An alloy surface detection system based on image recognition

Abstract

The application discloses an alloy surface detection system based on image recognition, comprising: an alloy surface image acquisition unit, which is used for reading alloy surface image information through an image acquisition machine; an alloy surface image recognition and extraction unit, which is used for pre-processing alloy surface image information, obtaining an alloy surface grayscale image and performing median filter noise reduction processing; obtaining contour information through an edge detection algorithm; an analysis unit, which is used for obtaining structural features of the contour information, analyzing alloy surface quality by detecting the structural features of the contour information, and more conveniently grasping problems of abnormal image information of the image surface by detecting the structural features of the contour, so as to reduce a loss rate in the production process of the alloy and facilitate systematic grasping of alloy surface detection.

Description

Technical Field

[0001] This invention relates to the field of alloy surface image recognition technology, and more specifically to an alloy surface inspection system based on image recognition. Background Technology

[0002] Image recognition refers to the process of using computers to process, analyze, and understand images. Image recognition technology is widely used in industrial production and product processing, and it has been widely applied in alloy surface inspection.

[0003] Traditional metal surface inspection is usually carried out by manual inspection or indirect inspection controlled by manual means. At present, image recognition is used to improve the accuracy, efficiency and intelligence of alloy surface inspection. However, there are still some problems with the application of image recognition technology in alloy surface inspection. Some factors affecting the quality of alloy surface are not classified and defined. It is difficult to distinguish the specific surface structure information in the images of abnormal alloy surfaces. Furthermore, the inspection process lacks further understanding of the specific structural features, which increases the difficulty of screening. Summary of the Invention

[0004] The purpose of this invention is to provide an alloy surface inspection system based on image recognition, solving the following technical problems:

[0005] (1) How to inspect the surface quality of an alloy based on its surface structure information;

[0006] (2) How to determine the specific problems existing in the surface structure of the alloy.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] An image recognition-based alloy surface inspection system includes:

[0009] An alloy surface image acquisition unit is used to read alloy surface image information through an image acquisition machine.

[0010] The alloy surface image recognition and extraction unit is used to preprocess alloy surface image information, obtain the alloy surface grayscale image and perform median filtering noise reduction; and obtain contour information through edge detection algorithm.

[0011] The analysis unit is used to acquire the structural features of the contour information and analyze the surface quality of the alloy by detecting the structural features of the contour information.

[0012] In some embodiments, the specific steps for the analysis unit to detect the surface quality of the alloy are as follows:

[0013] S100, Obtain the longest side of the contour of the alloy surface anomaly image. Shortest side Outline perimeter and outline area Through formula Calculate the contour structure value of the abnormal image , , These are the weighting coefficients. , All are greater than 0;

[0014] S200, The abnormal image contour structure value With preset threshold and A comparison was conducted, among which < :

[0015] like ≤ Then the outline of the abnormal image on the alloy surface is determined as follows: crack;

[0016] like < < If so, the abnormal image outline on the alloy surface is determined to be a coarse crack;

[0017] like ≥ If so, the abnormal image contour of the alloy surface is determined to be a pit or a bump.

[0018] In some implementations, the abnormal image contours on the alloy surface are determined to be pits or bumps by setting an illumination strategy. The specific detection steps are as follows:

[0019] S300: By acquiring the grayscale image of the abnormal image contour of the alloy surface, an illumination strategy is executed to obtain the ratio of white grayscale pixels to all pixels in the grayscale image under different illumination intensities, forming a variation curve.

[0020] S400: Compare the shape of the changing curve with the preset standard curve:

[0021] If the shape curve matches the standard curve, the abnormal profile of the alloy surface is a bump;

[0022] If the shape curve does not match the standard curve, the abnormal profile of the alloy surface will be a pit.

[0023] In some embodiments, the alloy surface image acquisition unit analyzes the sample of the acquired coordinate points and compares the coordinate points of the abnormal alloy surface image with a standard threshold: if they are within the standard threshold range, the detection is qualified; otherwise, it is unqualified.

[0024] In some embodiments, the analysis unit is also used to determine whether the alloy cutting edge is accurate. The specific operation steps are as follows:

[0025] SS100: Obtain n points evenly distributed along the boundary line of the cutting edge;

[0026] SS200: Calculate the distances between n points and the center point of the region to obtain the set. Extract the set In the standard set Overlap value ;

[0027] SS300, overlap value Compare with the standard threshold:

[0028] like If the cut edge is within the standard threshold range, it is considered acceptable.

[0029] like If the cut edge is not within the threshold range, it is deemed unqualified.

[0030] In some embodiments, the alloy surface image recognition and extraction unit includes sampling and detecting grayscale features at the target location and setting grayscale values. grayscale values Compared with standard grayscale threshold Compare:

[0031] like If so, the grayscale value setting is reasonable;

[0032] like If the grayscale value is set incorrectly, adjust the difference between the corresponding pixel and other surrounding pixels.

[0033] The beneficial effects of this invention are:

[0034] (1) By setting up an analysis unit, this invention ensures that the structural information of the abnormal image of the alloy surface is determined based on the alloy surface contour obtained in the previous step. In the process of producing alloys and other materials, there are handling, handover and transfer processes. Alloy surface detection is an important quality inspection process in the factory production process. The abnormal structure of the alloy surface usually includes scratches, bumps, pits, etc. This invention determines the detection quality of the detection contour by obtaining the longest side, shortest side, perimeter and area of ​​the detection contour; thereby determining that the abnormal structural features of the alloy surface are the thickness of the cracks and the pits or bumps. By using the structural features of the detection contour, it is easier to grasp the source of the abnormal image information of the image surface, ensuring that the loss rate is reduced during the alloy production process and improving the detection accuracy of the alloy surface.

[0035] (2) This invention further determines the outline of the abnormal image of the alloy surface by setting an illumination strategy. The illumination strategy measures the changes in light intensity at different levels and obtains grayscale images of the abnormal outline of the alloy surface under the corresponding light intensity conditions. The ratio of white grayscale to all pixel values ​​in the measured range in the grayscale images under different light intensities is statistically analyzed. A change curve is formed by recording the ratio under different light intensities. Since the pit reflects less light and presents a larger darkness, the white grayscale ratio is relatively small, while the bump reflects more light and presents a larger brightness, the white grayscale ratio is relatively large. Due to the characteristics of the pit and the bump, the change curve is compared with the shape corresponding to the preset standard curve. The curve formed by the numerical ratio corresponding to the bump is the preset standard curve. The degree of conformity with the standard curve is used to determine whether it is a bump or a pit. By setting the illumination strategy, a standard curve is formed under different light intensity conditions, so as to more comprehensively detect the specific structural type of the abnormality of the alloy surface.

[0036] (3) This invention sets up a detection process for the alloy cutting edge, and also sets up an alloy surface analysis unit for analysis. Since the cut alloy plate or shape has a specified template and shape standard, and the distance from the edge shape to the center has a specified dimension, the invention obtains points on the boundary line of the cutting edge, calculates the distance from the point to the standard center point, and judges whether the cutting edge is qualified. Specifically, it obtains n points evenly distributed on the boundary line of the cutting edge, and then calculates the distance between the n points and the center point of the surface region to obtain the set. Extract the set In the standard set Overlap value Then the overlap value By comparing the cut edge with a standard threshold to determine whether it falls within the threshold range, the system can assess whether the cut edge meets the requirements. By judging the cut edge, the system can more comprehensively determine the surface quality level of the alloy and improve the accuracy of image recognition.

[0037] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0038] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of a unit of an alloy surface inspection system based on image recognition according to the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Please see Figure 1 As shown, an image recognition-based alloy surface inspection system includes:

[0042] An alloy surface image acquisition unit is used to read alloy surface image information through an image acquisition machine.

[0043] The alloy surface image recognition and extraction unit is used to preprocess alloy surface image information, obtain the alloy surface grayscale image and perform median filtering noise reduction; and obtain contour information through edge detection algorithm.

[0044] The analysis unit is used to acquire the structural features of the contour information and analyze the surface quality of the alloy by detecting the structural features of the contour information.

[0045] The above technical solution involves: setting up an alloy surface image acquisition unit to collect alloy surface information via an image acquisition camera; filtering the alloy surface image coordinates to locate the alloy module; and transmitting the acquired information to a reader card for timely collection and transmission to the camera computer. The camera performs real-time acquisition and transmission of alloy surface information to ensure the stability of the read data. An alloy surface image recognition and extraction unit performs hierarchical preprocessing on the read alloy surface anomaly information. First, it improves the visual presentation of the original image by increasing image clarity, eliminating interference information, and converting the image into a mode that is easy to analyze and process. Then, based on the alloy surface region, median filtering is used to adjust pixels, and an appropriate filter window size is selected to sort the pixel values ​​within the neighborhood of odd-numbered pixels in the image, achieving the effect of removing isolated noise. To enhance image detection performance, this invention extracts the boundary between the detected object and the background image and uses local image differentiation technology to obtain edge measurement operators. The Canny algorithm is employed for convenient edge detection direction acquisition. An analysis unit is set up to determine the structural information of the alloy surface image based on the alloy surface contour obtained in the previous step. During the production of alloys and other materials, there are handling, handover, and transfer processes. Alloy surface inspection is a crucial quality control step in factory production. Common abnormal structures on alloy surfaces include scratches, bumps, and pits. This invention determines the detection quality of the detection contour by obtaining its longest and shortest sides, perimeter, and area. By understanding the structural features of the detection contour, it is easier to identify the source of abnormal image information on the surface, ensuring reduced loss during alloy manufacturing and improving the accuracy of alloy surface detection.

[0046] The specific steps for analyzing the surface quality of alloys are as follows:

[0047] S100, Obtain the longest side of the contour of the alloy surface anomaly image. Shortest side Outline perimeter and outline area Through formula Calculate the contour structure value of the abnormal image , , These are the weighting coefficients. , All are greater than 0;

[0048] S200, The abnormal image contour structure value With preset threshold and A comparison was conducted, among which < :

[0049] like ≤ Then the outline of the abnormal image on the alloy surface is determined as follows: crack;

[0050] like < < If so, the abnormal image outline on the alloy surface is determined to be a coarse crack;

[0051] like ≥ If so, the abnormal image contour of the alloy surface is determined to be a pit or a bump.

[0052] In one specific embodiment of this application, the contour features of the alloy anomaly image are detected and obtained by setting an analysis unit. This application can obtain corresponding outer contour parameters based on the alloy surface anomaly features. Specific parameters include the longest side length, shortest side length, contour perimeter, and area within the contour region. Typically, alloy surfaces are accompanied by cracks, bumps, and pits. Cracks are characterized by a large deviation between the longest and shortest side lengths and a large deviation between the area and perimeter, while bumps or pits have a small deviation between the longest and shortest side lengths and a large area. Therefore, the formula is used: Calculate the contour structure value of the abnormal image , to obtain abnormal image contour structure values With preset threshold and By comparing the data, we can make a preliminary judgment on the type of abnormal structure, which will facilitate adjustments during the processing and subsequent maintenance and repair.

[0053] By setting an illumination strategy, the abnormal image contours on the alloy surface are determined to be either pits or bumps. The specific detection steps are as follows:

[0054] S300: By acquiring the grayscale image of the abnormal image contour of the alloy surface, an illumination strategy is executed to obtain the ratio of white grayscale pixels to all pixels in the grayscale image under different illumination intensities, forming a variation curve.

[0055] S400: Compare the shape of the changing curve with the preset standard curve:

[0056] If the shape curve matches the standard curve, the abnormal profile of the alloy surface is a bump;

[0057] If the shape curve does not match the standard curve, the abnormal profile of the alloy surface will be a pit.

[0058] In one specific embodiment of this application, an illumination strategy is set to further determine the contour of anomalies on the alloy surface. The illumination strategy measures changes in light intensity at different levels, and grayscale images of the anomalies on the alloy surface are acquired under corresponding light intensity conditions. The ratio of white grayscale to all pixel values ​​in the measured range is statistically analyzed in the grayscale images under different light intensities. A change curve is formed by recording the ratios under different light intensities. Since pits reflect less light and appear darker, the proportion of white grayscale is relatively small, while bumps reflect more light and appear brighter, the proportion of white grayscale is relatively large. Due to the characteristics of pits and bumps, the change curves are compared with the shapes corresponding to preset standard curves. The curve formed by the numerical ratios corresponding to bumps is the preset standard curve. The degree of conformity with the standard curve determines whether it is a bump or a pit. By setting the illumination strategy, standard curves are formed under different light intensity conditions to facilitate a more comprehensive detection of the specific structural type of anomalies on the alloy surface.

[0059] In one embodiment of this application, the alloy surface image recognition and extraction unit selects the coordinate points of the collected coordinate points, selects the coordinate points of the corresponding abnormal images, and calculates whether the abnormal coordinates fall within the standard threshold range. Specifically, it compares the marked coordinate points with the standard recorded alloy point coordinates. If they are within the range, the sampling is deemed qualified; otherwise, it is deemed unqualified. By comparing the coordinate points, the location corresponding to the abnormal image of the target alloy surface can be screened, ensuring the accuracy of the screening.

[0060] The analysis unit is also used to determine whether the alloy cutting edge is accurate. The specific operation steps are as follows:

[0061] SS100: Obtain n points evenly distributed along the boundary line of the cutting edge;

[0062] SS200: Calculate the distances between n points and the center point of the region to obtain the set. Extract the set In the standard set Overlap value ;

[0063] SS300, overlap value Compare with the standard threshold:

[0064] like If the cut edge is within the standard threshold range, it is considered acceptable.

[0065] like If the cut edge is not within the threshold range, it is deemed unqualified.

[0066] The above technical solution includes not only quality inspection of the alloy surface but also inspection of the alloy cutting edges. It primarily utilizes an alloy surface analysis unit for analysis. Since the cut alloy plates or shapes have defined templates and shape standards, and the distance from the edge shape to the center has specified dimensions, the system obtains points on the boundary line of the cutting edge, calculates the distance from each point to the standard center point, and determines whether the cutting edge is qualified. Specifically, it obtains n points evenly distributed on the boundary line of the cutting edge, and then calculates the distances between these n points and the center point of the surface region to obtain a set. Extract the set In the standard set Overlap value Then the overlap value By comparing the cut edge with a standard threshold to determine whether it falls within the threshold range, the system can assess whether the cut edge meets the requirements. By judging the cut edge, the system can more comprehensively determine the surface quality level of the alloy and improve the accuracy of image recognition.

[0067] The alloy surface image recognition and extraction unit includes sampling and detecting grayscale features at the target location, and setting grayscale values. grayscale values Compared with standard grayscale threshold Compare:

[0068] like If so, the grayscale value setting is reasonable;

[0069] like If the grayscale value is set incorrectly, adjust the difference between the corresponding pixel and other surrounding pixels.

[0070] Through the above technical solution: In the alloy surface image recognition unit, the detected image needs to be preprocessed to obtain a grayscale image. In the embodiments of the present invention, scratches on the alloy surface are difficult to distinguish, or interference from dust and other debris on the alloy surface can easily lead to poor detection results. By setting a grayscale value standard, it is easier to analyze the effect of post-processing of the alloy surface after being captured by the camera. By sampling and detecting the grayscale value features of the target location, the grayscale value is set. grayscale values Compared with standard grayscale threshold A comparison is performed to determine the reasonableness of the grayscale value setting within the threshold range. If it does not meet the requirements, the difference between the calculated position and other surrounding pixels is adjusted to ensure that the detection results are more reasonable.

[0071] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in the claims, they should all fall within the protection scope of the present invention.

Claims

1. An alloy surface inspection system based on image recognition, characterized in that, include: An alloy surface image acquisition unit is used to read alloy surface image information through an image acquisition machine. The alloy surface image recognition and extraction unit is used to preprocess the alloy surface image information, obtain the alloy surface grayscale image, and perform median filtering and noise reduction processing. Contour information is obtained through edge detection algorithms; The analysis unit is used to acquire the structural features of the contour information and analyze the surface quality of the alloy by detecting the structural features of the contour information. The specific steps for the analysis unit to detect the surface quality of the alloy are as follows: S100, Obtain the longest side of the contour of the alloy surface anomaly image. Shortest side Outline perimeter and outline area Through formula Calculate the image contour structure value , , These are the weighting coefficients. , All are greater than 0; S200, Image contour structure value With preset threshold and A comparison was made, among which < : like ≤ Then the outline of the abnormal image on the alloy surface is determined as follows: crack; like < < If so, the abnormal image outline on the alloy surface is determined to be a coarse crack; like ≥ If so, the abnormal image contour of the alloy surface is determined to be a pit or a bump; The detection steps involve determining whether the abnormal image contours on the alloy surface are pits or bumps by setting an illumination strategy. S300: By acquiring the grayscale image of the abnormal image contour of the alloy surface, an illumination strategy is executed to obtain the ratio of white grayscale pixels to all pixels in the grayscale image under different illumination intensities, forming a variation curve. S400: Compare the shape of the changing curve with the preset standard curve: If the shape curve matches the standard curve, the abnormal profile of the alloy surface is a bump; If the shape curve does not match the standard curve, the abnormal profile of the alloy surface will be a pit.

2. The alloy surface inspection system based on image recognition according to claim 1, characterized in that, The alloy surface image acquisition unit analyzes the sample at the acquired coordinate points and compares the coordinate points of the alloy surface image with a standard threshold: if they are within the standard threshold range, the test is qualified; otherwise, it is unqualified.

3. The alloy surface inspection system based on image recognition according to claim 1, characterized in that, The analysis unit is also used to determine whether the alloy cutting edge is accurate. The specific operation steps are as follows: SS100: Obtain n points evenly distributed along the boundary line of the cutting edge; SS200: Calculate the distances between n points and the center point of the region to obtain the set. Extract the set In the standard set Overlap value ; SS300, overlap value Compare with the standard threshold: like If the cut edge is within the standard threshold range, it is considered acceptable. like If the cut edge is not within the threshold range, it is deemed unqualified.

4. The alloy surface inspection system based on image recognition according to claim 3, characterized in that, The alloy surface image recognition and extraction unit includes sampling and detecting grayscale features at the target location, and setting grayscale values. grayscale values Compared with standard grayscale threshold Compare: like If so, the grayscale value setting is reasonable; like If the grayscale value is set incorrectly, adjust the difference between the corresponding pixel and other surrounding pixels.

Images