Unbiased detection method and detector based on laser stripe centerline iteration method

Abstract

The present invention provides an unbiased detection method and detector of the center line of laser stripes by iterative method, which extracts the center of laser stripes multiple times by iterative extraction. During the extraction of laser center line, when the iterative criterion condition is met, the iterative operation stops, otherwise the laser stripe image is repeatedly iteratively processed; when the laser center line accuracy criterion condition is met, the unbiased detector of the iterative method of laser stripe center line stops operating, otherwise a new image is obtained according to the principle of removing image noise and impurity information and iterative operation is performed again; the center line accuracy criterion set by the method can explain the rationality of the laser stripe line extracted by the algorithm from the perspective of mathematical formula; after the iterative criterion is set by using the iterative idea, the center line of the laser stripe of the weld is extracted by using the grayscale centroid, which is particularly suitable for the occasion of structured light vision welding trajectory recognition in the welding field, and is used for the structured light vision welding trajectory recognition of butt joints, thin plates and medium and thick plates.

Description

Technical Field

[0001] The present invention belongs to the technical field of structured light visual image processing, and relates to an unbiased detection and extraction technology for the center line of curved structure laser stripes, and specifically to an unbiased detection method and detector for the center line of laser stripes by iterative method. Background Art

[0002] As the eye of industry, structured light vision is widely used in the equipment manufacturing industry. It can accurately identify the spatial position of the target, allowing the motion control mechanism to perceive the three-dimensional coordinates of the target object and complete some intelligent production operations. Structured light vision mainly includes laser emitters and industrial cameras. The laser emitter generates a line laser and projects it onto the surface of the target object. The industrial camera responsible for image acquisition obtains the laser stripe image. The laser stripe image contains some feature information of the target object, such as feature points or feature lines, and then uses image processing to obtain this feature information. Image processing is mainly divided into image preprocessing, laser stripe centerline extraction, and feature information extraction.

[0003] As a key step in image processing, the laser stripe centerline extraction methods include the Steger method and the center point method. However, if these methods are to extract the accurate laser stripe centerline, they are heavily dependent on the image preprocessing technology to remove image noise and image impurity information. The image preprocessing process includes image smoothing, threshold segmentation, morphological processing, etc. The traditional laser stripe centerline extraction method has two main drawbacks:

[0004] First, it is highly sensitive to image noise and image impurity information, and the accuracy of laser centerline extraction is heavily dependent on the purity of laser stripe information;

[0005] The second is that the algorithm extracts the center of the laser stripe only once, and does not determine the accuracy of the centerline extraction.

[0006] Therefore, in order to improve the stability and accuracy of the laser stripe centerline extraction algorithm, the traditional laser stripe centerline extraction algorithm must be improved to obtain an unbiased detection method for the laser stripe centerline with high robustness that can be used in actual situations. Summary of the invention

[0007] The purpose of the present invention is to provide an unbiased detection method for the center line of laser stripes by iteration. After setting the iteration criterion by using the iteration idea, the center line of the laser stripes of the weld is extracted by using the grayscale centroid. The method is particularly suitable for the structured light vision welding trajectory recognition occasions in the welding field, and is used for the structured light vision welding trajectory recognition of butt joints, thin plates and medium and thick plates.

[0008] The purpose of the present invention is achieved through the following technical solutions:

[0009] A laser stripe center line iteration method unbiased detection method, comprising the following steps:

[0010] S1, original laser stripe image img i Extract the center line and obtain the center curve function f i (x);

[0011] S2. Set the principle of removing image noise and impurity information to obtain a new laser stripe image img with effective information close to the theoretical laser stripe. i+1 , extract the center curve function f i+1 (x);

[0012] S3. Set the iteration criteria to determine the central curve function f i+1 (x) and f i (x) Whether the theoretical position relationship between them meets the iteration criterion;

[0013] When the iteration criterion is not met, a new laser stripe image img is obtained again according to the principle of removing image noise and impurity information. i+2 , and extract the central curve function f i+2 (x); determine the central curve function f i+2 (x) and f i+1 (x) whether the theoretical position between them meets the iteration criterion;

[0014] If it does not meet the criteria, the iteration operation is repeated until the current image center curve function and the previous iteration image center curve function meet the iteration criteria, and the iteration is terminated.

[0015] As a more optimal technical solution of the present invention, the principle of removing image noise and impurity information in step S2 is:

[0016]

[0017] h i =μ i h0 i=0,1,2……

[0018] h0=δ*{max[f0(x)]-min[f0(x)]}

[0019] Wherein max[f0(x)] and min[f0(x)] are the maximum and minimum values ​​of the central curve function f0(x) of the laser stripe image img0, respectively; δ is the correction factor, and δ>1; μ is the iteration factor, and μ<1.

[0020] As a more optimal technical solution of the present invention, the iterative criterion described in step S3 is set as:

[0021] D{abs[f i+1 (x)-fi (x)]}≤gap

[0022] Where gap is the iterative laser stripe center line variance threshold.

[0023] As a more optimal technical solution of the present invention, the centerline function extraction described in step S2 is as follows:

[0024] According to the grayscale centroid algorithm, the grayscale value of each pixel position in the "region" is regarded as the "quality" of the point. The grayscale value of the center of the "region" is equal to the weighted value of each point in the "region". The calculation formula of the center of the "region" is:

[0025]

[0026] Where: f(x, y) is the grayscale value of the pixel at coordinate (x, y), Ω is the set of target areas, is the coordinate of the center of the region, and the gray value is the energy center of the extracted region;

[0027] The image column pixels are designated as the light spot "area", and the laser center line with functional characteristics is obtained through grayscale centroid loop processing.

[0028] As a more optimal technical solution of the present invention, the central curve accuracy criterion described in step S2 is:

[0029] max[f i+1 (x)]-max[f i (x)]≤Z

[0030] Where z is the iterative laser stripe center line position deviation threshold, and z < 1, that is, at f i (x) and f i+1 The position deviation of (x) is within sub-pixel level.

[0031] Another object of the present invention is to provide a laser stripe centerline iteration method unbiased detector, comprising four modules; specifically:

[0032] The first module is a laser stripe image processing module; the purpose of setting the principle of removing image noise and impurity information is to obtain a new laser stripe image with effective information close to the theoretical laser stripe in the iterative processing process;

[0033] The principle of removing image noise and impurity information is:

[0034]

[0035] h i =μ i h0 i=0,1,2……

[0036] h0=δ*{max[f0(x)]-min[f0(x)]}

[0037] Where: max[f0(x)] and min[f0(x)] are the maximum and minimum values ​​of the central curve function f0(x) of the laser stripe image img0, respectively; δ is the correction factor, and δ>1; μ is the iteration factor, and μ<1;

[0038] The second module is a centerline extraction module; the centerline extraction module uses a grayscale centroid algorithm to treat the grayscale value of each pixel position in the "region" as the "quality" of the point, and the grayscale value of the center position of the "region" is equal to the weighted value of each point in the "region". The calculation formula of the center of the "region" is:

[0039]

[0040] Where: f(x, y) is the grayscale value of the pixel at coordinate (x, y), Ω is the set of target areas, is the coordinate of the center of the region, and the gray value is the energy center of the extracted region;

[0041] Specify the image column pixels as the light spot "area", and obtain the laser center line with function characteristics through grayscale centroid loop processing;

[0042] The third module is a laser centerline accuracy criterion module; the laser centerline accuracy criterion is:

[0043] max[f i+1 (x)]-max[f i (x)]≤Z

[0044] Where z is the iterative laser stripe center line position deviation threshold, and z < 1, that is, at f i (x) and f i+1 The position deviation of (x) is within the sub-pixel level;

[0045] The fourth module is an iterative criterion module, and the iterative criterion is set as:

[0046] D{abs[f i+1 (x)-f i (x)]}≤gap

[0047] Where: gap is the iterative laser stripe center line variance threshold.

[0048] The beneficial effects are:

[0049] The detector provided by the present invention has a strong anti-interference ability to the laser stripe image noise or impurity information, and the accuracy of the laser center line extraction is no longer heavily dependent on the purity of the laser stripe information.

[0050] The algorithm provided by the present invention performs multiple iterative extractions on the center of the laser stripe. During the extraction of the laser center line, the iterative operation stops when the iterative criterion condition is met, otherwise the laser stripe image is repeatedly iterated; when the laser center line accuracy criterion condition is met, the laser stripe center line iterative method unbiased detector stops operating, otherwise a new image is obtained based on the principle of image noise and impurity information removal and iterative operation is performed again; the center line accuracy criterion set by the algorithm can explain the rationality of the laser stripe line extracted by the algorithm from the perspective of mathematical formulas. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The drawings described herein are used to provide further understanding of the present invention and constitute a part of this application. The illustrative examples of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0052] Figure 1 This is a schematic diagram of an unbiased detector using the laser stripe centerline iteration method of the present invention;

[0053] Figure 2 It is a flow chart of the operation of the unbiased detector of the laser stripe centerline iteration method of the present invention;

[0054] Figure 3 It is a grayscale center of gravity centerline extraction operation flow chart of the present invention;

[0055] Figure 4 The centerline extraction effect of the laser stripe centerline iteration method of the present invention on the strong noise image without biased detector;

[0056] Figure 5 This is the effect of extracting the center line of a weak noise image using the unbiased detector of the laser stripe center line iteration method of the present invention. DETAILED DESCRIPTION

[0057] The details of the present invention and its specific implementation methods are further described below in conjunction with the accompanying drawings.

[0058] See also Figure 1 As shown, the unbiased detector of the laser stripe centerline iteration method of the present invention is extremely suitable for the structured light vision welding trajectory recognition occasions in the welding field, and is suitable for the field of structured light vision welding trajectory recognition of butt joints, thin plates, and medium and thick plates. It uses the iterative idea, sets the iterative criterion, and then uses the grayscale centroid to extract the weld laser stripe centerline.

[0059] The laser stripe center line iteration method unbiased detector described in the present invention and the laser stripe center line iteration method unbiased detection method are as follows:

[0060] For the laser stripe image img i Extract the center line and obtain the center curve function f i(x);

[0061] Set the principle of removing image noise and impurity information to obtain a new laser stripe image img with effective information close to the theoretical laser stripe i+1 , extract the central curve function f of the new laser stripe image i+1 (x);

[0062] Set the iteration criterion to determine the central curve function f i+1 (x) and f i (x) between the theoretical position relationship, when the center curve function f i+1 When the theoretical position relationship between (x) and fi(x) does not meet the iteration criterion, a new laser stripe image img is obtained again according to the principle of image noise and impurity information removal. i+2 , and extract the central curve function f i+2 (x);

[0063] Determine the central curve function f i+2 (x) and f i+1 If the theoretical position between the image and the previous iterative image meets the iterative criterion, the above iterative operation will be repeated until the center line of the current image meets the center line of the previous iterative image and the iteration is terminated.

[0064] The laser stripe center line iterative method unbiased detector described in the present invention includes an iterative processing theory, a theory capable of removing image noise and impurity information and extracting the laser center line unbiasedly, and a theory of laser center line accuracy criterion.

[0065] See also Figure 1-3 The laser stripe centerline iteration method unbiased detection method shown comprises the following steps:

[0066] That is, establish an iterative theory, extract the grayscale centerline, and establish the accuracy criterion of the laser centerline:

[0067] S1. Establish iterative theory:

[0068] The iterative theory mainly includes the determination of exchange and substitution physical quantities, the setting of image noise and impurity information removal principles, and the setting of iterative criteria.

[0069] S1.1. Determine the exchange substitution physical quantity;

[0070] The laser stripe image is used as an exchange substitute for physical quantities. During the process of extracting the center line by the unbiased detector using the laser stripe center line iteration method, the laser stripe image is repeatedly processed, and the new laser stripe image continuously approaches the theoretical laser center position.

[0071] S1.2, setting the principle of removing image noise and impurity information;

[0072] The purpose of setting the principle of removing image noise and impurity information is to obtain a new laser stripe image with new and effective information close to the theoretical laser stripe in the iterative processing process, so as to prepare for the next iterative cycle processing.

[0073] The principle of removing image noise and impurity information is:

[0074]

[0075] h i =μ i h0 i=0,1,2……

[0076] h0=δ*{max[f0(x)]-min[f0(x)]}

[0077] Wherein: max[f0(x)] and min[f0(x)] are the maximum and minimum values ​​of the central curve function f0(x) of the laser stripe image img0, respectively; δ is the correction factor, and δ>1; μ is the iteration factor, and μ<1.

[0078] S1.3, set the iteration criteria;

[0079] For a fixed laser stripe image, there is only one laser stripe center line at its theoretical position. Therefore, according to the principle of removing image noise and impurity information, the i-th laser stripe image img is obtained. i The center curve function f i (x) continuously approaches the i+1th laser stripe image img i+1 The center curve function f i+1 (x), that is, f i (x) and f i+1 There is a functional relationship between (x), and the variance of the absolute values ​​of their deviations is theoretically zero.

[0080] Therefore, the iteration criterion is set as:

[0081] D{abs[f i+1 (x)-f i (x)]}≤gap

[0082] Where: gap is the iterative laser stripe center line variance threshold.

[0083] S2, extract the grayscale centerline;

[0084] In the process of establishing the iterative theory, it is necessary to repeatedly process the laser stripe image to remove noise and impurity information and extract a new laser center line function. According to the grayscale centroid algorithm, the grayscale value of each pixel position in the "region" is regarded as the "quality" of the point, and the grayscale value of the center position of the "region" is equal to the weighted value of each point in the "region". The calculation formula of the center of the "region" is:

[0085]

[0086] Where: f(x, y) is the grayscale value of the pixel at coordinate (x, y), Ω is the set of target areas, are the coordinates of the center of the region, and the grayscale value is the energy center of the extracted region.

[0087] At the same time, in order to ensure that the extracted laser center line independent variable and dependent variable present a one-to-one functional relationship, the present invention specifies the image column pixels as the light spot "area", and obtains the laser center line with functional characteristics through grayscale centroid loop processing. Figure 3 .

[0088] S3. Establish the accuracy criterion of laser centerline;

[0089] According to f i (x) and f i+1 (x) The variance of the absolute value of the deviation between the two is not a necessary and sufficient condition for extracting the accurate center line of the laser stripe. Assuming f i (x) and f i+1 (x) is parallel, and the variance of the absolute value of the deviation is still zero. The accuracy criterion of the laser stripe center line set by the present invention is:

[0090] max[f i+1 (x)]-max[f i (x)]≤Z

[0091] Where: z is the iterative laser stripe center line position deviation threshold, and z < 1, that is, at f i (x) and f i+1 The position deviation of (x) is within sub-pixel level.

[0092] In the process of extracting the laser center line by the unbiased detector of the laser stripe center line iteration method of the present invention, the iterative operation stops when the iteration criterion condition is met, otherwise the laser stripe image is repeatedly iterated; when the laser center line accuracy criterion condition is met, the unbiased detector of the laser stripe center line iteration method stops operating, otherwise a new image is obtained according to the principle of removing image noise and impurity information and the iterative operation is performed again.

[0093] See also Figure 4The laser stripe centerline iterative method unbiased detector extracts the centerline of a strong noise image. The black line in the white stripe is the laser stripe centerline, which undergoes 6 iterative operations. It can be seen that the laser stripe centerline iterative method unbiased detector can continuously iteratively remove the noise and impurity information of the image, and the obtained laser stripe centerline is constantly approaching the theoretical position. The algorithm has a strong anti-interference ability against the noise and impurity information of the image.

[0094] See also Figure 5 The effect of extracting the center line of a weak noise image by an unbiased detector using the iterative method for the center line of the laser stripe is shown in the figure. The black line in the white stripe is the center line of the laser stripe. After 8 iterative operations, the unbiased detector using the iterative method for the center line of the laser stripe can accurately extract the center line of not only the strong noise image, but also the weak sound image. It has a certain adaptability.

[0095] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A laser stripe centerline iteration method unbiased detection method, characterized in that: The steps include: S1, original laser stripe image img i Extract the center line and obtain the center curve function f i (x); S2. Set the principle of removing image noise and impurity information to obtain a new laser stripe image img with effective information close to the theoretical laser stripe. i+1 , extract the center curve function f i+1 (x); The principle of removing image noise and impurity information is: h i =μ i h0i=0,1,2…… h0=δ*{max[f0(x)]-min[f0(x)]} Where max[f0(x)] and min[f0(x)] are the maximum and minimum values ​​of the central curve function f0() of the laser stripe image img0, respectively; δ is the correction factor, and δ>1; μ is the iteration factor, and μ<1; The centerline function is extracted as follows: According to the grayscale centroid algorithm, the grayscale value of each pixel position in the "region" is regarded as the "quality" of the pixel position, and the grayscale value of the center position of the "region" is equal to the weighted value of each point in the "region". The calculation formula of the center of the "region" is: Where: f(x, y) is the grayscale value of the pixel at coordinate (x, y), Ω is the set of target areas, is the coordinate of the center of the region, and the gray value is the energy center of the extracted region; S3. Set the iteration criteria to determine the central curve function f i+1 (x) and f i (x) Whether the theoretical position relationship between them meets the iteration criterion; The iterative criterion is set as: D{abs[f i+1 (x)-f i (x)]}≤gap Where gap is the iterative laser stripe centerline variance threshold; When the iteration criterion is not met, a new laser stripe image img is obtained again according to the principle of removing image noise and impurity information. i+2 , and extract the central curve function f i+2 (x); determine the central curve function f i+2 (x) and f i+1 (x) whether the theoretical position between them meets the iteration criterion; If it does not meet the criteria, the iteration operation is repeated until the current image center curve function and the previous iteration image center curve function meet the iteration criteria, and the iteration is terminated.

2. The unbiased detection method of the laser stripe centerline iteration method as claimed in claim 1, characterized in that: The central curve accuracy criterion described in step S2 is: max[f i+1 (x)]-max[f i (x)]≤Z Where Z is the iterative laser stripe center line position deviation threshold, and Z<1, that is, at f i (x) and f i+1 The position deviation of (x) is within sub-pixel level.

3. A laser stripe centerline iteration method unbiased detector, characterized in that; It includes four modules, namely: The first module is a laser stripe image processing module; the purpose of setting the principle of removing image noise and impurity information is to obtain a new laser stripe image with effective information close to the theoretical laser stripe in the iterative processing process; The principle of removing image noise and impurity information is: h i =μ i h0 i=0,1,2…… h0=δ*{max[f0(x)]-min[f0(x)]} Where: max[f0(x)] and min[f0(x)] are the maximum and minimum values ​​of the central curve function f0(x) of the laser stripe image img0 respectively; δ is the correction factor, and δ>1; μ is the iteration factor, and μ<1; The second module is a centerline extraction module; the centerline extraction module uses a grayscale centroid algorithm to take the grayscale value of each pixel position in the "region" as the "quality" of the pixel position, and the grayscale value of the center position of the "region" is equal to the weighted value of each point in the "region". The calculation formula of the center of the "region" is: Where: f(x, y) is the grayscale value of the pixel at coordinate (x, y), Ω is the set of target areas, is the coordinate of the center of the region, and the gray value is the energy center of the extracted region; Specify the image column pixels as the light spot "area", and obtain the laser center line with function characteristics through grayscale centroid loop processing; The third module is a laser centerline accuracy criterion module; the laser centerline accuracy criterion is: max[f i+1 (x)]-max[f i (x)]≤Z Where Z is the iterative laser stripe center line position deviation threshold, and Z<1, that is, at f i (x) and f i+1 The position deviation of (x) is within the sub-pixel level; The fourth module is an iterative criterion module, and the iterative criterion is set as: D{abs[f i+1 (x)-f i (x)]}≤gap Where: gap is the iterative laser stripe center line variance threshold.

Images