment of fluorescent magnetic particles on the uneven area of the bearing surface, so the peaks in the corresponding gray scale distribution are the same as The distance between the peaks is relatively large, and the peaks of the gray value in the suspected defect area corresponding to the uneven area are not uniform, and the gray level fluctuations of the pixels vary from large to small, as shown in area A in Figure 2, and each peak in area A is the gray value of the corresponding pixel in the suspected defect area. According to Figure 2, it can be found that the distribution characteristics of the peaks between the suspected defect area corresponding to the uneven area and the defect area are very different. And the peak is large. After passing the watershed algorithm, some peaks in the A region are smaller than the threshold and will not be segmented, while the peaks in the B region are greater than the segmentation threshold and will be segmented. The peaks in the B region have a higher degree of fluorescent magnetic particle aggregation big. Therefore, the suspected defect region with a greater degree of fluorescent magnetic particle aggregation is more likely to be a defect region. The cluster where the target pixel is located can fully reflect the distribution of high gray value peaks in the corresponding suspected defect area, so the cluster area where the target pixel is located is selected as the target area to analyze the target area. Because the distribution of gray value peaks in area A and area B is different, the number of pixels in the target area corresponding to each suspected defect area after clustering is different and the distribution of pixel points is also different. According to the number and distribution of pixels in the target area Obtain the degree of fluorescent magnetic particle accumulation for each target area.

Step S3: Obtain the degree of fluorescent magnetic powder aggregation in each target area according to the area of the suspected defect area corresponding to the target area and the position distribution and gradient distribution of pixels in the target area.

Specifically, the fluorescent area generated by the uneven area of the bearing surface will interfere with the recognition of the defect area, and it is impossible to accurately distinguish the fluorescent area without defects from the defect area only by relying on the gray scale distribution. However, as long as there are fluorescently marked areas, there may be defects. Therefore, the suspected defect area is obtained by segmenting the fluorescent area in the grayscale image, and then the suspected defect area is analyzed to obtain the accurate defect area.

The suspected defect area is analyzed, the suspected defect area without defect is merged with the background area, and the final segmented area is regarded as the defect area. It can be known from step S2 that a preliminary analysis of the suspected defect area can be carried out according to the aggregation degree of the fluorescent magnetic powder in the target area.

Preferably, the method for obtaining the aggregation degree of the fluorescent magnetic powder is: obtaining the coordinates of the pixel point corresponding to the maximum gray value in the target area as the first coordinate, and calculating the distance between the coordinate corresponding to each pixel point in the target area and the first coordinate Get the average distance between the pixels in the target area; negatively map and normalize the average distance, and the result is taken as the first result; get the mean gradient of the edge pixels in the target area as the average gradient; Obtain the number of pixels in the suspected defect region corresponding to the target region, that is, the area of the suspected defect region as the first area; calculate the first result, the product of the average gradient and the first area as the degree of fluorescent magnetic particle aggregation in the target region.

It should be noted that the fluorescent area caused by the unevenness of the bearing surface will only retain the area with obvious fluorescence after being segmented by the watershed algorithm, and the area with inconspicuous fluorescence will be segmented into the background area; the defect area will be segmented by the watershed algorithm The retained area is larger, therefore, the number of pixels with larger gray values in the defect area is larger, and the pixels with larger gray values are the peak points of each peak in area B in Figure 2, corresponding to The pixels in the target area correspond to the pixels in the suspected defect area that are close to the peak value of the maximum peak point, and correspond to a larger defect area on the bearing surface. Calculate the average distance of the pixel points in the target area to represent the degree of fluorescent magnetic powder aggregation in the target area. The greater the degree of fluorescent magnetic powder aggregation, the greater the possibility that the suspected defect area corresponding to the target area is a defect area. Use the gradient change of the edge pixels to correct the clustering results. The result of density clustering is not only to cluster the pixels close to the peak, but also to cluster the regions corresponding to the clustered pixels into one category. , the peaks in the defect area are more concentrated and the gray value is larger, and the larger the gradient of the edge pixel point is, the more obvious the edge feature is due to the aggregation of fluorescent magnetic powder in this area, and the greater the degree of fluorescent magnetic powder aggregation. After clustering, different peaks will be clustered into one area, that is, the target area. Therefore, the larger the area of the target area, that is, the larger the number of pixels in the target area, it means that the fluorescent magnetic powder at the corresponding position of the target area is more concentrated, and the fluorescent magnetic powder The greater the aggregation degree, the greater the possibility that the corresponding suspected defect region is a defect region.

After being segmented by the watershed algorithm, each suspected defect area is equivalent to an isolated area. The number of pixels in the suspected defect area corresponding to the defect area must be large, so the suspected defect area with more pixels is more likely to be a defect area. Large, and the suspected defect area without defects segmented due to unevenness is equivalent to some scattered points, the number of pixels in the corresponding suspected defect area is small, and the possibility of the suspected defect area being a defect area is smaller. Therefore, the number of pixels in the suspected defect area, that is, the area of the suspected defect area, is taken as a component for calculating the degree of aggregation of the fluorescent magnetic powder.

As an example, taking the target area j as an example, the coordinates of the pixel point corresponding to the maximum gray value in the target area j are obtained as the first coordinates. If there are at least two pixel points corresponding to the maximum gray value in the target area j, then The coordinates of a pixel point are arbitrarily selected as the first coordinates. Obtain the distance between each pixel in the target area j and the pixel corresponding to the first coordinate according to the coordinates of each pixel in the target area j. pass The operator obtains the gradient of each pixel in the target area j, and obtains the gradient of the edge pixels at the boundary of the area, where, The operator is a well-known technology, and details are not repeated here. According to the distance between each pixel point in the target area j and the pixel point corresponding to the first coordinate and the gradient of each pixel point, the degree of fluorescent magnetic powder aggregation in the target area j is obtained, and the degree of fluorescent magnetic powder aggregation in the target area j is obtained. The formula is:

in, is the aggregation degree of fluorescent magnetic particles in the target area j; is the number of pixels in the target area j; is the coordinate of the pixel point corresponding to the maximum gray value in the target area j, that is, the first coordinate; is the target area j in the The coordinates of a pixel point; is the target area j in the The gradient of edge pixels; is the area of the suspected defect area where the target area j is located, i.e. the first area; is a natural constant.

It should be noted that the average distance The smaller the value, the more concentrated the position distribution of the pixels in the target area j is, The larger; the average gradient The larger the value, the greater the grayscale change of the edge pixels in the target area j, so the edge features produced by the aggregation of fluorescent magnetic powder are more obvious. bigger; The larger the value, the larger the area of the suspected defect region where the target region j is located, that is, the larger the number of pixels, so that the more pixels satisfy the clustering conditions, the better the clustering effect. larger; therefore, The larger the , the higher the concentration of fluorescent magnetic powder in the target area j is, and the more likely the suspected defect area where the target area j is located is a defect area.

According to the method of acquiring the degree of accumulation of fluorescent magnetic powder in the target area j, the degree of accumulation of fluorescent magnetic powder in each target area is acquired. The greater the degree of fluorescent magnetic particle aggregation in the target area, the more uneven the target area is, and the more uneven the area is marked by the fluorescent magnetic particle method, the more obvious the marking effect is; the smaller the degree of fluorescent magnetic particle aggregation in the target area, the more uneven the target area is. The more normal areas are retained due to the unevenness of the bearing surface. It is inaccurate to judge the correlation of the target area by the position of the pixel point. It is also necessary to correct the clustering result according to the gradient change between each peak, and then judge the correspondence according to the degree of fluorescent magnetic particle aggregation in each target area. Whether the suspected defect area can be merged with the background area.

Step S4: Acquiring the grayscale difference between the target area and the grayscale image as the target difference; acquiring the similarity between the target area and the background area according to the aggregation degree of the fluorescent magnetic powder and the target difference.

Specifically, in order to determine the optimal flooding position of the watershed algorithm, the suspected defect area without defects is merged with the background area, and then the accurate defect area is obtained, and the target area and the background area are obtained according to the degree of fluorescent magnetic particle aggregation in the target area similarity.

Preferably, the method for obtaining the similarity between the target area and the background area is: calculating the average gray value in the target area as the first value; calculating the average gray value in the gray image as the second value; combining the first value with The difference of the second value serves as the target difference between the target area and the grayscale image. The product of the degree of fluorescent magnetic particle aggregation and the target difference is negatively correlated and normalized as the similarity between the target area and the background area.

It should be noted that by calculating the similarity between the suspected defect area and the background area, the suspected defect area with a large similarity is merged with the background area, so as to exclude some suspected defects that do not have defects but affect the judgment of the defect area area. When the segmentation is performed by the watershed algorithm, there will not be too much interference of suspected defect areas without defects, so that the segmentation of the defect areas in the grayscale image is more accurate.

As an example, taking the target area j in step S3 as an example, according to the gray value of each pixel in the target area j, the average gray value of the target area j is the first value; For the gray value of each pixel, obtain the average gray value of the gray image, which is the second value, because the pixels in the target area j are all pixels marked with fluorescent magnetic powder, so the gray value is larger, that is, the first value Larger, and the pixels in the grayscale image also include the pixels in the background area, so the second value must be smaller than the first value, and the difference between the first value and the second value, that is, the target difference must be a positive number. Get target difference for target region j The formula is:

in, is the target difference between the target area j and the grayscale image; is the average gray value of the target area j, i.e. the first value; is the number of pixels in the grayscale image; for the grayscale image gray value of a pixel.

It should be noted, The larger the value, the greater the difference between the average gray value of the target area j and the average gray value of the gray image; the gray value of the defect area is larger than that of the normal area when the fluorescent magnetic particle method is used to mark. And the larger the gray value, the more obvious the defect; therefore, The larger the value, the greater the difference between the target area j and the background area.

According to the aggregation degree of fluorescent magnetic powder difference from target Get the similarity between the target area j and the background area The formula is:

in, is the similarity between the target area j and the background area; is the aggregation degree of fluorescent magnetic particles in the target area j; is the target difference of the target region j; is a natural constant.

It should be noted, The larger the value, the more likely the suspected defect area where the target area j is located is a defect area, smaller; The larger the value, the greater the difference between the gray value in the target area j and the gray value in the background area, the less similar the target area j is to the background area, and the more likely the target area j is a defect area. smaller; therefore, The smaller the , the less similar the target area j is to the background area, and the suspected defect area where the target area j is located is more likely to be a defect area.

According to the method of obtaining the similarity between the target area j and the background area, the similarity between each target area and the background area is obtained. Set the appropriate segmentation threshold in the watershed algorithm according to the similarity, so that the final segmented area is the defect area.

Step S5: Merge the suspected defective region corresponding to the target region with the background region according to the similarity, and determine the defective region in the grayscale image.

Set the similarity threshold. When the similarity is greater than the similarity threshold, there is no defect in the suspected defect area where the target area is located, and the suspected defect area is merged with the background area; the unmerged area is the defect area, and the defect area is verified.

In the embodiment of the present invention, the similarity threshold is set to 0.32, which can be adjusted by the implementer according to different actual situations. When the similarity is greater than the similarity threshold, it means that the similarity between the target area and the background area is large, and there is no defect in the suspected defect area where the target area is located, then the suspected defect area is merged with the background area, and the grayscale image is processed by the watershed algorithm. During segmentation, the suspected defect area will not be segmented.

After merging the suspected defect areas, the suspected defect areas without defects are eliminated, and the remaining unmerged suspected defect areas are defect areas. Marking the defect areas facilitates manual verification of the defect areas and improves the accuracy of defect detection. efficiency.

So far, the embodiment of the present invention ends.

In summary, the embodiment of the present invention obtains the grayscale image of the bearing surface after fluorescent magnetic powder spraying; obtains the suspected defect area in the grayscale image through the watershed algorithm; obtains the pixel corresponding to the maximum gray value in each suspected defect area As the target pixel, cluster the pixels in each suspected defect area, and use the area of the cluster where the target pixel is located as the target area; according to the area of the suspected defect area corresponding to the target area and the number of pixels in the target area The position distribution and gradient distribution obtain the degree of fluorescent magnetic particle aggregation of each target area; obtain the gray level difference between the target area and the gray image as the target difference; obtain the similarity between the target area and the background area according to the degree of fluorescent magnetic particle aggregation and the target difference ; Determine the defect area in the grayscale image based on the similarity. Improved efficiency in detecting defective areas.

It should be noted that: the order of the above embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. Multitasking and parallel processing are also possible or may be advantageous in certain embodiments.

Each embodiment in this specification is described in a progressive manner, the same and similar parts of each embodiment can be referred to each other, and each embodiment focuses on the differences from other embodiments.

Claims (4)

1. A method for intelligent detection of automobile parts processing information, characterized in that the method comprises the following steps: Obtain a grayscale image of the bearing surface after fluorescent magnetic particle spraying; The suspected defect area in the grayscale image is obtained by the watershed algorithm; the pixel point corresponding to the maximum gray value in each of the suspected defect areas is obtained as the target pixel point, and the pixel point in each of the suspected defect area is obtained. Carrying out clustering, using the region of the cluster where the target pixel is located as the target region; Acquiring the degree of fluorescent magnetic powder aggregation of each target area according to the area of the suspected defect area corresponding to the target area and the position distribution and gradient distribution of pixels in the target area; Obtaining the grayscale difference between the target area and the grayscale image as the target difference; acquiring the similarity between the target area and the background area according to the aggregation degree of the fluorescent magnetic powder and the target difference; merging the suspected defect region corresponding to the target region with the background region according to the similarity, and determining the defect region in the grayscale image; The method for obtaining the aggregation degree of the fluorescent magnetic powder includes: Acquire the coordinates of the pixel corresponding to the maximum gray value in the target area as the first coordinate, calculate the distance between the coordinate corresponding to each pixel in the target area and the first coordinate, and obtain the target area The average distance between the pixels in ; the average distance is negatively correlated and normalized, and the result is taken as the first result; Obtaining the gradient mean value of the edge pixels in the target area as the average gradient; Obtaining the number of pixels in the suspected defective region corresponding to the target region, that is, the area of the suspected defective region, as the first area; calculating the product of the first result, the average gradient and the first area as the aggregation degree of fluorescent magnetic particles in the target area; The method for obtaining the target difference includes: calculating the average gray value in the target area as the first value; calculating the average gray value in the gray image as a second value; The difference between the first value and the second value is used as the target difference between the target area and the grayscale image. 2. A kind of automobile parts processing information intelligent detection method as claimed in claim 1, is characterized in that, the acquisition method of described similarity comprises: The product of the degree of fluorescent magnetic particle aggregation and the target difference is negatively correlated and normalized as the similarity between the target area and the background area. 3. A kind of automobile parts processing information intelligent detection method as claimed in claim 1, is characterized in that, the method for determining the defect area in the grayscale image according to the similarity includes: Set the similarity threshold. When the similarity is greater than the similarity threshold, there is no defect in the suspected defect area where the target area is located, and the suspected defect area is merged with the background area; the unmerged area is the defect area, and the defect area is verified. 4. A kind of automobile parts processing information intelligent detection method as claimed in claim 1, is characterized in that, the method for clustering the pixels in each of the suspected defect regions comprises: Use the DBSCAN density clustering algorithm to cluster the pixels in the suspected defect area according to the position and gray value of the pixels.

Images