A method for detecting internal defects of mud pumps based on optical analysis

Abstract

The present invention relates to the field of optical detection technology, and specifically to a method for detecting internal defects of a mud pump based on optical analysis, which optically analyzes the acquired internal radiographic images / image data of the mud pump, combines image processing technology, studies the pixel characteristics in the radiographic images, and thereby obtains appropriate smoothing filter weight coefficients to smooth the radiographic images, thereby identifying the internal defect areas of the mud pump to be detected. Through the above method, the noise interference in the original radiographic images can be eliminated to compensate for the loss effect of the process of the radiation transmitting the mud particles, thereby improving the contrast and clarity of the image, and finally obtaining a more reliable and accurate internal structural defect area of ​​the mud pump.

Description

Technical Field

[0001] The invention relates to the technical field of optical detection, and in particular to a method for detecting internal defects of a mud pump based on optical analysis. Background Art

[0002] Mud pumps play a key role in industrial production, used to transport mud and other fluids, so their stable operation is very important for the normal progress of the production process. During the operation of the mud pump, a variety of internal problems may occur, such as impeller damage, seal failure, pump casing wear, etc. Optical analysis technology, through the introduction of optical sensors, cameras and other equipment, combined with image processing, can achieve internal defect detection of the mud pump, in advance and prevent situations that may cause mud pump failure.

[0003] In the traditional optical analysis-based internal defect detection of mud pumps, the internal structure of the mud pump has complex environmental conditions during use or after a period of use. For example, there is a large amount of mud residue on the surface of the internal structure of the mud pump, which interferes with the optical imaging technology, resulting in low quality images of the internal structure of the mud pump, thereby affecting the accuracy and reliability of the system's internal defect detection. Summary of the invention

[0004] In order to solve the technical problem that the internal environmental conditions of the mud pump interfere with the optical imaging technology, resulting in low quality of the internal structure image of the mud pump, which affects the internal defect detection, the purpose of the present invention is to provide a mud pump internal defect detection method based on optical analysis. The technical solution adopted is as follows:

[0005] A method for detecting internal defects of a mud pump based on optical analysis, the method comprising: obtaining an internal radiographic image of a mud pump to be detected; dividing the radiographic image in the radiographic image into multiple local areas, analyzing the images of adjacent frames of each local area to obtain a transmission loss probability factor of the local area; obtaining multiple local abnormal areas from the multiple local areas according to the transmission loss probability factor, performing image analysis on the multiple local abnormal areas to obtain an abnormal area collapse factor; obtaining multiple local collapsed areas from the multiple local abnormal areas according to the abnormal area collapse factor, performing image analysis on the multiple local collapsed areas to obtain a transmission loss degree factor; setting a smoothing filter weight coefficient according to the transmission loss probability factor and the transmission loss degree factor to smooth the radiographic image, and identifying the internal defect area of ​​the mud pump to be detected according to the smoothed radiographic image.

[0006] The scheme of the present invention divides the radiographic image in the internal radiographic image of the mud pump to be inspected to form several local areas, and then uses the local area as a comparison unit to perform specific analysis and comparison on the image pixels in the comparison unit, thereby obtaining the local abnormal area and the local collapsed area, and then combining the transmission loss probability of the image of the local abnormal area and the transmission loss degree of the image of the local collapsed area to obtain the smoothing filter weight coefficient. Then, the internal radiographic image of the mud pump to be inspected is smoothed by the smoothing filter weight coefficient to eliminate the noise interference in the original radiographic image, and then the defective area is identified, and finally a more reliable and accurate internal structural defective area of ​​the mud pump is obtained.

[0007] In view of the scenario in which the quality and clarity of the internal image of the mud pump obtained in the traditional optical analysis method is reduced, the present invention analyzes the transmission loss generated by the ray transmission process, sets the smoothing filter weight coefficient as the scale factor of the Wiener filter, and smoothes the internal flaw detection image of the mud pump, which can control the balance between retaining image details and removing noise, and also makes the filtering efficiency higher and the effect better.

[0008] Furthermore, before segmenting the radiographic image, the radiographic image in the radiographic image is preprocessed, and the preprocessing includes denoising and graying.

[0009] Furthermore, the radiographic image in the radiographic image is segmented into a plurality of local regions, specifically comprising: using a superpixel region segmentation algorithm to segment the preprocessed radiographic image into a plurality of superpixel regions, wherein the superpixel regions are used as the local regions.

[0010] Furthermore, the transmission loss probability factor of each local area is obtained by analyzing the images of adjacent frames of the local area, specifically including: analyzing the same local area through the radiation images of two adjacent frames, and obtaining the fluidity factor and steady-state factor of the local area according to the differences in pixel value changes, the number of corner points, and the changes in corner point coordinates in the current local area; and obtaining the norm of the ratio of the fluidity factor to the steady-state factor to obtain the transmission loss probability factor of the local area.

[0011] Furthermore, the fluidity factor is the mean square error of the set of pixel values ​​of the local area of ​​two adjacent frames; the steady-state factor is the absolute value of the difference in stable structure factors of the local area of ​​two adjacent frames, and the stable structure factor is the ratio of the number of corner points whose pixel coordinates of two adjacent frames overlap to the number of all corner points of the previous frame in the two adjacent frames.

[0012] Furthermore, image analysis is performed on the multiple local abnormal areas to obtain an abnormal area collapse factor, specifically including: detecting the edge of the abnormal area, analyzing the quantitative relationship between the number of pixels inside the abnormal area and the number of edge pixels, and obtaining a texture feature factor; obtaining a brightness highlighting factor based on the difference between the average pixel value of the abnormal area and the average pixel value of the surrounding area; multiplying the texture feature factor and the brightness highlighting factor to obtain the abnormal area collapse factor.

[0013] In the step, the texture feature factor is the difference between 1 and the ratio of the number of pixels, wherein the ratio of the number of pixels refers to the ratio of the number of edge pixels to the number of internal pixels; the brightness highlighting factor is the average of the brightness difference factors between the abnormal area and all surrounding areas; wherein the brightness difference factor is the absolute value of the difference between the average pixel value of the abnormal area and the average pixel value of its surrounding areas.

[0014] Furthermore, a plurality of local collapsed regions are obtained from the plurality of local abnormal regions, specifically, the local abnormal regions whose collapse factors of the abnormal regions are greater than a preset threshold are marked as local collapsed regions.

[0015] Furthermore, image analysis is performed on the multiple local collapsed areas to obtain a transmission loss degree factor, specifically including: the transmission loss degree factor is the product of the collapse factor of the abnormal area and the density of the collapsed area; wherein the density of the collapsed area is: the average of the distance differences between the local collapsed area and all the local areas closest to it.

[0016] Furthermore, the product of the transmission loss probability factor and the transmission loss degree factor is used as a smoothing filter weight coefficient, and the smoothing filter weight coefficient is used as a scale factor of Wiener filtering, and the radiation image is smoothed by Wiener filtering.

[0017] The present invention has the following beneficial effects:

[0018] During the use of the mud pump or after a period of use, the complex environmental conditions of its internal structure affect the image quality obtained by the optical imaging technology, resulting in the inability to accurately identify the defective area. The scheme of the present invention is based on the analysis of the radiographic image based on whether there is mud residue attached to the surface of the defective area. When there is mud residue attached to the surface of the defective area, the rays will have a certain loss during transmission, and the transmission loss probability factor generated therein is analyzed. At the same time, when there are a large number of large mud particles inside the mud residue, it will cause more collapse areas when the rays are transmitted. Therefore, the collapse density of the local area is analyzed to obtain the ray transmission loss degree factor of the area. The transmission loss probability factor and the ray transmission loss degree factor are combined to obtain the line transmission loss factor, which is used as the smoothing filter weight coefficient, and an adaptive smoothing filter is set to perform smoothing processing to eliminate the noise interference in the original radiographic image, thereby identifying the defective area and obtaining a more reliable and accurate internal structural defect area of ​​the mud pump.

[0019] The scheme of the present invention fully considers the interference factors such as mud particles existing on the surface of the internal structure of the mud pump, designs the smoothing filter weight coefficient, and uses an adaptive smoothing filter to filter the original image of the internal structure of the mud pump to compensate for the loss effect of the ray transmission process of the mud particles, thereby improving the contrast and clarity of the image. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] 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.

[0021] Figure 1 A flow chart of a method for detecting internal defects of a mud pump based on optical analysis is provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION

[0022] 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 the specific implementation method, structure, features and effects of a mud pump internal defect detection method based on optical analysis proposed by the present invention 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.

[0023] 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.

[0024] The following is a detailed description of a method for detecting internal defects of a mud pump based on optical analysis provided by the present invention in conjunction with the accompanying drawings. The present invention proposes a method for detecting internal defects of a mud pump based on optical analysis in view of the fact that the quality clarity of the image obtained from the inside of the mud pump is reduced in the traditional optical analysis method. Figure 1 , the method specifically includes:

[0025] Obtain an X-ray image of the interior of the mud pump to be inspected. The rays in the present application scheme refer to a type of rays that can be used for optical analysis of the internal defect detection method of the mud pump. Such rays have penetrability (can penetrate the metal shell and internal materials of the mud pump to obtain information about the internal structure) and absorptivity (different materials have different absorptivity for different rays. This can be used to detect areas of different materials or densities inside the mud pump to find possible defects or foreign matter). For the sake of convenience, the embodiments of the present application take X-rays as an example for explanation.

[0026] X-ray flaw detection technology is a non-destructive detection method commonly used to detect internal defects of materials. X-ray flaw detection equipment usually includes an X-ray generator, a detection detector, an image acquisition system and an image processing system. Based on a specific practical scenario, this embodiment installs an X-ray generator on the surface of the mud pump to be detected, and uses the penetrating properties of X-rays to obtain information about the internal structure of the mud pump by irradiating the mud pump to be detected and measuring the intensity of the projected X-rays. When the X-rays pass through the object, the density and thickness of different materials will affect the degree of transmission of the X-rays, thereby producing different degrees of attenuation, forming an X-ray image of the internal structure of the mud pump. This embodiment optically analyzes the acquired internal X-ray image / image data of the mud pump, combines image processing technology, studies the pixel characteristics in the X-ray image, and thus obtains an appropriate smoothing filter weight coefficient to smooth the X-ray image, thereby identifying the internal defect area of ​​the mud pump to be detected.

[0027] The obtained internal radiographic image of the mud pump to be inspected, namely the X-ray image, is preprocessed, including but not limited to operations such as denoising and graying, so as to facilitate subsequent identification of defective areas of the internal structure of the mud pump.

[0028] When using X-ray flaw detection technology to obtain images of the internal structure of a mud pump, particles in the mud (such as mud particles) will affect the transmission of X-rays. When X-rays pass through mud particles, they absorb and scatter the X-rays, which reduces the intensity of X-rays that are finally transmitted to the detector, resulting in a decrease in the contrast between the defect area and the surrounding mud particle area in the X-ray flaw detection image, blurring the edges of the defect area in the image, and thus affecting the detection and positioning of internal structural defects in the mud pump.

[0029] In this case, an adaptive smoothing filter can be designed and used to filter the original image of the internal structure of the mud pump according to interference factors such as mud particles on the surface of the internal structure of the mud pump to compensate for the loss effect of the X-ray transmission process of the mud particles, thereby improving the contrast and clarity of the image.

[0030] The ray image in the ray image is divided into multiple local areas, and the preprocessed ray image is segmented into multiple superpixel areas using the superpixel area segmentation algorithm. The superpixel area is used as the local area, and the image of the adjacent frames of each local area is analyzed to obtain the transmission loss probability factor of the local area. The same local area is analyzed by the ray images of two adjacent frames, and the liquidity factor and steady-state factor of the local area are obtained according to the differences in pixel value changes, the number of corner points, and the change of corner point coordinates in the current local area; the norm of the ratio of the liquidity factor to the steady-state factor is calculated to obtain the transmission loss probability factor of the local area. Specifically, the liquidity factor is the mean square error of the pixel value set of the local area of ​​two adjacent frames; the steady-state factor is the absolute value of the difference of the stable structure factor of the local area of ​​two adjacent frames, and the stable structure factor is the ratio of the number of corner points whose pixel coordinates overlap between the two adjacent frames to the number of all corner points in the previous frame of the two adjacent frames.

[0031] According to the radiographic image obtained from the internal radiographic image of the mud pump to be inspected, i.e., the X-ray flaw detection image, the preprocessed X-ray flaw detection image is segmented into several super-pixel regions using the super-pixel region segmentation algorithm. Each super-pixel region represents a tight pixel combination, i.e., a region with similar local features. Subsequent analysis is performed on each super-pixel region. Each super-pixel region is regarded as a local region.

[0032] Normally, the impact of mud residue on the defective area will affect its X-ray transmission, causing the edge of the defective area to become blurred. Since the mud residue area has a certain fluidity, it is manifested as a large change in the pixel values ​​of some areas at adjacent moments. At the same time, the defective area inside the mud pump does not have fluidity. Therefore, for the area with mud residue attached to the surface, it can be judged whether the area is an area where the defective area and the mud residue area coexist, and then it is smoothed to reduce the edge blurring caused by the mud residue area.

[0033] Specifically, by comparing the X-ray flaw detection images at adjacent moments, the pixel value changes in the same superpixel region (i.e., a local region) of two adjacent frames of flaw detection images in the time series are determined; at the same time, the number and coordinates of all corner points in each superpixel region of the target frame are obtained, and the number and coordinates of all corner points in the superpixel region at the same position in the next frame are obtained, and the number of corner points whose coordinates of two adjacent frames overlap is calculated, and the number is divided by the number of all corner points in the superpixel region in the target frame to obtain a stable structure factor, denoted as D. The larger the value, the more stable and unchanging structure points there are in two adjacent frames.

[0034] Construct a specific local area X-ray transmission loss probability factor formula:

[0035]

[0036] In the formula, i represents the i-th abnormal local area in the flaw detection image, a represents the a-th moment in the time series, b represents the left moment adjacent to the a-th moment in the time series, and norm() represents the number of all abnormal local areas in the flaw detection image. represents the pixel value set of the ith abnormal local area in the flaw detection image at the ath moment, represents the pixel value set of the i-th abnormal local area in the flaw detection image at the b-th moment, represents the stable structure factor of the i-th abnormal local area in the flaw detection image at the a-th moment, represents the stable structure factor of the i-th abnormal local area in the flaw detection image at the b-th moment, R i represents the X-ray transmission loss probability factor of the i-th abnormal local area.

[0037] In summary, It represents the mean square error between the pixel value set of the ith abnormal local area in the flaw detection image at the ath moment and the pixel value set of the ith abnormal local area in the flaw detection image at the bth moment, i.e., the fluidity factor. The larger the value, the greater the change in pixel value of the ith abnormal local area between adjacent frames, i.e., the ith abnormal local area has a certain fluidity, and there is mud residue in the area, and the greater the X-ray transmission loss probability factor of the area; It represents the absolute value of the difference between the stable structural factor of the ith abnormal local area in the flaw detection image at the ath moment and the stable structural factor of the ith abnormal local area in the flaw detection image at the bth moment, i.e., the steady-state factor. The larger the value, the smaller the possibility that the shape or structure of the abnormal local area has changed significantly between adjacent moments, and the smaller the possibility that the area has a defective area. It is subtracted from 1 to correct the logic.

[0038] It represents the ratio of the fluidity factor to the steady-state factor of the ith abnormal local area in the flaw detection image. The larger the value, the greater the fluidity of the ith abnormal local area, that is, the mud residue area, and the greater the stability of the defect area. The greater the possibility that the area is a defect area and the mud residue is attached to the surface, the greater the probability of loss when the X-ray penetrates the area, and the greater the probability of loss when the X-ray penetrates the area. i The larger the value of is, the greater the smoothing effect should be set for this area during the subsequent image smoothing process to ensure the clarity of the edge of the defect area.

[0039] Generally speaking, when there are interference factors such as larger mud particles in the mud residue, X-rays will produce more transmission losses when transmitting larger mud particles, resulting in the formation of local collapsed areas in the flaw detection image, which is specifically manifested as a difference in brightness from the surrounding areas and fewer detailed texture features in the local collapsed areas.

[0040] Specifically, the Canny edge detection algorithm is used to detect the edge in the obtained local abnormal region (ie, the local abnormal superpixel region), and the number of edge pixels in the local abnormal region is counted for subsequent analysis.

[0041] According to the transmission loss probability factor R i , multiple local abnormal regions are obtained from multiple local regions, and image analysis is performed on the multiple local abnormal regions to obtain the abnormal region collapse factor.

[0042] Construct a specific local area collapse factor formula:

[0043]

[0044] Where M represents the total number of local abnormal regions around the ith local abnormal region, m represents the mth local abnormal region around the ith local abnormal region, and L i represents the average pixel value of the i-th local abnormal area, L m represents the average pixel value of the mth local abnormal area, B i represents the number of edge pixels inside the i-th local abnormal region, E i represents the number of pixels inside the i-th local abnormal area, P i Represents the collapse factor of the i-th local abnormal area.

[0045] In summary, |L i ―L m | represents the absolute value of the difference between the average pixel value of the ith local abnormal area and the average pixel value of the mth local abnormal area, i.e., the brightness difference factor. The larger the value, the greater the brightness difference between the ith local abnormal area and the mth local abnormal area. It represents the brightness highlighting factor, which is the average of the brightness difference factors between the ith local abnormal area and all surrounding areas. The larger the value, the greater the average brightness difference between the ith local abnormal area and the surrounding abnormal areas, that is, the more significant the brightness difference between this area and the surrounding areas, indicating that the visual characteristics of the ith local abnormal area and the surrounding areas are quite different, which may indicate the existence of collapse abnormality. It represents the difference between 1 and the ratio of the number of edge pixels in the ith local abnormal region to the number of pixels in the ith local abnormal region, i.e., the texture feature factor. The larger the value, the more edge pixels there are in the local abnormal region, i.e., the edge features in the region are more prominent. However, in general, the collapsed region caused by mud particles does not have many edge features. Therefore, 1 is subtracted from it to correct the logic.

[0046] It represents the product of the brightness highlighting factor and the texture feature factor of the ith local abnormal area. The larger the value, the greater the brightness difference between the ith local abnormal area and the surrounding area. At the same time, the local abnormal area has fewer edge curve features. The greater the possibility that the local abnormal area is a collapsed area, the larger the value of Pi.

[0047] According to the collapse factor of the abnormal area, multiple local collapse areas are obtained from multiple local abnormal areas, and image analysis is performed on the multiple local collapse areas to obtain the transmission loss degree factor, which specifically includes: detecting the edge of the local abnormal area, analyzing the quantitative relationship between the number of internal pixels and the number of edge pixels in the local abnormal area, and obtaining the texture feature factor; according to the difference relationship between the average pixel value of the local abnormal area and the average pixel value of the surrounding area, the brightness highlighting factor is obtained; the texture feature factor and the brightness highlighting factor are multiplied to obtain the collapse factor of the abnormal area. The texture feature factor is the difference between 1 and the ratio of the number of pixels, where the ratio of the number of pixels refers to the ratio of the number of edge pixels to the number of internal pixels; the brightness highlighting factor is the average of the brightness difference factors between the abnormal area and all surrounding areas; where the brightness difference factor is the absolute value of the difference between the average pixel value of the abnormal area and the average pixel value of the surrounding area.

[0048] The transmission loss degree factor is obtained by performing image analysis on multiple collapsed areas, specifically including: the transmission loss degree factor is the product of the collapse factor of the abnormal area and the density of the collapsed area; wherein the density of the collapsed area is: the average of the distance differences between the collapsed area and all the local areas closest to it.

[0049] According to the abnormal area collapse factors of all local abnormal areas obtained, based on scene experience, the threshold T=0.5 is set, and all local abnormal areas in the overall flaw detection radiographic image are judged. Those greater than the threshold are marked as local collapsed areas, otherwise they are not marked.

[0050] Specifically, when there are more local collapse areas among all the local areas, it means that the mud residues in the local area have larger particles and other factors, which will cause greater losses during X-ray transmission. Therefore, a larger smoothing effect needs to be set in the subsequent smoothing process to compensate for the losses caused by the existence of local collapse areas.

[0051] Construct a specific local area X-ray transmission loss factor formula:

[0052]

[0053] Where P i represents the collapse factor of the ith local abnormal region, U represents the 5 local collapsed regions closest to the ith local abnormal region, j represents the jth local region among the 5 local collapsed regions closest to the ith local abnormal region, and Z i represents the centroid coordinates of the ith local abnormal area, Z j represents the centroid coordinates of the jth local collapse area, τ i Represents the transmission loss factor of the i-th local abnormal area.

[0054] In summary, It represents the average of the distance difference between the ith local abnormal area and all the local areas closest to it. The larger the value, the denser the local collapse area is around the ith local abnormal area, and the more large mud particles there are in the area, resulting in a greater degree of transmission loss of X-rays in the local area. It represents the product of the collapse factor of the ith local abnormal area and the density of the collapsed area. The larger the value, the more likely the area is to be a local collapsed area. At the same time, there are more local collapsed areas around, that is, there are more large mud particles in the area, resulting in a greater degree of transmission loss of X-rays in the local area. Then τ i The larger the value, the more smoothing strength is required in the subsequent smoothing process to ensure the contrast and clarity of the local area.

[0055] According to the transmission loss probability factor and the transmission loss degree factor, the smoothing filter weight coefficient is set to smooth the radiographic image, and the internal defect area of ​​the mud pump to be inspected is identified according to the smoothed radiographic image.

[0056] In areas where the mud particles are denser, the filter places more emphasis on preserving details and edge information, while in areas where the mud particles are sparser, the filter places more emphasis on smoothing to reduce noise and blur.

[0057] Construct a specific smoothing filter weight coefficient formula:

[0058] W i =R i ×τ i

[0059] In the formula, R i represents the X-ray transmission loss probability factor of the i-th local abnormal area, τ i represents the transmission loss factor of the i-th local abnormal area, W i Represents the smoothing filter weight coefficient of the i-th local abnormal area.

[0060] In summary, R i ×τ i It represents the X-ray transmission loss probability factor and transmission loss degree factor of the i-th local abnormal area. The larger the value, the more likely it is that the area is a defect area with mud residues attached to the surface. At the same time, there are influencing factors such as large mud particles in the mud residues. Therefore, in the subsequent smoothing process, it is necessary to set a larger smoothing filter weight to enhance the smoothing effect to ensure the clarity of the edge of the defect area. Then W i The larger the value of .

[0061] Wiener filtering is used for smoothing according to the smoothing filter weight coefficient. Wiener filtering is a classic image restoration technology, which is usually used to remove noise in images. By adjusting the scale factor of the Wiener filter, the balance between retaining image details and removing noise can be controlled. The smoothing filter weight coefficient is used as the scale factor of the Wiener filter to adjust the filtering strength of the Wiener filter. The scale factor is generally set by the image signal-to-noise ratio, but the image signal-to-noise ratio cannot be obtained without a clear reference image. Therefore, in this case, the smoothing filter weight coefficient of the local area is used as the scale factor of the Wiener filter, which can make the Wiener filter most efficient and have the best filtering effect when filtering out the noise in the internal structure flaw detection image of the mud pump.

[0062] Based on the smoothed internal flaw detection image of the mud pump, edge detection is used to extract possible defect area features from the image, and the extracted features are identified to detect and identify the defect area, obtain a more accurate defect area, and finally visualize the defect area to generate a report for further analysis and processing, so as to better evaluate and process the internal structure of the mud pump.

[0063] In a possible embodiment, the product of the transmission loss probability factor and the transmission loss degree factor is used as a smoothing filter weight coefficient, and the smoothing filter weight coefficient is used as a scale factor of Wiener filtering, and the radiation image is smoothed by Wiener filtering.

[0064] The embodiment of the present application analyzes the X-ray image based on whether there is mud residue attached to the surface of the defective area. When there is mud residue attached to the surface of the defective area, the X-ray will produce a certain loss during transmission, and the transmission loss probability factor generated therein is analyzed. At the same time, when there are a large number of large mud particles inside the mud residue, it will cause more collapsed areas when the X-ray is transmitted. Therefore, the collapse density of the local area (i.e., the super-pixel area) is analyzed to obtain the ray transmission loss degree factor of the local area. The transmission loss probability factor and the ray transmission loss degree factor are combined to obtain the X-ray transmission loss factor, and the ray transmission loss factor is used as the smoothing filter weight coefficient. An adaptive smoothing filter is set to perform smoothing processing to eliminate noise interference in the original X-ray image, thereby identifying the defective area and obtaining a more reliable and accurate internal structural defect area of ​​the mud pump.

[0065] The scheme of the present invention fully considers the interference factors such as mud particles existing on the surface of the internal structure of the mud pump, designs the smoothing filter weight coefficient, and uses an adaptive smoothing filter to filter the original image of the internal structure of the mud pump to compensate for the loss effect of the ray transmission process of the mud particles, thereby improving the contrast and clarity of the image.

[0066] 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.

[0067] 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

1. A method for detecting internal defects of a mud pump based on optical analysis, characterized in that: The method comprises: Obtaining the internal radiographic image of the mud pump to be inspected; The radiographic image in the radiographic image is divided into a plurality of local regions, and images of adjacent frames of each local region are analyzed to obtain a transmission loss probability factor of the local region; According to the transmission loss probability factor, a plurality of local abnormal regions are obtained from the plurality of local regions, and image analysis is performed on the plurality of local abnormal regions to obtain an abnormal region collapse factor; According to the abnormal area collapse factor, a plurality of local collapse areas are obtained from the plurality of abnormal areas, and image analysis is performed on the plurality of local collapse areas to obtain a transmission loss degree factor; According to the transmission loss probability factor and the transmission loss degree factor, a smoothing filter weight coefficient is set to smooth the radiographic image, and an internal defect area of ​​the mud pump to be inspected is identified according to the smoothed radiographic image; The transmission loss probability factor of each local area is obtained by analyzing the images of adjacent frames of each local area, specifically including: Analyze the same local area through two adjacent frames of ray images, and obtain the flow factor and steady-state factor of the local area according to the differences in pixel value changes, corner point number, corner point coordinate changes, etc. in the current local area; Calculating the norm of the ratio of the mobility factor to the steady-state factor to obtain the transmission loss probability factor of the local area; The mobility factor is the mean square error of the pixel value set of the local area in two adjacent frames; The steady-state factor is the absolute value of the difference between the stable structure factors of the local area in two adjacent frames, and the stable structure factor is the ratio of the number of corner points whose pixel coordinates of two adjacent frames overlap to the number of all corner points of the previous frame in the two adjacent frames; Performing image analysis on the multiple local abnormal regions to obtain abnormal region collapse factors specifically includes: Detect the edge of the local abnormal area, analyze the quantitative relationship between the number of pixels inside the local abnormal area and the number of edge pixels, and obtain a texture feature factor; obtain a brightness highlighting factor according to the difference between the average pixel value of the local abnormal area and the average pixel value of its surrounding area; Multiplying the texture feature factor and the brightness highlight factor to obtain the abnormal area collapse factor; The texture feature factor is the difference between 1 and the ratio of the number of pixels, wherein the ratio of the number of pixels refers to the ratio of the number of edge pixels to the number of internal pixels; The brightness highlighting factor is the average of the brightness difference factors between the local abnormal area and all surrounding areas; The brightness difference factor is the absolute value of the difference between the average pixel value of the local abnormal area and the average pixel value of its surrounding area; Performing image analysis on the multiple local collapse areas to obtain a transmission loss degree factor specifically includes: The transmission loss degree factor is the product of the collapse factor of the abnormal area and the density of the collapsed area; The density of the collapsed area is: the average of the distance difference between the local collapsed area and all the local areas closest to it; Identifying the internal defect area of ​​the mud pump to be inspected according to the smoothed radiographic image includes: extracting defect area features in the smoothed radiographic image by edge detection, identifying the defect area features, and obtaining the internal defect area.

2. A method for detecting internal defects of a mud pump based on optical analysis according to claim 1, characterized in that: Before segmenting the radiographic image, the radiographic image in the radiographic image is preprocessed, and the preprocessing includes denoising and graying.

3. A method for detecting internal defects of a mud pump based on optical analysis according to claim 2, characterized in that: Segmenting the radiographic image in the radiographic image into a plurality of local regions specifically includes: The preprocessed radiographic image is segmented into a plurality of superpixel regions using a superpixel region segmentation algorithm, and the superpixel regions are used as the local regions.

4. A method for detecting internal defects of a mud pump based on optical analysis according to any one of claims 1 to 3, characterized in that: A plurality of local collapse regions are obtained from the plurality of local abnormal regions, specifically: The local abnormal region whose abnormal region collapse factor is greater than a preset threshold is marked as a local collapsed region.

5. The method for detecting internal defects of a mud pump based on optical analysis according to claim 1, characterized in that: The product of the transmission loss probability factor and the transmission loss degree factor is used as a smoothing filter weight coefficient, and the smoothing filter weight coefficient is used as a scale factor of Wiener filtering, and the radiation image is smoothed by Wiener filtering.

Images