CN118570072A - Burr detection method and system in titanium metal processing process - Google Patents

Abstract

The present invention relates to the field of image analysis, and in particular to a method and system for detecting burrs in a titanium metal processing process. The method first obtains a grayscale image of the surface of the titanium metal to be processed, and performs edge detection to obtain processing edge lines, suspected burr edge lines and texture edge lines in the grayscale image, obtains the burr possibility of the suspected burr edge line according to the change of the grayscale of each pixel point on the suspected burr edge line, and the difference in the grayscale change of the pixel point on the processing edge line, analyzes the gradient value of each pixel point on the reference texture edge line of the suspected burr edge line, obtains the burr fuzzy factor of the suspected burr edge line, and then filters each suspected burr edge line based on the obtained noise influence factor to obtain an enhanced image, and performs edge detection on the enhanced image to obtain the burr position in the enhanced image. The present invention eliminates the interference of noise on burr detection and improves the detection accuracy of burrs on the titanium metal processing surface.

Description

A burr detection method and system during titanium metal processing

Technical Field

The invention relates to the field of image analysis, and in particular to a burr detection method and system in a titanium metal processing process.

Background Art

During the titanium metal cutting process, due to the thrust of the tool, part of the material is squeezed and stretched to form relatively prominent burrs. The presence of burrs will increase the difficulty and cost of subsequent processing steps and affect the performance of titanium metal products. Therefore, it is necessary to detect the burrs that appear during the titanium metal processing process to ensure the quality of titanium metal products.

In the related technology, machine vision detection technology is usually used to detect the burr parts of titanium metal processing surface images. However, the presence of noise in the image will cause changes in the grayscale value of the pixel points, resulting in the inability of existing methods to accurately distinguish between noise and burrs, thereby failing to effectively remove the noise in the image, thereby reducing the detection accuracy of burrs on the titanium metal processing surface.

Summary of the invention

In order to solve the technical problem that the existing methods cannot accurately distinguish between noise and burrs, thus cannot effectively remove noise in the image, and reduce the detection accuracy of burrs on the titanium metal processing surface, the purpose of the present invention is to provide a burr detection method and system in the titanium metal processing process, and the technical scheme adopted is as follows:

The present invention proposes a method for detecting burrs in a titanium metal processing process, the method comprising:

Obtain a grayscale image of the titanium metal surface to be processed, perform edge detection on the grayscale image, and obtain processing edge lines, suspected burr edge lines, and texture edge lines in the grayscale image;

According to the distribution of the grayscale values of each pixel point in a preset neighborhood centered on each pixel point, a grayscale variation factor of each pixel point is obtained; any suspected burr edge line is used as a target edge line, and two processed edge lines adjacent to the target edge line are used as reference processed edge lines of the target edge line, and according to the grayscale variation factor of each pixel point on the target edge line, and the difference in the grayscale variation factor of the pixel point between the target edge line and the reference processed edge line, the burr possibility of the target edge line is obtained;

Obtain the centroid of the target edge line and each texture edge line, and use the distance between the centroid of the target edge line and each texture edge line as the distance parameter of each texture edge line; select a preset number of texture edge lines with the smallest distance parameters as reference texture edge lines of the target edge line; obtain the burr fuzzy factor of the target edge line according to the gradient value of each pixel point on each reference texture edge line and the distance parameter of each reference texture edge line;

According to the difference between the burr possibility and the burr fuzzy factor of the target edge line, a noise impact factor of the target edge line is obtained; based on the noise impact factor of each suspected burr edge line, a grayscale image is filtered to obtain an enhanced image;

Perform edge detection on the enhanced image to obtain the burr location in the enhanced image.

Furthermore, obtaining the processed edge lines, suspected burr edge lines and texture edge lines in the grayscale image includes:

Based on the edge detection algorithm, edge detection is performed on the grayscale image to obtain the processing edge line and texture edge line in the grayscale image;

The processing edge line is compared with a preset processing trajectory, and a missing portion of the processing edge line relative to the preset processing trajectory is used as a suspected burr edge line.

Furthermore, obtaining the grayscale change factor of each pixel includes:

Any pixel in the grayscale image is taken as the target pixel, and any pixel in the preset neighborhood centered on the target pixel is taken as the pixel to be tested; if the grayscale value of the pixel to be tested is greater than or equal to the grayscale values of all pixels in the preset window centered on the pixel to be tested, the pixel to be tested is marked as a maximum pixel, and all maximum pixels in the preset neighborhood are obtained;

Take any maximum pixel as the target maximum pixel, and take other maximum pixels closest to the target maximum pixel as reference maximum pixels of the target maximum pixel;

The average value of the grayscale values of the target maximum pixel and the corresponding reference maximum pixel is used as the first grayscale parameter of the target maximum pixel; the overall level of the grayscale values of all pixels on the line between the target maximum pixel and the corresponding reference maximum pixel is analyzed to obtain the second grayscale parameter of the target maximum pixel; according to the first grayscale parameter and the second grayscale parameter, an initial grayscale change parameter of the target maximum pixel is obtained, wherein the initial grayscale change parameter is positively correlated with the first grayscale parameter, and the initial grayscale change parameter is negatively correlated with the second grayscale parameter;

Performing negative correlation mapping on the distance between the target maximum pixel and the corresponding reference maximum pixel to obtain a first distance weight of the target maximum pixel, and using the first distance weight of the target maximum pixel to weight the initial grayscale change parameter to obtain a true grayscale change parameter of the target maximum pixel;

The overall level of the true grayscale change parameters of all maximum pixel points in a preset neighborhood is analyzed and normalized to obtain the grayscale change factor of the target pixel point.

Furthermore, the possibility of obtaining the burr of the target edge line includes:

Analyzing the overall level of the grayscale change factor of all pixels on the target edge line or each reference processed edge line to obtain the overall grayscale change factor of the target edge line or each reference processed edge line;

According to the difference of the overall grayscale change factor between the target edge line and each reference processed edge line, a grayscale change difference value of each reference processed edge line is obtained, and the maximum value of the grayscale change difference values of all reference processed edge lines is used as the grayscale change characteristic value of the target edge line;

According to the overall grayscale change factor and the grayscale change characteristic value of the target edge line, the burr possibility of the target edge line is obtained, and the burr possibility is positively correlated with the grayscale change characteristic value of the target edge line, and the burr possibility is negatively correlated with the overall grayscale change factor of the target edge line. The burr possibility is a normalized value.

Furthermore, obtaining the burr fuzzy factor of the target edge line includes:

Analyzing the overall level of the gradient values of all pixels on each texture edge line to obtain the gradient distribution value of each texture edge line, and analyzing the overall level of the gradient distribution values of all texture edge lines to obtain the overall gradient distribution parameter;

According to the overall gradient distribution parameter and the gradient distribution value of each reference texture edge line, an initial gradient characteristic value of each reference texture edge line is obtained; the initial gradient characteristic value is positively correlated with the overall gradient distribution parameter, and the initial gradient characteristic value is negatively correlated with the gradient distribution value of each reference texture edge line;

Performing negative correlation mapping on the distance parameter of each reference texture edge line to obtain a second distance weight of each reference texture edge line; using the second distance weight of each reference texture edge line, weighting the initial gradient eigenvalue of each reference texture edge line to obtain a true gradient eigenvalue of each reference texture edge line;

The overall level of the true gradient feature values of all reference texture edge lines is analyzed and normalized to obtain the burr fuzzy factor of the target edge line.

Furthermore, the noise impact factor of obtaining the target edge line includes:

The difference between the burr possibility and the burr fuzzy factor according to the target edge line is normalized to obtain the noise impact factor of the target edge line.

Further, obtaining the enhanced image includes:

The noise impact factor of each suspected burr edge line is used as a scale factor used by the Wiener filtering algorithm, and based on the Wiener filtering algorithm, the area where the minimum circumscribed circle of each suspected burr edge line in the grayscale image is located is filtered to obtain an enhanced image.

Furthermore, the performing edge detection on the enhanced image to obtain the burr location in the enhanced image includes:

Perform edge detection on the enhanced image to obtain the optimized processed edge line in the enhanced image;

The optimized processing edge line is compared with the preset processing trajectory, and the missing part of the optimized processing edge line relative to the preset processing trajectory is used as the burr part in the enhanced image.

Furthermore, the preset number is in the range of 10 to 30 integers.

The present invention also proposes a burr detection system during titanium metal processing, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements any one of the steps of a burr detection method during titanium metal processing when executing the computer program.

The present invention has the following beneficial effects:

The present invention takes into account that the presence of noise in an image may cause changes in the grayscale values of pixels, resulting in the inability of existing methods to accurately distinguish between noise and burrs, thereby failing to effectively remove noise in the image and reducing the detection accuracy of burrs on the titanium metal processing surface. Therefore, the present invention first obtains a grayscale image of the titanium metal surface to be processed, and performs edge detection on the grayscale image to obtain processing edge lines, suspected burr edge lines and texture edge lines in the grayscale image. Considering that the noise and burrs existing in the grayscale image will cause changes in the grayscale values of pixels in the local area, the grayscale change factor can be used to reflect the change characteristics of the local grayscale value of each pixel, thereby providing a data basis for the subsequent distinction between noise and burrs. Compared with the suspected burr edge line formed by noise, the grayscale change factor of the pixel on the suspected burr edge line formed by the burr is smaller, and is consistent with the pixel on the reference processing edge line. The difference in the grayscale change factors of the points is large, so the burr possibility can be used to reflect the possibility of the existence of burrs on the target edge line. Since light will reduce the contrast of the area around the burr, the visual effect of the surrounding area is blurred, resulting in the decrease of the pixel gradient value on the texture edge line in the surrounding area, while the pixel gradient value on the texture edge line in the area around the suspected burr edge line formed by noise will not change significantly. Therefore, the burr fuzzy factor obtained can be used to reflect the possibility that the target edge line is affected by burrs rather than noise. At the same time, the presence of noise will cause a large difference between the burr possibility and the burr fuzzy factor of the target edge line. Therefore, the noise influence factor obtained can be used to reflect the degree to which the target edge line is affected by noise, and then the grayscale image is filtered to eliminate the interference of noise in the grayscale image on burr detection, thereby improving the detection accuracy of burrs on the titanium metal processing surface.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

FIG1 is a flow chart of a method for detecting burrs in a titanium metal processing process provided by an embodiment of the present invention;

FIG2 is a schematic diagram showing the distribution of a processing edge line, a suspected burr edge line and a texture edge line provided by an embodiment of the present invention;

FIG3 is a flow chart of a method for obtaining the possibility of burrs on a target edge line provided by one embodiment of the present invention;

FIG. 4 is a flow chart of a method for obtaining a burr fuzzy factor of a target edge line provided by an embodiment of the present invention.

DETAILED DESCRIPTION

In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a burr detection method and system in the titanium metal processing process proposed by the present invention, its specific implementation, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

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

The specific scheme of a burr detection method and system in titanium metal processing provided by the present invention is described in detail below with reference to the accompanying drawings.

Please refer to FIG. 1 , which shows a flow chart of a burr detection method during titanium metal processing provided by one embodiment of the present invention. The method comprises:

Step S1: obtaining a grayscale image of the titanium metal surface to be processed, performing edge detection on the grayscale image, and obtaining processing edge lines, suspected burr edge lines, and texture edge lines in the grayscale image.

During the titanium metal cutting process, due to the thrust of the tool, part of the material is squeezed and stretched to form relatively prominent burrs. The presence of burrs will increase the difficulty and cost of subsequent processing steps and affect the performance of titanium metal products. Therefore, it is necessary to detect the burrs appearing in the titanium metal processing process to ensure the quality of titanium metal products. In the related art, machine vision detection technologies such as edge detection or threshold segmentation are usually used to detect the burr parts of the titanium metal processing surface image. However, the presence of noise in the image will cause the grayscale value of the pixel to change, resulting in the inability of the existing method to accurately distinguish between noise and burrs, thereby failing to effectively remove the noise in the image, thereby reducing the detection accuracy of burrs on the titanium metal processing surface. Therefore, an embodiment of the present invention proposes a burr detection method in the titanium metal processing process to solve this problem.

The embodiment of the present invention first uses an industrial camera to collect the original image of the titanium metal surface to be processed. Since the original image is generally a multi-channel image, it will increase the complexity of subsequent calculations. Therefore, in order to reduce the amount of calculation of subsequent image processing and improve the processing speed, in one embodiment of the present invention, the collected original image is grayed and converted into a single-channel gray image. It should be noted that graying is a technical means well known to those skilled in the art and will not be described in detail here.

During the cutting process of titanium metal, burrs will appear at certain local positions on the edge formed after cutting. Since the burr surface is relatively rough, the cutting edge features presented in the grayscale image at the burr position are masked. In other words, the cutting edge features at the burr position are not particularly obvious. Based on this feature, edge detection can be performed on the grayscale image to obtain the processing edge line and suspected burr edge line in the grayscale image. At the same time, there are texture features on the surface of titanium metal. Therefore, after edge detection, the texture edge line of its surface can also be obtained. Subsequently, the suspected burr edge line, the processing edge line and the texture edge line can be further analyzed to effectively remove the noise in the grayscale image and improve the detection accuracy of the burr. It should be noted that since the noise in the image will also mask the edge features after cutting, there are two factors that form the suspected burr edge line. One is the suspected burr edge line formed by the burr itself, and the other is the suspected burr edge line formed by noise.

Preferably, in one embodiment of the present invention, the method for obtaining the processed edge line, the suspected burr edge line and the texture edge line in the grayscale image specifically includes:

Based on an edge detection algorithm, such as the Canny edge detection algorithm or other edge detection methods, edge detection is performed on the grayscale image to obtain processing edge lines and texture edge lines in the grayscale image. Since the burrs and noise in the grayscale image will mask the edge features formed during the cutting process, the edges formed during the processing are not continuous. That is to say, multiple processing edge lines can be detected by the edge detection algorithm, and the disconnection and missing phenomenon between the processing edge lines is caused by the burrs and noise. Therefore, the processing edge line can be compared with the preset processing trajectory, and the missing part of the processing edge line relative to the preset processing trajectory is taken as a suspected burr edge line, wherein the preset processing trajectory is a known processing trajectory designed before the titanium metal is processed. In the ideal case where there are no burrs and noise, the edge line formed by the titanium metal after processing is complete and the same as the preset processing trajectory. Please refer to Figure 2, which shows a distribution diagram of a processing edge line, a suspected burr edge line and a texture edge line provided by an embodiment of the present invention, wherein lines A~F are processing edge lines, lines a~f are suspected burr edge lines, and lines 1~7 are texture edge lines.

Step S2: according to the distribution of grayscale values of each pixel point in a preset neighborhood centered on each pixel point, obtain the grayscale variation factor of each pixel point; take any suspected burr edge line as the target edge line, and take the two processed edge lines adjacent to the target edge line as the reference processed edge lines of the target edge line, and according to the grayscale variation factor of each pixel point on the target edge line, and the difference in the grayscale variation factor of the pixel point between the target edge line and the reference processed edge line, obtain the burr possibility of the target edge line.

From the above analysis, it can be seen that there are two types of suspected burr edge lines among all the obtained suspected burr edge lines, one is the suspected burr edge line formed by burrs, and the other is the suspected burr edge line formed by noise, which causes noise to affect the accurate detection of burrs. Compared with burrs, noise is more random, and the degree of change in the grayscale value of local pixels caused by noise is greater, resulting in differences in the local distribution characteristics of the grayscale values of pixels on the two suspected burr edge lines, and the local distribution characteristics of the grayscale values of pixels between the two suspected burr edge lines and the processed edge lines are also different. Therefore, the distribution of the grayscale values of each pixel in a preset neighborhood centered on each pixel can be analyzed first, and the local grayscale change characteristics of each pixel can be reflected by the obtained grayscale change factor, so as to provide a data basis for the subsequent distinction between noise and burrs, wherein the size of the preset neighborhood is generally an odd number greater than 3. In one embodiment of the present invention, the size of the preset neighborhood is set to 5, that is, the preset neighborhood is a The specific size of the preset neighborhood can also be set by the implementer according to the specific implementation scenario and is not limited here.

Preferably, in an embodiment of the present invention, the method for obtaining the grayscale variation factor of each pixel point specifically includes:

First, any pixel point in the grayscale image is taken as the target pixel point, and any pixel point in the preset neighborhood centered on the target pixel point is taken as the pixel point to be tested; if the grayscale value of the pixel point to be tested is greater than or equal to the grayscale values of all pixels in the preset window centered on the pixel point to be tested, the pixel point to be tested is marked as a maximum pixel point, and all maximum pixels in the preset neighborhood are obtained, wherein the size of the preset window needs to be smaller than the size of the preset neighborhood. In one embodiment of the present invention, the size of the preset window is set to 3, and the specific size of the preset window can also be set by the implementer according to the specific implementation scenario, which is not limited here.

It should be noted that, for the pixel points at the boundary of the preset neighborhood, it is impossible to establish a preset window with them as the center. In this case, the boundary of the preset neighborhood can be expanded. For example, in one embodiment of the present invention, the size of the preset neighborhood is 5, and the size of the preset window is 3. In this case, the size of the preset neighborhood can be expanded to 7, wherein the expanded pixel points do not belong to the preset neighborhood, but are just convenient for establishing a preset window with the pixel points at the boundary of the preset neighborhood as the center, so that the analysis can proceed smoothly.

Then, any maximum pixel point is taken as the target maximum pixel point, and the other maximum pixel points closest to the target maximum pixel point are taken as the reference maximum pixel points of the target maximum pixel point; the average value of the grayscale value of the target maximum pixel point and the corresponding reference maximum pixel point is taken as the first grayscale parameter of the target maximum pixel point; the overall level of the grayscale values of all pixels on the line between the target maximum pixel point and the corresponding reference maximum pixel point is analyzed to obtain the second grayscale parameter of the target maximum pixel point. The larger the first grayscale parameter of the target maximum pixel point is relative to the second grayscale parameter, the greater the degree of change of the grayscale value of the pixel point between the target maximum pixel point and the reference maximum pixel point. Therefore, the initial grayscale change parameter of the target maximum pixel point can be obtained according to the first grayscale parameter and the second grayscale parameter. The initial grayscale change parameter is positively correlated with the first grayscale parameter, and the initial grayscale change parameter is negatively correlated with the second grayscale parameter. The larger the initial grayscale change parameter is, the greater the degree of change of the grayscale value of the pixel point between the target maximum pixel point and the reference maximum pixel point.

In an embodiment of the present invention, the average or median of the grayscale values of all pixels on the line between the target maximum pixel and the corresponding reference maximum pixel can be used as the second grayscale parameter of the target maximum pixel to achieve an overall level analysis of the grayscale values of all pixels on the line between the target maximum pixel and the corresponding reference maximum pixel, which is not limited here.

In one embodiment of the present invention, the first grayscale parameter of the target maximum pixel point can be used as the numerator, the second grayscale parameter of the target maximum pixel point can be used as the denominator, and the ratio can be used as the initial grayscale change parameter of the target maximum pixel point.

Finally, considering that the closer the distance between the target maximum pixel and the corresponding reference maximum pixel is, the faster the grayscale value of the pixel between the two changes, the distance between the target maximum pixel and the corresponding reference maximum pixel can be negatively correlated to obtain the first distance weight of the target maximum pixel, and the initial grayscale change parameter is weighted using the first distance weight of the target maximum pixel to obtain the real grayscale change parameter of the target maximum pixel. The real grayscale change parameter of each maximum pixel in the preset neighborhood can be obtained by the same method as above. The larger the real grayscale change parameter, the more drastic the change in the grayscale value of the pixel between the maximum pixel and the reference maximum pixel. Then, the overall level of the real grayscale change parameters of all the maximum pixels in the preset neighborhood is analyzed and normalized to obtain the grayscale change factor of the target pixel. The larger the grayscale change factor, the greater the degree of change in the grayscale value of the pixel in the preset neighborhood of the target pixel.

In the embodiment of the present invention, a negative correlation function such as an inverse proportional function may be used to perform negative correlation mapping, which is not limited herein.

In an embodiment of the present invention, the average value or median of the true grayscale change parameters of all maximum pixels in a preset neighborhood can be used as the grayscale change factor of the target pixel to achieve an overall level analysis of the true grayscale change parameters of all maximum pixels in a preset neighborhood, which is not limited here.

In one embodiment of the present invention, the normalization processing can be specifically, for example, maximum and minimum value normalization processing, and the normalization in subsequent steps can all adopt maximum and minimum value normalization processing. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of numerical values, which will not be described in detail.

The expression of the grayscale change factor of the target pixel point can be specifically, for example, as follows:

in, Indicates the grayscale change factor of the target pixel; Indicates the first The first grayscale parameter of the maximum pixel; Indicates the first The second grayscale parameter of the maximum pixel; Indicates the first The initial grayscale change parameter of the maximum pixel; Indicates the first The distance between a maximum pixel and the corresponding reference maximum pixel can be calculated using the Euclidean distance; Indicates the first The first distance weight of the maximum pixel; Indicates the first The real grayscale change parameter of the maximum pixel point; Indicates the number of maximum pixels in a preset neighborhood centered on the target pixel; Represents the normalization function.

The grayscale change factor of each pixel in the grayscale image can be obtained by the same method as above.

From the above analysis, it can be seen that compared with burrs, noise is more random, and the degree of change in the grayscale value of local pixels caused by noise is greater, resulting in differences in the local distribution characteristics of the grayscale values of pixels on the two suspected burr edge lines. At the same time, the grayscale value distribution of the local area of the pixels on the processed edge line is relatively uniform and smooth, so the local distribution characteristics of the grayscale values of the pixels between the two suspected burr edge lines and the processed edge line are also different, which is specifically manifested as follows: compared with the suspected burr edge line formed by noise, the degree of change in the local grayscale value of the pixels on the suspected burr edge line formed by burrs is smaller, and the difference in the change of the local grayscale value of the pixels on the processed edge line is larger. Therefore, in the subsequent process, any suspected burr edge line can be targeted first. The target edge line is marked, and the two processed edge lines adjacent to the target edge line are used as the reference processed edge lines of the target edge line, so as to facilitate the subsequent comparative analysis. Please refer to Figure 2. Assuming that the suspected burr edge line a is the target edge line, the processed edge lines A and F are the reference processed edge lines of the suspected burr edge line a. Then, the grayscale change factor of each pixel point on the target edge line and the difference in the grayscale change factor of the pixel point between the target edge line and the reference processed edge line are analyzed. The burr possibility obtained reflects the possibility that the target edge line is formed by a burr. The greater the burr possibility, the greater the possibility that there is a burr on the target edge line. Further analysis can be made based on the burr possibility, thereby effectively removing the noise in the grayscale image and improving the accuracy of burr detection on the titanium metal processing surface.

Preferably, in one embodiment of the present invention, the method for obtaining the burr possibility of the target edge line specifically includes:

Please refer to FIG. 3 , which shows a flow chart of a method for obtaining the burr possibility of a target edge line provided by an embodiment of the present invention.

Step S201: Analyze the overall level of the grayscale variation factor of all pixels on the target edge line and each reference processed edge line respectively to obtain the overall grayscale variation factor of the target edge line and each reference processed edge line.

From the above analysis, it can be seen that compared with the suspected burr edge line formed by noise, the degree of change of the local grayscale value of the pixels on the suspected burr edge line formed by burrs is smaller, and the difference with the change of the local grayscale value of the pixels on the processed edge line is larger. Therefore, the overall level of the grayscale change factor of all pixels on the target edge line and each reference processed edge line can be analyzed respectively to obtain the overall grayscale change factor of the target edge line and each reference processed edge line. The larger the overall grayscale change factor of the target edge line or each reference processed edge line, the greater the degree of local grayscale change of the pixels on the target edge line or each reference processed edge line, which provides a data basis for the subsequent calculation and analysis of the possibility of burrs on the target edge line.

In an embodiment of the present invention, the average value or median of the grayscale change factors of all pixels on the target edge line or each reference processing edge line can be used as the overall grayscale change factor of the target edge line or each reference processing edge line, so as to realize the overall level analysis of the grayscale change factors of all pixels on the target edge line or each reference processing edge line. No limitation is made here, and specifically: the average value or median of the grayscale change factors of all pixels on the target edge line is used as the overall grayscale change factor of the target edge line, and the average value or median of the grayscale change factors of all pixels on each reference processing edge line is used as the overall grayscale change factor of each reference processing edge line.

Step S202: According to the difference in the overall grayscale change factor between the target edge line and each reference processed edge line, the grayscale change difference value of each reference processed edge line is obtained, and the maximum value of the grayscale change difference values of all reference processed edge lines is used as the grayscale change characteristic value of the target edge line.

The pixel points on the suspected burr edge line formed by the burrs have a large difference in the change of local grayscale values from the pixel points on the processed edge line. Specifically, the difference in the overall grayscale change factor between the suspected burr edge line formed by the burrs and the processed edge line is large. Therefore, the grayscale change difference value of each reference processed edge line can be obtained according to the difference in the overall grayscale change factor between the target edge line and each reference processed edge line. The larger the grayscale change difference value, the greater the difference in the change of local grayscale values of the pixel points between each reference processed edge line and the target edge line. The maximum value of the grayscale change difference values of all reference processed edge lines is used as the grayscale change characteristic value of the target edge line. The larger the grayscale change characteristic value, the greater the difference in the change of local grayscale values between the pixel points on the target edge line and the pixel points on the reference processed edge line, which further indicates that burrs are more likely to exist on the target edge line.

In one embodiment of the present invention, the absolute value of the difference in the overall grayscale change factor between the target edge line and each reference processing edge line can be used as the grayscale change difference value of each reference processing edge line, thereby realizing the analysis of the difference in the overall grayscale change factor between the target edge line and each reference processing edge line.

Step S203: According to the overall grayscale change factor and grayscale change characteristic value of the target edge line, the burr possibility of the target edge line is obtained. The burr possibility is positively correlated with the grayscale change characteristic value of the target edge line, and negatively correlated with the overall grayscale change factor of the target edge line. The burr possibility is a normalized value.

Compared with the suspected burr edge line formed by noise, the degree of change of the local grayscale value of the pixel points on the suspected burr edge line formed by burrs is smaller, and the difference with the change of the local grayscale value of the pixel points on the processed edge line is larger. Therefore, the smaller the overall grayscale change factor of the target edge line obtained above and the larger the grayscale change characteristic value of the target edge line, the more likely the target edge line is formed by burrs, and the more likely burrs are present on the target edge line, and the less likely noise is present. Therefore, the burr possibility of the target edge line can be obtained according to the overall grayscale change factor and grayscale change characteristic value of the target edge line, wherein the burr possibility is positively correlated with the grayscale change characteristic value of the target edge line, and the burr possibility is negatively correlated with the overall grayscale change factor of the target edge line, and the burr possibility is the normalized value.

In one embodiment of the present invention, the grayscale change characteristic value of the target edge line can be used as the numerator, the overall grayscale change factor of the target edge line can be used as the denominator, the ratio can be used as the burr parameter of the target edge line, and the burr parameter of the target edge line can be normalized to obtain the burr possibility of the target edge line.

The expression of the burr possibility of the target edge line can be specifically expressed as:

in, Indicates the possibility of burr on the edge line of the target; Represents the overall grayscale change factor of the target edge line; Represents the overall grayscale change factor of the first reference processed edge line of the target edge line; The overall grayscale variation factor of the second reference processed edge line representing the target edge line; Indicates the grayscale change difference value of the first reference processed edge line; Indicates the grayscale change difference value of the second reference processed edge line; Represents the grayscale change characteristic value of the target edge line; Indicates the burr parameter of the target edge line; represents the function of taking the maximum value; Represents the normalization function.

At this point, the burr possibility of the target edge line is obtained. Subsequently, based on the burr possibility of the target edge line, the degree to which the target edge line is affected by noise can be analyzed, thereby effectively removing the noise in the grayscale image and improving the detection accuracy of burrs on the titanium metal processing surface.

Step S3: Obtain the centroid of the target edge line and each texture edge line, and use the distance between the centroid of the target edge line and each texture edge line as the distance parameter of each texture edge line; select a preset number of texture edge lines with the smallest distance parameters as reference texture edge lines for the target edge line; obtain the burr fuzzy factor of the target edge line according to the gradient value of each pixel point on each reference texture edge line and the distance parameter of each reference texture edge line.

Since the burrs are relatively prominent, when illuminated, the illumination will reduce the contrast of the area around the burrs, making the visual effect of the area around the burrs more blurred, thereby causing the pixel gradient value on the texture edge line of the surrounding area to become smaller. Therefore, the pixel gradient value on the texture edge line of the area around the suspected burr edge line formed by the burrs is relatively small, but the pixel gradient value on the texture edge line of the area around the suspected burr edge line formed by noise will not change significantly. Therefore, the target edge line and the centroid of each texture edge line can be obtained first, wherein the method for obtaining the centroid of the edge line in the image is a technical means well known to those skilled in the art, which will not be elaborated here. The distance between the centroid of the target edge line and each texture edge line is used as the distance parameter of each texture edge line. The distance between the centroids can be specifically the Euclidean distance, and then a preset number of texture edge lines with the smallest distance parameters are selected as reference texture edge lines in the area around the target edge line. Subsequently, the burr fuzziness factor of the target edge line can be calculated and analyzed based on the pixel gradient value on the reference texture edge line.

The preset number has a value range of integers from 10 to 30. In one embodiment of the present invention, the preset number is set to 20. The specific value of the preset number can also be set by the implementer according to the specific implementation scenario and is not limited here.

From the above analysis, it can be seen that the gradient value of the pixel points on the texture edge line in the area around the suspected burr edge line formed by the burrs is relatively small, but the gradient value of the pixel points on the texture edge line in the area around the suspected burr edge line formed by the noise will not change significantly. Therefore, the gradient value of each pixel point on each reference texture edge line can be analyzed, and the burr fuzzy factor of the target edge line can be obtained by combining the distance parameter of each reference texture edge line. The burr fuzzy factor reflects the possibility that the target edge line is affected by burrs rather than noise. Subsequently, based on the burr fuzzy factor of the target edge line and the burr possibility obtained above, the degree of influence of noise on the target edge line is analyzed.

Preferably, in one embodiment of the present invention, the method for obtaining the burr possibility of the target edge line specifically includes:

Please refer to FIG. 4 , which shows a flow chart of a method for obtaining a burr fuzzy factor of a target edge line provided by an embodiment of the present invention.

Step S301: Analyze the overall level of the gradient values of all pixels on each texture edge line to obtain the gradient distribution value of each texture edge line, and analyze the overall level of the gradient distribution values of all texture edge lines to obtain the overall gradient distribution parameter.

First, the overall level of the gradient values of all pixels on each texture edge line is analyzed to obtain the gradient distribution value of each texture edge line. The gradient distribution value reflects the overall distribution of the gradient values of each pixel on each texture edge line. The gradient value of the pixel can be calculated by existing gradient operators such as the Sobel operator or the Scharr operator, which will not be elaborated here. The overall level of the gradient distribution values of all texture edge lines is further analyzed to obtain the overall gradient distribution parameters. The overall gradient distribution parameters reflect the gradient distribution of each texture edge line, providing a data basis for subsequent calculations.

In an embodiment of the present invention, the average or median of the gradient values of all pixels on each texture edge line can be used as the gradient distribution value of each texture edge line to achieve an overall level analysis of the gradient values of all pixels on each texture edge line, which is not limited here.

In the embodiment of the present invention, the average value or median of the gradient distribution values of all texture edge lines can be used as the overall gradient distribution parameter to implement the overall level analysis of the gradient distribution values of all texture edge lines, which is not limited here.

Step S302: obtaining an initial gradient eigenvalue of each reference texture edge line according to the overall gradient distribution parameter and the gradient distribution value of each reference texture edge line; the initial gradient eigenvalue is positively correlated with the overall gradient distribution parameter, and the initial gradient eigenvalue is negatively correlated with the gradient distribution value of each reference texture edge line.

The pixel gradient values on the texture edge lines in the area surrounding the suspected burr edge lines formed by the burrs are relatively small. Therefore, the smaller the gradient distribution value of each reference texture edge line of the target edge line is relative to the overall gradient distribution parameter, the smaller the pixel gradient values on the texture edge lines in the area surrounding the target edge line are, which further indicates that the possibility that the gradient value of the texture edge lines in the surrounding area is reduced due to the presence of burrs on the target edge line is greater. Therefore, the initial gradient eigenvalue of each reference texture edge line can be obtained according to the overall gradient distribution parameter and the gradient distribution value of each reference texture edge line; the initial gradient eigenvalue is positively correlated with the overall gradient distribution parameter, and the initial gradient eigenvalue is negatively correlated with the gradient distribution value of each reference texture edge line.

In one embodiment of the present invention, the overall gradient distribution parameter may be used as the numerator, the gradient distribution value of each reference texture edge line may be used as the denominator, and the ratio may be used as the initial gradient feature value of each reference texture edge line.

Step S303: negatively correlate the distance parameters of each reference texture edge line to obtain a second distance weight for each reference texture edge line; and weight the initial gradient eigenvalue of each reference texture edge line using the second distance weight for each reference texture edge line to obtain a true gradient eigenvalue for each reference texture edge line.

Considering that the smaller the distance between the centroid of each reference texture edge line and the centroid of the target edge line, the greater the reference value of the reference texture edge line, the distance parameter of each reference texture edge line can be negatively correlated mapped to obtain the second distance weight of each reference texture edge line, and then the second distance weight of each reference texture edge line can be used to weight the initial gradient eigenvalue of each reference texture edge line to obtain the true gradient eigenvalue of each reference texture edge line. Subsequently, the burr fuzzy factor of the target edge line can be analyzed based on the true gradient eigenvalues of all reference texture edge lines.

Step S304: Analyze and normalize the overall level of the true gradient feature values of all reference texture edge lines to obtain the burr fuzzy factor of the target edge line.

The larger the true gradient eigenvalue of the reference texture edge line is, the greater the possibility that the gradient value of the texture edge line in the surrounding area is reduced due to the presence of burrs on the target edge line, that is, the visual effect of the surrounding area is blurred due to the presence of burrs on the target edge line. Therefore, the overall level of the true gradient eigenvalues of all reference texture edge lines of the target edge line can be analyzed and normalized to obtain the burr fuzziness factor of the target edge line.

In the embodiment of the present invention, the overall level analysis of the true gradient eigenvalues of all reference texture edge lines of the target edge line can be achieved by calculating the average or median of the true gradient eigenvalues of all reference texture edge lines of the target edge line, which is not limited here.

The expression of the burr fuzzy factor of the target edge line can be specifically, for example, as follows:

in, Indicates the burr fuzziness factor of the target edge line; The target edge line The distance parameter of the reference texture edge line; The target edge line A second distance weight for a reference texture edge line; represents the overall gradient distribution parameter; The target edge line The gradient distribution value of the reference texture edge line; The target edge line The initial gradient eigenvalues of the reference texture edge lines; The target edge line The true gradient feature value of the reference texture edge line; Represents the normalization function.

So far, the burr fuzziness factor of each target edge line is obtained.

Step S4: according to the difference between the burr possibility and the burr fuzziness factor of the target edge line, the noise impact factor of the target edge line is obtained; based on the noise impact factor of each suspected burr edge line, the grayscale image is filtered to obtain an enhanced image.

Under ideal conditions without being affected by noise, there will be a positive correlation between the burr possibility and the burr fuzzy factor of the target edge line, but noise will break this positive correlation, resulting in a difference between the burr possibility and the burr fuzzy factor of the target edge line. Therefore, the difference between the burr possibility and the burr fuzzy factor of the target edge line can be analyzed, and the noise influence factor obtained can be used to reflect the degree to which the target edge line is affected by noise, so as to facilitate subsequent filtering of different intensities on each suspected burr edge line based on the noise influence factor, thereby removing the influence of noise.

Preferably, in one embodiment of the present invention, the method for obtaining the noise impact factor of the target edge line specifically includes:

The greater the impact of noise on the target edge line, the greater the difference between the burr possibility and the burr fuzzy factor of the target edge line. Therefore, the difference between the burr possibility and the burr fuzzy factor of the target edge line can be normalized to obtain the noise impact factor of the target edge line.

In one embodiment of the present invention, the difference analysis between the burr possibility of the target edge line and the burr fuzzy factor can be realized by calculating the absolute value of the difference between the two.

The expression of the noise impact factor of the target edge line can be specifically expressed as:

in, Represents the noise impact factor of the target edge line; Indicates the possibility of burr on the edge line of the target; Indicates the burr fuzziness factor of the target edge line; Represents the normalization function.

The noise influence factor of each suspected burr edge line can be obtained by the same method as mentioned above. The larger the noise influence factor of each suspected burr edge line is, the greater the degree of influence of noise on the suspected burr edge line is. Therefore, based on the noise influence factor of each suspected burr edge line, the grayscale image can be filtered to obtain an enhanced image, eliminate the interference of noise on burr detection, and improve the subsequent detection accuracy of burrs on the titanium metal processing surface.

Preferably, in one embodiment of the present invention, the method for acquiring an enhanced image specifically includes:

The embodiment of the present invention selects a Wiener filtering algorithm with good information retention capability to filter the grayscale image. Since the larger the noise impact factor, the greater the degree to which the suspected burr edge line is affected by the noise, the noise impact factor of each suspected burr edge line can be used as the proportional factor used by the Wiener filtering algorithm. Based on the Wiener filtering algorithm, the area where the minimum circumscribed circle of each suspected burr edge line in the grayscale image is located is filtered to obtain an enhanced image, so that the target edge lines affected by different degrees of noise are filtered with different intensities, while removing the noise and retaining the characteristics of the burr as much as possible, thereby improving the subsequent detection accuracy of burrs on the titanium metal processing surface. The Wiener filtering algorithm is a technical means well known to those skilled in the art and will not be elaborated here.

At this point, an enhanced image with better quality is obtained, and the burr locations can be accurately detected in the enhanced image later.

Step S5: Perform edge detection on the enhanced image to obtain the burr location in the enhanced image.

Since the noise in the enhanced image is effectively removed, the edge detection of the enhanced image can be performed to obtain the burr location in the enhanced image, thereby improving the detection accuracy of burrs on the titanium metal processing surface.

Preferably, in one embodiment of the present invention, the method for obtaining the burr location specifically includes:

Based on edge detection algorithms such as the Canny edge detection algorithm, edge detection is performed on the enhanced image to obtain the optimized processing edge line in the enhanced image. Since the noise in the enhanced image has been effectively removed, the disconnection and missing phenomenon in the optimized processing edge line is caused by burrs. Therefore, the optimized processing edge line can be compared with the preset processing trajectory, and the missing part of the optimized processing edge line relative to the preset processing trajectory is taken as the burr location in the enhanced image.

An embodiment of the present invention provides a burr detection system during titanium metal processing, the system comprising a memory, a processor and a computer program, wherein the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and when the computer program runs in the processor, the method described in steps S1 to S5 can be implemented.

In summary, the embodiment of the present invention first obtains a grayscale image of the titanium metal surface to be processed, performs edge detection on the grayscale image, and obtains processing edge lines, suspected burr edge lines, and texture edge lines in the grayscale image; obtains the grayscale variation factor of each pixel point according to the distribution of the grayscale values of each pixel point in a preset neighborhood centered on each pixel point; takes any suspected burr edge line as the target edge line, and takes the two processing edge lines adjacent to the target edge line as the reference processing edge lines of the target edge line; obtains the burr possibility of the target edge line according to the grayscale variation factor of each pixel point on the target edge line and the difference in the grayscale variation factor of the pixel point between the target edge line and the reference processing edge line; obtains the target edge line The target edge line is compared with the centroid of each texture edge line, and the distance between the centroid of the target edge line and each texture edge line is used as the distance parameter of each texture edge line; a preset number of texture edge lines with the smallest distance parameters are selected as the reference texture edge lines of the target edge line; the burr fuzzy factor of the target edge line is obtained according to the gradient value of each pixel point on each reference texture edge line and the distance parameter of each reference texture edge line; the noise influence factor of the target edge line is obtained according to the difference between the burr possibility of the target edge line and the burr fuzzy factor; based on the noise influence factor of each suspected burr edge line, the grayscale image is filtered to obtain an enhanced image; edge detection is performed on the enhanced image to obtain the burr part in the enhanced image.

It should be noted that the sequence 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 drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims (9)

1. A method for detecting burrs in a titanium metal machining process, the method comprising:

acquiring a gray level image of the surface of the titanium metal to be processed, and performing edge detection on the gray level image to obtain a processed edge line, a suspected burr edge line and a texture edge line in the gray level image;

Obtaining a gray scale change factor of each pixel point according to the distribution of gray scale values of each pixel point in a preset neighborhood taking each pixel point as a center; taking any suspected burr edge line as a target edge line, taking two processing edge lines adjacent to the target edge line as reference processing edge lines of the target edge line, and obtaining the burr possibility of the target edge line according to the gray scale change factors of all pixel points on the target edge line and the difference of the gray scale change factors of the pixel points between the target edge line and the reference processing edge line;

Acquiring the mass centers of the target edge line and each texture edge line, and taking the distance between the target edge line and each texture edge line as the distance parameter of each texture edge line; selecting a preset number of texture edge lines with minimum distance parameters as reference texture edge lines of target edge lines; obtaining a burr blurring factor of the target edge line according to the gradient value of each pixel point on each reference texture edge line and the distance parameter of each reference texture edge line;

obtaining a noise influence factor of the target edge line according to the burr possibility of the target edge line and the difference of the burr blurring factors; filtering the gray level image based on the noise influence factors of each suspected burr edge line to obtain an enhanced image;

performing edge detection on the enhanced image to obtain a burr part in the enhanced image;

The obtaining an enhanced image includes:

And taking the noise influence factor of each suspected burr edge line as a scale factor used by a wiener filtering algorithm, and filtering the area where the minimum circumcircle of each suspected burr edge line in the gray level image is located based on the wiener filtering algorithm to obtain an enhanced image.

2. The method for detecting burrs during titanium metal processing as defined in claim 1, wherein said obtaining processing edge lines, suspected burrs edge lines and texture edge lines in gray scale images includes:

Performing edge detection on the gray level image based on an edge detection algorithm to obtain a processing edge line and a texture edge line in the gray level image;

And comparing the processing edge line with a preset processing track, and taking the part of the processing edge line, which is missing relative to the preset processing track, as a suspected burr edge line.

3. The method for detecting burrs during titanium metal processing of claim 1, wherein said obtaining a gray scale variation factor for each pixel point comprises:

Taking any pixel point in the gray level image as a target pixel point, and taking any pixel point in a preset adjacent area taking the target pixel point as a center as a pixel point to be measured; if the gray value of the pixel to be detected is greater than or equal to the gray value of all the pixels in a preset window taking the pixel to be detected as the center, marking the pixel to be detected as a maximum pixel, and acquiring all the maximum pixels in a preset adjacent area;

taking any one maximum pixel point as a target maximum pixel point, and taking other maximum pixel points closest to the target maximum pixel point as reference maximum pixel points of the target maximum pixel point;

taking the average value of the gray values of the target maximum pixel point and the corresponding reference maximum pixel point as a first gray parameter of the target maximum pixel point; analyzing the integral level of gray values of all pixel points on the connecting line between the target maximum pixel point and the corresponding reference maximum pixel point to obtain a second gray parameter of the target maximum pixel point; acquiring an initial gray scale variation parameter of a target maximum pixel point according to the first gray scale parameter and the second gray scale parameter, wherein the initial gray scale variation parameter is positively correlated with the first gray scale parameter, and the initial gray scale variation parameter is negatively correlated with the second gray scale parameter;

Performing negative correlation mapping on the distance between the target maximum pixel point and the corresponding reference maximum pixel point to obtain a first distance weight of the target maximum pixel point, and weighting the initial gray level change parameter by using the first distance weight of the target maximum pixel point to obtain a real gray level change parameter of the target maximum pixel point;

analyzing the overall level of the real gray scale change parameters of all the maximum pixel points in the preset neighborhood, and carrying out normalization processing to obtain the gray scale change factor of the target pixel point.

4. The method for detecting burrs during titanium metal processing of claim 1, wherein said obtaining the probability of burrs of a target edge line comprises:

analyzing the overall level of the gray scale change factors of all pixel points on the target edge line or each reference processing edge line to obtain the overall gray scale change factors of the target edge line or each reference processing edge line;

According to the difference of the integral gray scale change factors between the target edge line and each reference processing edge line, gray scale change difference values of each reference processing edge line are obtained, and the maximum value of the gray scale change difference values of all the reference processing edge lines is used as a gray scale change characteristic value of the target edge line;

and obtaining the burr possibility of the target edge line according to the integral gray scale change factor and the gray scale change characteristic value of the target edge line, wherein the burr possibility is positively correlated with the gray scale change characteristic value of the target edge line, the burr possibility is negatively correlated with the integral gray scale change factor of the target edge line, and the burr possibility is a numerical value after normalization processing.

5. The method for detecting burrs during titanium metal processing of claim 1, wherein said obtaining a burr blur factor of a target edge line comprises:

Analyzing the overall level of the gradient values of all pixel points on each texture edge line to obtain a gradient distribution value of each texture edge line, and analyzing the overall level of the gradient distribution value of all texture edge lines to obtain an overall gradient distribution parameter;

Obtaining an initial gradient characteristic value of each reference texture edge line according to the overall gradient distribution parameters and the gradient distribution value of each reference texture edge line; the initial gradient characteristic value is positively correlated with the overall gradient distribution parameter, and the initial gradient characteristic value is negatively correlated with the gradient distribution value of each reference texture edge line;

performing negative correlation mapping on the distance parameters of each reference texture edge line to obtain a second distance weight of each reference texture edge line; weighting the initial gradient characteristic value of each reference texture edge line by using the second distance weight of each reference texture edge line to obtain a real gradient characteristic value of each reference texture edge line;

analyzing the overall level of the real gradient characteristic values of all the reference texture edge lines, and carrying out normalization processing to obtain the burr fuzzy factor of the target edge line.

6. The method for detecting burrs during titanium metal processing of claim 1, wherein said obtaining noise influencing factors of a target edge line comprises:

And carrying out normalization processing on the difference of the burr possibility and the burr blurring factor according to the target edge line to obtain a noise influence factor of the target edge line.

7. The method for detecting burrs in a titanium metal working process of claim 1, wherein said performing edge detection on the enhanced image to obtain burrs in the enhanced image comprises:

performing edge detection on the enhanced image to obtain an optimized processing edge line in the enhanced image;

Comparing the optimized machining edge line with a preset machining track, and taking the missing part of the optimized machining edge line relative to the preset machining track as a burr part in the enhanced image.

8. The method for detecting burrs in a titanium metal machining process according to claim 1, wherein the preset number is an integer ranging from 10 to 30.

9. A system for detecting burrs during titanium metal machining, said system comprising a memory, a processor and a computer program stored in said memory and executable on said processor, characterized in that said processor implements the steps of the method according to any one of claims 1-8 when executing said computer program.