An underwater robot operation automatic deviation correction method, system, medium and product

Abstract

An underwater robot operation automatic deviation correction method, system, medium and product, relate to general control or regulation system field, the method comprises: obtaining a weld image, adjusting the filtering parameter according to the noise intensity of local image in the weld image, and obtaining a noise reduction image;The double-layer threshold segmentation is executed to the noise reduction image, and the binary image is obtained;Multiple detection areas are set in the binary image, the effective detection area is determined, the edge feature point in the effective detection area is converted into coordinate data, the partition weighted fitting is carried out, and the initial weld direction is obtained;Dynamic error zone is constructed, coordinate points outside the dynamic error zone are used as candidate abnormal points, the continuity criterion of adjacent feature points is combined to determine the final abnormal point in the candidate abnormal point, the remaining coordinate points are refitted, and the target weld direction is obtained;The target weld direction is compared with the reference direction, and the hierarchical correction control signal is generated according to the deviation degree. The application can improve the correction accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of general control or regulation systems, and particularly relates to a work automatic deviation correction method and system for an underwater robot, a medium and a product. BACKGROUND

[0002] In the field of nuclear industry, the spent fuel pool is a key facility for spent fuel storage and treatment, and its structural integrity is directly related to nuclear safety. The spent fuel pool is composed of a bottom surface and four side surfaces, and the weld seams between the surfaces are important guarantees for structural strength. In order to ensure the safe operation of the pool, the weld seams need to be detected and maintained regularly. Since the pool is in an underwater environment and the position of the weld seams is fixed, an underwater robot needs to be used to carry detection equipment to complete the detection task of the weld seams.

[0003] In the related art, the underwater robot obtains a weld seam image through a camera, removes underwater noise by using a fixed parameter image filtering method, and then extracts weld seam features by using a global threshold segmentation method. After feature extraction, a simple edge detection algorithm is used to obtain weld seam contour points, and the direction of the weld seam is calculated by linear fitting, and finally the motion direction of the robot is adjusted according to the calculation result.

[0004] However, in the actual underwater environment, due to the influence of water flow disturbance and suspended particles caused by the movement of the robot, the imaging quality of the weld seam image presents obvious differences in different areas, and the feature processing method in the related art is easily disturbed by local imaging quality fluctuations, resulting in a large deviation in the calculation of the weld seam direction. SUMMARY

[0005] The present application provides a work automatic deviation correction method and system for an underwater robot, a medium and a product, which can improve the deviation correction accuracy.

[0006] In a first aspect, the application provides a method for automatically correcting deviation of an underwater robot, applied to a control system, comprising: obtaining a weld seam image collected by a camera of the underwater robot, adjusting a filtering parameter according to a noise intensity of a local image in the weld seam image to obtain a denoising image; performing double-layer threshold segmentation on the denoising image to obtain a binary image; a first layer threshold of the double-layer threshold is used to extract a main contour of the weld seam, and a second layer threshold of the double-layer threshold is used to retain a feature of a transition region of the weld seam edge; setting a plurality of detection regions in the binary image, independently extracting edge feature points in each detection region, and determining an effective detection region according to a feature point quantity distribution of each detection region, and converting the edge feature points in the effective detection region into coordinate data; determining a weight coefficient according to a distribution density of the edge feature points in the detection region, and performing partition weighted fitting on the coordinate data to obtain an initial weld seam direction; constructing a dynamic error band based on the initial weld seam direction, taking coordinate points outside the dynamic error band as candidate abnormal points, determining final abnormal points in the candidate abnormal points in combination with a continuity criterion of adjacent feature points, re-fitting remaining coordinate points outside the final abnormal points to obtain a target weld seam direction; and comparing the target weld seam direction with a reference direction, and generating a hierarchical deviation correction control signal according to a deviation degree.

[0007] In the above embodiment, the control system adjusts the filtering parameter according to the noise intensity of the local image to improve the adaptability of image preprocessing, the double-layer threshold segmentation retains the main body and the edge transition feature of the weld seam, the multiple detection regions independently extract the feature points to avoid local interference, the weighted fitting based on the distribution density improves the contribution degree of the key region, and the dynamic error band screens the abnormal points to enhance the stability of the fitting result, thereby improving the accuracy of underwater weld seam tracking.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the step of performing double-layer threshold segmentation on the denoising image to obtain a binary image specifically comprises: performing gray scale normalization processing on the denoising image to obtain a standard gray scale image; calculating a gray scale histogram and a cumulative distribution function of the standard gray scale image to obtain a gray scale distribution feature; determining a first layer threshold based on the gray scale distribution feature to perform primary segmentation to obtain a weld seam candidate region; calculating a second layer threshold according to a gray scale distribution of the weld seam candidate region to perform secondary segmentation to retain an edge transition region; marking a region higher than the first layer threshold as a foreground pixel value, marking a region lower than the second layer threshold as a background pixel value, and obtaining a binary image.

[0009] In the above embodiment, the gray scale normalization processing of the control system eliminates the brightness difference of different regions, the gray scale histogram and the cumulative distribution function reflect the overall feature distribution of the image, the first layer threshold ensures the complete extraction of the main body of the weld seam, the second layer threshold retains the detail features of the edge transition region, and the binary processing of the foreground and the background lays a reliable foundation for subsequent feature extraction, thereby improving the segmentation accuracy of the weld seam features in the underwater environment.

[0010] In some embodiments of the first aspect, in the step of determining the weight coefficient according to the distribution density of the edge feature points in the detection region, partitioning and weighting fitting the coordinate data to obtain the initial weld direction, specifically comprising: calculating the projection distribution of the edge feature points in each detection region in the horizontal direction and the vertical direction to obtain a two-dimensional density distribution matrix; calculating the feature point aggregation degree of each region according to the two-dimensional density distribution matrix, and determining the weight coefficient according to the feature point aggregation degree; dividing the detection region along the weld direction into a plurality of equal-length sub-regions, and applying the corresponding weight coefficient to the feature point coordinates in each sub-region; performing least squares fitting on all weighted feature point coordinates to obtain the spatial equation of the weld, and determining the direction vector of the spatial equation as the initial weld direction.

[0011] In the above embodiment, the two-dimensional density matrix calculated by the control system through the projection distribution reflects the spatial distribution characteristics of the feature points, the weight coefficient determined based on the feature point aggregation degree highlights the contribution of high-quality regions, the sub-region division and weighting processing balances the influence of different regions, and the least squares fitting ensures the accuracy of the spatial equation.

[0012] In some embodiments of the first aspect, after the step of obtaining the weld image collected by the underwater robot camera, adjusting the filtering parameters according to the noise intensity of the local image in the weld image to obtain the denoising image, the method further comprises: obtaining the attitude data collected by the attitude sensor of the underwater robot, establishing a spatial transformation matrix of the image plane and the weld plane; performing perspective correction on the weld image according to the spatial transformation matrix to eliminate the perspective deviation; extracting the motion bubble trajectory in the weld image, establishing a bubble motion vector field, and determining the water flow disturbance direction; adjusting the image sampling timing according to the water flow disturbance direction to avoid the peak moment of water flow disturbance.

[0013] In the above embodiment, the attitude data and the spatial transformation matrix of the control system realize the perspective correction of the image, the bubble trajectory analysis establishes the water flow disturbance model, and the adjustment of the sampling timing avoids the interference peak, thereby improving the image acquisition quality.

[0014] In some embodiments of the first aspect, the step of extracting the motion bubble trajectory in the weld seam image, establishing a bubble motion vector field, and determining the water flow disturbance direction specifically comprises: performing inter-frame difference operation on the continuously collected multiple frames of images to extract the position changes of the moving targets, and screening according to the target area and gray scale distribution characteristics to obtain bubble targets; establishing a time sequence tracking sequence for the bubble targets, calculating the motion speed and acceleration of each bubble target, and constructing local bubble motion vectors; performing spatial interpolation on the local bubble motion vectors in the image plane to obtain a continuous bubble motion vector field distribution; performing cluster analysis on the bubble motion vector field to extract the water flow disturbance direction, and identifying the period parameters of the water flow disturbance according to the gradient changes of the vector field.

[0015] In the above embodiments, the inter-frame difference operation of the control system extracts the moving targets, and the area and gray scale feature screening ensures the accurate identification of the bubble targets, the time sequence tracking sequence reflects the bubble motion law, the vector field spatial interpolation constructs a continuous disturbance distribution, the cluster analysis extracts the disturbance direction and identifies the period parameters, and the accurate modeling of the water flow disturbance is realized.

[0016] In some embodiments of the first aspect, after the step of comparing the target weld seam direction with the reference direction and generating a hierarchical correction control signal according to the deviation degree, the method further comprises: obtaining contact pressure data of the robot suction device to generate a wall surface contact state distribution map; identifying a weld seam contour abnormal area according to the contact state distribution map, determining the weld seam structure type of the weld seam contour abnormal area; updating the detection area division scheme when the weld seam structure type changes; and selecting a corresponding correction execution strategy according to the weld seam structure type.

[0017] In the above embodiments, the contact pressure data of the control system reflects the wall surface state distribution, the identification of the weld seam contour abnormal area provides the structure type information, the dynamic update of the detection area division scheme adapts to the structure changes, and the selection of the correction execution strategy ensures the control accuracy under different conditions, thereby improving the adaptability of the underwater robot operation.

[0018] In some embodiments of the first aspect, in some embodiments, the step of selecting a corresponding deviation rectification execution strategy according to the weld structure type specifically comprises: classifying the weld structure type into a flat weld, a fillet weld and a lap weld according to the geometric features and the pressure distribution features of the weld contour abnormal area; when the weld structure type is the flat weld, a one-way progressive deviation rectification strategy is adopted, the robot pose is adjusted step by step according to the deviation direction until the reference direction is returned; when the weld structure type is the fillet weld, a two-way coordinated deviation rectification strategy is adopted, the lateral position and the steering angle of the robot are adjusted at the same time to keep the relative position between the detection area and the weld contour stable; when the weld structure type is the lap weld, a step transition deviation rectification strategy is adopted, the deviation rectification action is paused at the weld step, and the deviation rectification reference is re-established after the robot crosses the step.

[0019] In the above embodiments, the control system classifies the weld structure according to the geometric features and the pressure distribution features, adopts corresponding deviation rectification strategies for different types of welds, and flexibly applies one-way progressive, two-way coordinated and step transition strategies, thereby realizing accurate deviation rectification control in a complex environment.

[0020] In the second aspect, the embodiments of the present application provide a control system, which comprises: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program codes, the computer program codes comprise computer instructions, and the one or more processors invoke the computer instructions to make the control system execute the method described in the first aspect and any possible implementation manner of the first aspect.

[0021] In the third aspect, the embodiments of the present application provide a computer program product comprising instructions, which, when executed on a control system, cause the control system to perform the method described in the first aspect and any possible implementation manner of the first aspect.

[0022] In the fourth aspect, the embodiments of the present application provide a computer readable storage medium comprising instructions, which, when executed on a control system, cause the control system to perform the method described in the first aspect and any possible implementation manner of the first aspect.

[0023] It can be understood that the control system provided in the second aspect, the computer program product provided in the third aspect and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiments of the present application. Therefore, the beneficial effects that can be achieved are referred to the beneficial effects in the corresponding method, which will not be described here.

[0024] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0025] 1. Due to the adoption of dynamic filtering parameter adjustment based on local noise intensity, double-layer threshold segmentation, multi-region independent feature extraction, and weighted fitting based on distribution density, the filtering effect can be adaptively optimized according to the imaging quality of different regions in the image preprocessing stage, the main contour of the weld is retained while the detail information of the edge transition area is protected in the feature extraction stage, and the independent processing of multiple detection regions avoids the propagation of local interference in the feature analysis stage, while the weight coefficient is determined by the feature point distribution density to highlight the contribution of high-quality regions, effectively solving the problems of excessive smoothing caused by fixed parameter filtering, loss of edge details due to single threshold segmentation, and influence of local interference on global feature extraction in the prior art, thereby realizing high precision and high robustness of underwater weld tracking and improving the reliability of underwater robot operation.

[0026] 2. Due to the adoption of gray scale normalization processing, double-layer threshold determination based on gray scale distribution characteristics, and binary processing of foreground and background separation, the brightness differences of different regions are eliminated in the image preprocessing stage, the feature distribution law of the image is accurately grasped by analyzing the gray scale histogram and the cumulative distribution function in the threshold determination stage, the first layer threshold is used to ensure the complete extraction of the main weld, and the second layer threshold is used to retain the detail information of the edge transition area, effectively solving the problems of uneven illumination in underwater environment, difficulty in balancing main integrity and edge details with single threshold, and unstable segmentation results in the prior art, thereby realizing high-quality segmentation of weld features in underwater environment and laying a reliable data foundation for subsequent feature extraction and direction calculation.

[0027] 3. Due to the adoption of wall surface state analysis based on contact pressure data, weld contour abnormal region identification, detection region dynamic division, and hierarchical correction strategy, the contact state of the robot and the wall surface can be monitored in real time, the weld structure change is discovered in time and its type is identified, the division scheme of the detection region is adjusted according to the structure change, and the most suitable correction strategy is selected for different types of weld structure, effectively solving the problems of delayed response to weld structure change, feature extraction failure caused by fixed detection region division, and single correction strategy difficult to adapt to complex environment in the prior art, thereby realizing accurate correction control of underwater robots in complex environment and improving the operation adaptability and reliability. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a flowchart of the operation automatic correction method of the underwater robot in the embodiment of the present application;

[0029] Figure 2 is another flowchart of the operation automatic correction method of the underwater robot in the embodiment of the present application;

[0030] Figure 3 is a schematic diagram of an entity device structure of a control system in an embodiment of the present application. DETAILED DESCRIPTION

[0031] The terminology used in the following embodiments of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used in the description of the application, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or", as used in the description of the application, refers to any or all possible combinations of one or more of the associated listed items.

[0032] Hereinafter, the terms "first", "second", "third", etc. are used only for the purpose of description and are not to be understood as implying or suggesting relative importance or a specific number of technical features indicated. Thus, the features defined with "first", "second", etc. can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise stated.

[0033] In order to facilitate understanding, the application scenarios of the embodiments of the present application are introduced as follows.

[0034] In the spent fuel pool maintenance project of a certain nuclear power plant, automatic welding repair needs to be performed on the underwater storage racks and pipelines. Due to the special environment of the spent fuel pool, the water depth is usually more than 10 meters, and there is a radiation environment, so a remotely operated underwater robot must be used for operation. The circulating water system in the pool will cause complex water flow disturbance, and the thermal convection caused by the decay heat of the spent fuel will also cause water disturbance. In addition, the radiation interference to the camera equipment makes the underwater field of view blurred and the image noise serious. Especially in the area close to the spent fuel assembly, strong radiation will cause spot noise of the image sensor, and the Cherenkov radiation reflected by the pool wall will cause uneven blue light interference. These factors seriously affect the accuracy of the weld tracking and increase the difficulty and risk of the repair operation.

[0035] In the related art, feature extraction and tracking of underwater welds can be achieved by using an image filtering algorithm with fixed parameters and a single threshold segmentation method. Specifically, first, the image is subjected to Gaussian filtering for noise reduction, then the OTSU algorithm is used to determine the global threshold for binaryzation segmentation, and finally the weld direction is obtained by detecting the straight line feature through Hough transform. The following introduces the scene of using the automatic deviation correction method of the underwater robot in the related art.

[0036] The Institute of Nuclear and New Energy Technology of Tsinghua University uses traditional image processing methods for spent fuel pool welding operations. They use a fixed parameter median filter to remove pulse noise caused by radiation, and a single threshold Canny operator for edge detection. However, in practical applications, when the underwater robot approaches areas with different radiation intensities, the image quality changes, and fixed processing parameters cannot adapt. At the same time, the blue light interference caused by Cerenkov radiation can affect the accuracy of edge detection, leading to unstable weld feature extraction. In addition, due to the use of global feature extraction method, once a certain area is disturbed by strong radiation, the overall tracking effect will be affected, causing the robot to deviate from the predetermined trajectory.

[0037] The underwater robot operation automatic deviation correction method in the embodiments of the present application adjusts the filter parameters and uses double-layer threshold segmentation to retain edge details, and at the same time uses multiple detection areas to independently extract feature points to avoid local disturbance propagation, achieving accurate tracking of underwater welds, not only improving the accuracy of feature extraction, but also enhancing the adaptability of the system to changes in the underwater environment. The following describes the scene in which the underwater robot operation automatic deviation correction method in the embodiments of the present application is used.

[0038] By applying the basic scheme of the present application, a nuclear power station has implemented a radiation-proof adaptive weld tracking system in spent fuel pool maintenance. By analyzing the radiation interference degree of the image in real time, the system can adjust the filter parameters to suppress radiation noise. The double-layer threshold segmentation method not only effectively separates the interference of Cerenkov radiation, but also ensures the integrity of the weld contour. The system divides the detection area into multiple independent units, so even if some areas are affected by strong radiation, other areas can still provide reliable feature information. The weighted fitting method based on radiation intensity distribution preferentially uses feature data from low radiation areas, and the dynamic error band can effectively filter out abnormal points affected by radiation interference.

[0039] As can be seen, by using the underwater robot operation automatic deviation correction method in the embodiments of the present application, accurate extraction of weld features can be achieved while effectively solving the problems of underwater environmental noise interference, unstable image quality, and local feature abnormalities, thereby achieving high-precision and high-reliability tracking control of underwater welding robots.

[0040] For ease of understanding, the method provided by the present embodiment will be described in the following flowchart in conjunction with the above-mentioned scene. Please refer to Figure 1 , which is a flowchart of the underwater robot operation automatic deviation correction method in the embodiments of the present application.

[0041] S101, acquire the weld image collected by the underwater robot camera, adjust the filter parameters according to the noise intensity of the local image in the weld image, and obtain a denoised image.

[0042] Wherein, the underwater robot camera represents an image sensor device installed on the underwater robot for image acquisition; the weld image refers to a digital image containing the weld area collected by the camera; the local image noise intensity represents the noise interference degree of a specific area in the image; the filtering parameter is used to represent the control parameters in the image filtering algorithm, including window size, weight coefficient, etc.; the denoising image refers to the image after noise removal processing.

[0043] When the underwater robot starts work, the control system first needs to obtain a clear weld image as the basis for subsequent processing. Specifically, the control system collects the weld image in real time through the underwater robot camera, and divides the image into multiple local areas. For each local area, the control system calculates the variance of the image gray value and the gradient distribution to evaluate the noise intensity of the area. According to the size of the noise intensity, the control system adjusts the filtering algorithm parameters applied to the area, and the area with larger noise intensity adopts larger filtering strength, and the area with smaller noise intensity adopts smaller filtering strength, so as to remove noise while retaining image detail information to the greatest extent.

[0044] In actual application, due to the uneven illumination in underwater environment, the noise characteristics of different areas in the same image may be different. For this, the control system can introduce a local contrast enhancement mechanism: first, calculate the local contrast distribution map of the image, identify the areas with abnormal contrast, and perform adaptive histogram equalization processing on these areas to improve the local contrast of the image. Then perform noise evaluation and filtering processing, so as to ensure that the adjustment of the filtering parameter is more accurate and effective.

[0045] S102, perform double-layer threshold segmentation on the denoising image to obtain a binary image.

[0046] Wherein, the denoising image represents a digital image after noise removal processing; the double-layer threshold segmentation refers to a processing method of image segmentation using two different thresholds; the first layer threshold is used to represent the gray threshold for extracting the main contour of the weld; the second layer threshold is used to represent the gray threshold for retaining the transition area of the weld edge; the binary image refers to the processing result of converting the image pixel value to two discrete values, usually 0 represents background and 1 represents target.

[0047] After obtaining the denoised image, the control system needs to accurately segment the weld region from the background. Specifically, the control system first performs gray scale normalization on the denoised image to eliminate the brightness difference between different regions. Then, the gray scale histogram and cumulative distribution function of the normalized image are calculated to analyze the gray scale distribution characteristics of the image. Based on the gray scale distribution characteristics, the control system uses the OTSU algorithm to determine the first layer threshold value for extracting the main body region of the weld. Next, the control system analyzes the gray scale distribution of the weld candidate region and uses an adaptive threshold algorithm to calculate the second layer threshold value for retaining the transition region of the weld edge. Finally, the control system marks the region above the first layer threshold value as foreground and the region below the second layer threshold value as background to obtain the final binary image.

[0048] It should be noted that the double-layer threshold segmentation is an adaptive segmentation method for underwater weld image features. The system first performs gray scale normalization preprocessing on the image, and then determines two key threshold values by analyzing the gray scale histogram and cumulative distribution function. The first layer threshold value is mainly based on the overall gray scale distribution characteristics of the image, and is determined by maximizing the inter-class variance, which is used to reliably extract the main contour of the weld; the second layer threshold value is determined by analyzing the local gray scale distribution of the weld candidate region, which is used to retain the transition region characteristics of the weld edge. The region between the two threshold values is considered as the edge transition zone, and the system will dynamically determine the pixel attribution based on the local region characteristics. When the turbidity of the underwater environment changes, the system will automatically adjust the interval between the two threshold values, while ensuring the integrity of the main body of the weld and retaining as much edge detail information as possible. For example, when the water body is relatively turbid, the system will appropriately reduce the interval between the two threshold values to prevent feature loss caused by excessive segmentation; when the water body is clear, the threshold interval will be expanded to obtain more rich edge features.

[0049] In practical applications, the turbidity of the underwater environment may change over time, resulting in unstable image contrast. To this end, the control system can use a dynamic threshold adjustment mechanism: by tracking the gray scale distribution changes of multiple consecutive images, a dynamic model of the gray scale distribution is established. When detecting changes in the gray scale distribution, the control system automatically adjusts the threshold calculation strategy, which can be adjusting the inter-class variance weight of the OTSU algorithm or modifying the local window size of the adaptive threshold, to ensure the stability of the segmentation result. For example, when the turbidity of the water body increases, the interval between the two threshold values can be appropriately reduced to better retain the weld features.

[0050] S103、In the binary image, a plurality of detection regions are set, each detection region independently extracts edge feature points, and the number of feature points in each detection region is determined to determine an effective detection region, and the edge feature points in the effective detection region are converted into coordinate data.

[0051] Among them, the binary image represents the black and white image after double-layer threshold segmentation; the detection area refers to the local rectangular area divided in the image for feature extraction; the edge feature points are used to represent the key pixel points on the edge of the weld contour; the distribution of the number of feature points represents the density and distribution of edge feature points in each detection area; the effective detection area refers to the detection area containing effective weld features; the coordinate data is used to represent the position information of the edge feature points in the image coordinate system.

[0052] After obtaining the binary image, the control system needs to accurately extract the weld features. Specifically, the control system first sets up multiple rectangular detection areas at equal intervals along the general direction of the weld in the binary image. The size of each detection area is determined according to the image resolution and the weld width. For each detection area, the control system independently applies an edge detection algorithm to extract feature points, including steps such as calculating pixel gradients, non-maximum suppression, and dual threshold detection. Then, the control system analyzes the number and spatial distribution of feature points in each detection area, identifies areas with abnormal feature point distribution through cluster analysis, and excludes them to obtain effective detection areas. Finally, the control system maps the edge feature points in the effective detection area to the image coordinate system to obtain standardized coordinate data.

[0053] In actual applications, rust or attachments may exist on the weld surface, resulting in abnormal distribution of feature points in certain detection areas. To address this, the control system can introduce a feature point quality assessment mechanism: first, a feature template of the ideal weld edge is established, including the density distribution and spatial arrangement pattern of the feature points. For each feature point extracted from the detection area, its matching degree with the template is calculated. When the matching degree is lower than the threshold, the control system automatically expands the detection range of the area or adjusts the feature extraction parameters until a set of feature points that meets the quality requirements is obtained. For example, for areas with sparse feature points, the gradient threshold of edge detection can be lowered, while adding neighborhood constraints to ensure the reliability of the feature points.

[0054] S104 , determining a weight coefficient according to the distribution density of edge feature points in the detection area, performing partition weighted fitting on the coordinate data, and obtaining an initial weld direction.

[0055] Among them, the distribution density indicates the degree of aggregation of edge feature points in the detection area; the weight coefficient is used to indicate the importance of feature points in different regions in the fitting calculation; partition weighted fitting refers to the method of assigning different weights to feature points in different regions for curve fitting; the initial weld direction represents the preliminary estimate of the spatial direction of the weld obtained by fitting; the coordinate data refers to the position information of the edge feature points in the image coordinate system.

[0056] After obtaining the coordinates of the feature points in the effective detection area, the control system needs to accurately calculate the weld direction. Specifically, the control system first calculates the horizontal and vertical projection distribution of the feature points in each detection area and constructs a two-dimensional density distribution matrix. Based on this matrix, the control system calculates the aggregation index of the feature points in each area, including local density, spatial uniformity and other characteristics. According to the aggregation index, the control system assigns a weight coefficient to each detection area, and the area with the more concentrated and uniform distribution of feature points obtains a higher weight. Then, the control system divides the detection area into several equal-length sub-areas along the estimated weld direction, and applies the corresponding weight coefficient to weight the feature points in each sub-area. Finally, the control system uses the least squares method to perform spatial curve fitting on all weighted feature points to obtain the spatial equation of the weld, and determines the direction vector of the equation as the initial weld direction.

[0057] In practical applications, welds may contain curved or deformed areas, resulting in large errors in linear fitting. To address this, the control system can adopt a segmented weighted fitting strategy: First, based on the spatial distribution characteristics of the characteristic points, the turning points or curvature change points of the weld are identified. The weld area is then divided into multiple subsegments using these characteristic points as boundaries. Weight calculation and curve fitting are performed separately for each subsegment, and polynomial models of different orders can be selected. Finally, the control system smoothly connects the fitting results of each subsegment through methods such as spline interpolation to obtain a complete description of the weld direction. For example, for sharp turns, a higher-order polynomial model can be used, while increasing the weights of the characteristic points in this area to improve fitting accuracy.

[0058] S105. Construct a dynamic error band based on the initial weld direction, use the coordinate points outside the dynamic error band as candidate outlier points, determine the final outlier point among the candidate outlier points based on the continuity criterion of adjacent feature points, refit the remaining coordinate points outside the final outlier point, and obtain the target weld direction.

[0059] Among them, the dynamic error band represents a variable width tolerance area constructed according to the initial weld direction; the candidate outlier point refers to the suspicious feature point that falls outside the error band; the continuity criterion is used to represent the criterion for evaluating the spatial continuity of adjacent feature points; the final outlier point refers to the outlier feature point confirmed by the continuity judgment; the target weld direction refers to the final weld direction obtained by refitting after removing the outlier point.

[0060] After obtaining the initial weld seam direction, the control system needs to further optimize the fitting result. Specifically, the control system first constructs a dynamic error band along the initial weld seam direction, and the error band width is adjusted according to the dispersion degree of the local feature point distribution. The more concentrated the feature point distribution in a region, the narrower the error band. Then, the control system identifies feature points falling outside the error band and marks them as candidate abnormal points. For each candidate abnormal point, the control system analyzes the spatial distance, angle, and curvature between it and adjacent feature points, and establishes a continuity evaluation model. Based on the model, the control system determines whether the candidate abnormal point violates the continuity feature of the weld seam contour, and determines the points that violate continuity as final abnormal points. Finally, the control system removes these abnormal points from the feature point set, and re-fits the remaining feature points to obtain a more accurate target weld seam direction.

[0061] It should be noted that the dynamic error band can be adaptively screened based on the weld feature distribution. The system takes the weld seam direction obtained by initial fitting as a reference to construct a tolerance region with variable width. The width of the error band is not a fixed value, but is dynamically adjusted according to the distribution characteristics of the local feature points: in the region where the feature points are densely and regularly distributed, the error band is narrowed to improve the screening accuracy; in the region where the feature points are sparsely or irregularly distributed, the error band is appropriately widened to accommodate reasonable fluctuations. For coordinate points falling outside the error band, the system further analyzes the continuity features such as spatial distance, angle, and curvature between them and adjacent feature points, and comprehensively judges whether to mark them as abnormal points. In actual application, when a large range of continuous abnormal points are detected in a region, the system will combine the pressure distribution characteristics and historical data of the region to evaluate whether it is caused by weld structure changes, to avoid misjudgment of normal structure feature points. This dynamic adjustment mechanism ensures that in complex underwater environment, the true abnormal points can be accurately identified and removed, while the effective points reflecting the actual weld features are preserved.

[0062] In actual application, local defects or bubbles on the weld seam surface may cause a large number of continuous feature points to be misjudged as abnormal points. For this purpose, the control system can introduce an abnormal point verification mechanism based on regional consistency: first, the detected abnormal points are clustered according to their spatial positions to form abnormal regions. For each abnormal region, the shape feature and the distribution pattern of the internal feature points are analyzed. If the abnormal region shows a similar pattern to the weld defect or bubble feature, the control system will re-evaluate the effectiveness of the feature points in the region, and if necessary, adjust the error band parameters or the continuity criterion threshold to avoid excessive deletion of effective feature information. For example, for a region suspected to be a weld pit, the continuity criterion constraint can be appropriately relaxed to retain feature points reflecting the actual weld shape.

[0063] S106, compare the target weld seam direction with the reference direction, and generate a hierarchical correction control signal according to the deviation degree.

[0064] Wherein, the target weld direction represents the final determined spatial orientation of the weld; the reference direction is used to represent the ideal weld motion trajectory, which is set before the operation begins and input into the control system, and is a path pre-planned through analysis of the design drawings of the workpiece or through teaching; the deviation degree refers to the angle and positional deviation between the target direction and the reference direction; the hierarchical correction control signal represents different levels of control instructions generated according to the deviation degree, and is used to guide the underwater robot to adjust the posture.

[0065] After obtaining the target weld direction, the control system needs to generate corresponding correction control instructions. Specifically, the control system first converts the target weld direction into a direction vector in the standard coordinate system, and obtains the preset reference direction. Then, the control system calculates the included angle and perpendicular distance between the two direction vectors, and comprehensively evaluates the deviation degree of the weld tracking. Based on the deviation degree, the control system divides the deviation state into multiple levels, such as slight deviation, moderate deviation and severe deviation. For each deviation level, the control system designs a corresponding correction strategy, including adjusting the motion speed, turning angle and posture parameters. Finally, the control system converts the correction strategy into a specific control signal sequence and sends it to the execution mechanism of the underwater robot.

[0066] It should be noted that the reference direction, as the ideal motion trajectory reference in the underwater robot weld tracking process, its determination process includes two stages of static presetting and dynamic updating. In the static presetting stage, the control system first extracts the theoretical direction data of the weld from the construction drawings and pool structure design documents, and establishes an initial reference direction database combined with historical detection records. In the actual operation process, the control system will dynamically update the reference direction according to multi-source sensing data: first, the statistical average direction is calculated using continuous multiple frames of weld features obtained by image processing, and the space is corrected combined with the attitude sensor data, and then the local reference parameters are adjusted according to the structure characteristics feedback by the pressure sensor. Through the data fusion algorithm, the control system combines the static preset value and the dynamic detection result by weighting, to obtain the final optimized reference direction. The weight distribution will be adaptively adjusted according to the weld quality requirement, the degree of environmental disturbance and the state of the robot.

[0067] For example, in the maintenance task of a spent fuel pool in a nuclear power plant, when the underwater robot passes through a straight weld segment, the control system mainly uses the preset theoretical direction as the reference; when approaching the corner area, the system will increase the weight of the real-time detection result, predict the corner curvature through multi-frame image analysis, and adjust the reference direction in advance. For the area disturbed by water flow, the system will combine the bubble motion feature analysis to apply compensation correction to the reference direction, to ensure the stability of the tracking control.

[0068] In practical applications, underwater robots' posture adjustments may experience response lag due to mechanical inertia or water flow disturbances. To address this, the control system can employ an adaptive control gain strategy: by monitoring the robot's motion state and environmental parameters in real time, a dynamic response model is established. When response lag is detected, the control system automatically adjusts the control gain parameters, increasing the control gain appropriately for directions with slower response and reducing it appropriately for directions prone to overshoot. Simultaneously, a velocity feedforward compensation mechanism is introduced to estimate the impact of water flow velocity and direction on the robot's motion and preemptively apply compensatory control. For example, in strong water flow, the system increases the control gain perpendicular to the flow direction and calculates the required compensation torque based on the water flow velocity to ensure the accuracy and stability of the corrective action.

[0069] This dynamic response model is used to adjust the control gain during the underwater robot's posture adjustment. During the training phase, the system collects the robot's motion state data (including velocity, acceleration, and attitude angle) and environmental parameters (such as water velocity and direction) under different water flow conditions as training samples. The deviation between the actual response time and the ideal response time is used as the loss function for training. The model itself is a state predictor. Its inputs include the current motion state parameter vector and the environmental parameter vector. It uses a multi-layer perceptron structure to perform feature extraction and nonlinear mapping, and outputs predicted response characteristics. During the operation phase, the control system inputs the real-time state data and environmental data into the model. Based on the predicted response characteristics, it calculates the control gain adjustment coefficients for each direction, increasing the gain in directions with slow response and reducing the gain in directions prone to overshoot. The system also calculates the required compensation torque based on the water flow velocity vector. For example, if a vertical response lag is detected, the model outputs a larger vertical gain coefficient and simultaneously calculates the required compensation torque based on the lateral component of the water flow, thereby achieving precise control of posture adjustment.

[0070] The above embodiment mainly describes basic feature extraction and direction estimation methods. In practical applications, functional modules such as posture compensation, water flow analysis, and structure recognition can be combined to further improve the system's environmental adaptability and control accuracy. The following supplements the scenarios of this embodiment.

[0071] In the maintenance project of the grid of the spent fuel pool of a nuclear power plant, the engineering team further optimizes the system performance. By introducing a real-time radiation dose rate monitoring module, the system can predict high radiation areas and adjust the work path in advance. The image sampling strategy automatically avoids strong radiation periods while adjusting the exposure parameters to reduce the impact of radiation on the sensor. The system can also adaptively update the image processing parameters according to the radiation intensity distribution at different depths of the pool. For example, in the high radiation area near the bottom of the pool, the system increases the sampling frequency and enables stronger filtering strength; at the corner weld, the system selects the optimal detection area layout according to the local radiation environment to ensure the stability of feature extraction. These optimization measures improve the system's work reliability in strong radiation environments, ensuring the maintenance quality while reducing the radiation exposure of workers.

[0072] In combination with the above scenarios, the following further more specific flow description of the method provided by the present embodiment is provided. Please refer to Figure 2 , another flow diagram of the automatic deviation correction method for the underwater robot in the embodiments of the present application.

[0073] S201, obtain the weld image collected by the underwater robot camera, adjust the filtering parameters according to the noise intensity of the local image in the weld image, and obtain the denoising image.

[0074] Referring to step S101, the control system dynamically optimizes the filtering effect according to the local noise intensity of the weld image.

[0075] In some embodiments, the control system optimizes image acquisition through pose data and bubble trajectory analysis, that is, the control system obtains the pose data collected by the pose sensor of the underwater robot, establishes a spatial transformation matrix between the image plane and the weld plane; performs perspective correction on the weld image according to the spatial transformation matrix to eliminate the perspective deviation; extracts the motion bubble trajectory in the weld image, establishes a bubble motion vector field, and determines the water flow disturbance direction; adjusts the image sampling timing according to the water flow disturbance direction to avoid the peak moment of water flow disturbance.

[0076] Wherein, the pose sensor represents a sensing device for measuring the spatial pose of the underwater robot; the pose data refers to the spatial pose parameters of the robot such as pitch angle, roll angle and heading angle; the spatial transformation matrix is used to represent the conversion relationship between different coordinate systems; the perspective correction refers to eliminating the image distortion caused by the change of viewing angle; the motion bubble trajectory represents the spatial path of the motion of the bubble in water; the bubble motion vector field refers to the vector distribution describing the motion trend of the bubble; the water flow disturbance direction is used to represent the motion direction of the local water flow; the sampling timing represents the time sequence of image acquisition.

[0077] Before image processing, the control system needs to eliminate the perspective deviation and optimize the sampling strategy. Specifically, the control system first reads the attitude angle data collected by the attitude sensor in real time, calculates the spatial transformation relationship between the image plane and the weld plane based on the camera calibration parameters. Based on this transformation relationship, the control system constructs a perspective transformation matrix to correct the original image geometrically and eliminate the image distortion caused by the change of the robot posture. Then, the control system identifies the moving bubbles in the water through image sequence analysis, and records the motion trajectory of the bubbles. Through statistical analysis of multiple bubble trajectories, the control system establishes a vector field model reflecting the local water flow motion characteristics. Finally, according to the time-varying characteristics of the water flow vector field, the control system predicts the peak time of water flow disturbance, adjusts the time interval of image sampling, and avoids image acquisition during strong disturbance.

[0078] In some embodiments, the posture compensation and sampling optimization can be achieved in various ways: optionally, the control system can use a posture estimation method based on Kalman filtering, first fuse the measurement data of the accelerometer and the gyroscope to predict the posture change trend of the robot, and then adaptively adjust the perspective transformation parameters according to the prediction results, while optimizing the image sampling timing to match the posture change period; optionally, the control system can use a motion compensation method based on optical flow analysis, estimate the relative motion by calculating the optical flow field between consecutive image frames, construct a compensation model combined with the attitude data, and adjust the image correction parameters and sampling strategy. It can be understood that other ways of motion compensation and sampling control can also be used, which are not limited here.

[0079] In some embodiments, the control system will track the bubble motion characteristics through multi-frame image analysis, that is, the control system will perform inter-frame difference operation on the continuously collected multiple image frames, extract the position change of the moving target, and obtain the bubble target according to the target area and gray distribution characteristics; a time sequence tracking sequence is established for the bubble target, the motion speed and acceleration of each bubble target are calculated, and a local bubble motion vector is constructed; the local bubble motion vectors are spatially interpolated on the image plane to obtain a continuous bubble motion vector field distribution; the bubble motion vector field is clustered and analyzed to extract the water flow disturbance direction, and the cycle parameters of the water flow disturbance are identified according to the gradient change of the vector field.

[0080] Wherein, the inter-frame difference operation means calculating the pixel change between adjacent image frames; the moving target refers to the object in the image that changes position; the target area indicates the number of pixels occupied by the bubble in the image; the gray distribution characteristics are used to represent the brightness distribution pattern of the bubble; the time sequence tracking sequence refers to the time sequence recording the motion state of the bubble; the local bubble motion vector represents the motion direction and speed of a single bubble; the spatial interpolation means constructing a continuous distribution between discrete observation points; the clustering analysis is used to represent the grouping of similar motion characteristics; the cycle parameter refers to the time characteristic of the water flow disturbance.

[0081] After acquiring the image sequence, the control system needs to analyze the water flow disturbance characteristics. Specifically, the control system first performs inter-frame difference calculation on the continuously collected image sequence to extract dynamic targets in the image. By setting area threshold and gray feature constraints, the control system filters out targets that meet the bubble characteristics and removes other interference targets. For each identified bubble target, the control system establishes a motion state sequence containing position, velocity and acceleration, and smoothes the motion state estimation through Kalman filtering. Then, the control system projects the motion vector of each bubble onto the image plane, uses radial basis function for spatial interpolation, and constructs a continuous motion vector field. Finally, the control system performs density clustering analysis on the vector field to extract the main water flow direction, and identifies the periodic characteristics of the water flow disturbance through Fourier analysis.

[0082] In practical applications, the motion trajectories of multiple bubbles may interfere with each other or intersect, affecting the extraction accuracy of water flow characteristics. For this purpose, the control system can use a hierarchical trajectory management strategy: first, establish a multi-level trajectory tracking model, the bottom layer is responsible for the state estimation of a single bubble, the middle layer handles trajectory interaction and association, and the top layer analyzes the overall motion pattern. When detecting trajectory intersection or occlusion, the system will start trajectory prediction and re-association mechanism to restore trajectory continuity through motion consistency and appearance similarity. For example, for temporarily disappeared bubbles, the system will maintain their trajectory hypothesis for a period of time, and wait for their re-appearance to connect the trajectory, ensuring the integrity of water flow feature extraction.

[0083] S202, perform gray scale normalization processing on the denoised image to obtain a standard gray scale image.

[0084] Wherein, the denoised image represents the original image after noise suppression processing; gray scale normalization represents the process of mapping image pixel values to a standard range; the standard gray scale image refers to an image whose pixel values are normalized to the interval [0, 1] or [0, 255]; pixel intensity represents the brightness value of each point in the image; gray scale distribution feature is used to represent the brightness distribution of the whole image; dynamic range refers to the difference between the maximum and minimum gray scale values of the image; contrast represents the difference in brightness of the image; linear mapping function is used to represent the mathematical relationship of gray scale conversion; histogram equalization refers to the processing method of optimizing gray scale distribution; normalization coefficient represents the proportion factor of gray scale mapping.

[0085] After obtaining the denoised image, the control system needs to perform gray scale standardization processing to eliminate the influence of uneven illumination. Specifically, the control system first analyzes the gray scale histogram distribution of the denoised image, calculates the maximum gray scale value, minimum gray scale value and average gray scale value of the image. Then, the control system selects the appropriate normalization strategy according to the dynamic range characteristics of the image: when the image contrast is low, the control system uses linear stretching method to expand the gray scale dynamic range by remapping; when the image has local dark or bright areas, the control system uses piecewise linear mapping to use different mapping coefficients for different gray scale intervals; when the gray scale distribution of the image is uneven, the control system optimizes the gray scale distribution combined with histogram equalization technology. Finally, the control system maps the processed gray scale value to the standard range to generate a standardized gray scale image. To ensure the stability of the processing, the control system will perform amplitude limiting processing on abnormal gray scale values to avoid overflow.

[0086] S203, calculate the gray scale histogram and cumulative distribution function of the standard gray scale image to obtain the gray scale distribution characteristics.

[0087] Among them, the standard gray scale image represents the image after gray scale normalization processing; the gray scale histogram is used to represent the statistical distribution of the number of pixels of each gray scale level in the image; the cumulative distribution function refers to the cumulative sum curve of the gray scale histogram; the gray scale distribution characteristics represent the statistical characteristics reflecting the overall distribution law of the image gray scale value.

[0088] After obtaining the standard gray scale image, the control system needs to analyze the gray scale distribution characteristics of the image. Specifically, the control system first counts the number of pixels corresponding to each gray scale level in the image to generate a gray scale histogram. Then, the control system starts from the minimum gray scale value to accumulate the number of pixels of each gray scale level to calculate the cumulative distribution function. By analyzing the peak value, valley value position and change trend of the cumulative distribution function of the gray scale histogram, the control system obtains the key gray scale characteristics of the image, including the gray scale distribution range of the background region and the target region, the gray scale span of the edge transition region and other information.

[0089] In actual application, there may be local strong light or shadow areas in the image, resulting in abnormal peaks in the gray scale distribution. For this purpose, the control system can adopt an adaptive histogram smoothing strategy: first, detect the abnormal fluctuations in the gray scale histogram, and pre-process the histogram through median filtering or Gaussian smoothing. Then analyze the local change trend of the smoothed histogram to identify the true gray scale distribution characteristics. For example, for the area with strong light, the system will correct the abnormal peak value by analyzing the neighborhood gray scale distribution to ensure that the obtained distribution characteristics are more reliable.

[0090] S204, determine the first layer threshold value based on the gray scale distribution characteristics, and perform initial segmentation to obtain the weld candidate region.

[0091] wherein the gray scale distribution feature represents the statistical distribution law of the image gray scale value; the first layer threshold value is used to represent the gray scale critical value for distinguishing the weld main body and the background; the initial segmentation refers to the image segmentation operation using the first layer threshold value; and the weld candidate region represents the image region that may contain the weld after the initial segmentation.

[0092] After obtaining the gray scale distribution feature, the control system needs to determine a suitable segmentation threshold value. Specifically, the control system first analyzes the bimodal feature of the gray scale histogram, wherein the higher gray scale value corresponds to the weld region, and the lower gray scale value corresponds to the background region. Then, the control system uses the OTSU algorithm to calculate the inter-class variance, and determines the optimal segmentation threshold value by maximizing the inter-class variance. In order to improve the segmentation reliability, the control system also considers the rate of change of the cumulative distribution function, and finds the position with the most steep gray scale change near the maximum point of the inter-class variance, and takes the gray scale value of the position as the first layer threshold value. The image is segmented using the threshold value to obtain the preliminary weld candidate region.

[0093] In some embodiments, the determination of the first layer threshold value can be realized in various ways: optionally, the control system can use an improved iterative threshold method, first estimate an initial threshold value, repeatedly calculate the average gray scale value of the target region and the background region, and obtain a stable segmentation threshold value through iterative optimization, and correct it in combination with the image entropy information; optionally, the control system can use a threshold selection method based on local self-adaptation, divide the image into multiple sub-regions, calculate the local optimal threshold value respectively, and obtain the global threshold value through weighted fusion. It can be understood that other ways can also be used to realize the determination of the first layer threshold value, which is not limited here.

[0094] In actual application, the reflection or oxidation of the weld surface may cause uneven gray scale distribution, affecting the accuracy of threshold selection. In this regard, the control system can introduce a threshold optimization mechanism based on morphological features: first, use multiple candidate threshold values for exploratory segmentation, analyze the morphological features of the obtained candidate region, including connectivity, boundary smoothness and region compactness, etc. Select the threshold value that can produce the most consistent weld morphological features as the final first layer threshold value. For example, when it is detected that the segmentation result appears excessive fragmentation, the system will appropriately adjust the threshold value to maintain the integrity of the weld region.

[0095] S205, calculating a second layer threshold value according to the gray scale distribution of the weld candidate region, and performing secondary segmentation to reserve the edge transition region.

[0096] wherein the weld candidate region represents the potential weld region obtained by the initial segmentation; the second layer threshold value is used to represent the gray scale critical value for reserving the edge transition feature; the secondary segmentation refers to the fine segmentation operation using the second layer threshold value; and the edge transition region represents the gradual change region between the weld main body and the background.

[0097] After obtaining the weld candidate region, the control system needs to further retain the edge detail information. Specifically, the control system first extracts the boundary of the weld candidate region, and sets a local analysis window centered on the boundary. Within each window, the control system calculates the gradient distribution and direction information of the gray value, and constructs a gray change model of the edge region. Based on the model, the control system analyzes the continuity characteristics of the gray transition to determine the optimal threshold value that can retain the edge transition characteristics. To adapt to the edge characteristics of different regions, the control system adjusts the second layer threshold value according to the local gradient intensity, while ensuring the integrity of the edge and avoiding the introduction of too much noise.

[0098] In some embodiments, the calculation of the second layer threshold value can be implemented in various ways: optionally, the control system can use an adaptive threshold method based on edge intensity, first calculate the gradient amplitude map of the candidate region boundary, determine the gradient threshold value through statistical analysis, and adjust the second layer threshold value combined with the local gray distribution characteristics; optionally, the control system can use a threshold determination method based on region growing, take the main body region of the weld as the seed region, and gradually expand to the edge transition region by controlling the gray similarity threshold of the growth condition. It can be understood that other ways can also be used to implement the calculation of the second layer threshold value, which is not limited here.

[0099] In actual application, the weld edge may have irregular changes or fuzzy regions, resulting in unstable edge feature extraction. For this purpose, the control system can use a multi-scale edge analysis strategy: through multi-scale decomposition of the image by Gaussian pyramid, the consistency of edge features at different scales is analyzed. When the edge features at a certain scale are detected to be unstable, the system will correct it by referring to the analysis results at other scales. For example, for a fuzzy edge region, the system will determine the approximate boundary position at a coarse scale, and then accurately position the edge transition region at a fine scale, and finally determine the appropriate second layer threshold value.

[0100] S206, mark the region higher than the first layer threshold value as foreground pixel value, and mark the region lower than the second layer threshold value as background pixel value, to obtain a binary image.

[0101] Wherein, the foreground pixel value represents the binary value used to mark the weld region; the background pixel value is used to represent the binary value of the non-weld region; the binary image refers to the processing result image containing only two kinds of pixel values; the first layer threshold value and the second layer threshold value respectively represent the gray critical value for distinguishing the main body of the weld and the edge transition region.

[0102] After determining the two thresholds, the control system needs to generate the final binarization result. Specifically, the control system first judges the gray value of each pixel point: for the pixel points with a gray value higher than the first layer threshold, it is determined as the main body area of the weld, and marked as a foreground pixel value (usually 255 or 1); for the pixel points with a gray value lower than the second layer threshold, it is determined as the background area, and marked as a background pixel value (usually 0). For the pixel points with a gray value between the two thresholds, the control system analyzes its spatial relationship with the marked area and the gray continuity, and dynamically determines its attribution according to the local area characteristics. Finally, the control system performs connectivity analysis on the marked result to eliminate isolated small areas and obtain a complete binarization image.

[0103] In practical applications, the transition area between the two thresholds may contain complex gray value changes, resulting in fragmentation or discontinuity of the marked result. To this end, the control system can introduce an optimization mechanism based on topological structure: first, analyze the topological features of the binarization result, including the number of connected components, area shape and boundary curvature, etc. When detecting abnormal topological structure, the system will automatically adjust the marking strategy of the intermediate area. For example, for the weld area with a broken part, the system will analyze the morphological features on both sides of the broken point and restore the area connectivity through an intelligent completion algorithm. At the same time, the system will also consider the overall trend of the weld to ensure the continuity and integrity of the marked result.

[0104] S207, set multiple detection areas in the binarization image, independently extract edge feature points in each detection area, and determine the effective detection area according to the number distribution of feature points in each detection area, and convert the edge feature points in the effective detection area into coordinate data.

[0105] Referring to step S103, the control system will extract effective feature points in multiple detection areas and convert them into coordinate data.

[0106] S208, determine the weight coefficient according to the distribution density of the edge feature points in the detection area, and perform partition weighted fitting on the coordinate data to obtain the initial weld direction.

[0107] Referring to step S104, the control system will calculate the weld direction based on the weighted fitting of the feature point distribution density.

[0108] In some embodiments, the control system determines the region weight coefficient according to the feature point density distribution, that is, the control system calculates the projection distribution of the edge feature points in the horizontal direction and the vertical direction in each detection region to obtain a two-dimensional density distribution matrix; calculates the feature point aggregation degree of each region according to the two-dimensional density distribution matrix, and determines the weight coefficient according to the feature point aggregation degree; divides the detection region along the weld direction into a plurality of equal-length sub-regions, and applies the corresponding weight coefficient to the feature point coordinates in each sub-region; performs least square fitting on all weighted feature point coordinates to obtain a spatial equation of the weld, and determines the direction vector of the spatial equation as the initial weld direction.

[0109] wherein the detection region represents a local image region for feature extraction; the edge feature point refers to a key pixel point on the weld contour; the projection distribution is used to represent the density distribution of the feature points in the horizontal and vertical directions; the two-dimensional density distribution matrix represents a numerical expression of the spatial distribution of the feature points; the feature point aggregation degree refers to the spatial concentration degree of the feature points; the weight coefficient is used to represent the importance of different regions in the fitting calculation; the sub-region refers to an equal-length local region divided along the weld direction; the spatial equation represents a mathematical expression describing the weld trajectory; and the direction vector refers to a unit vector representing the weld direction.

[0110] After the edge feature point extraction is completed, the control system needs to determine the spatial direction of the weld. Specifically, the control system first counts the number of projections of the feature points in the horizontal and vertical directions in each detection region to generate a two-dimensional matrix reflecting the density of the feature point distribution. Based on the matrix, the control system calculates the feature point aggregation degree index of each region, including local density mean, variance, and peak value, etc. statistical characteristics. According to the aggregation degree index, the control system assigns a weight coefficient to each region, and the more concentrated and uniform the feature point distribution, the higher the weight. Then, the control system divides the detection region into a plurality of sub-regions along the estimated weld direction, and applies the corresponding weight coefficient to the feature point coordinates in each sub-region for weighted transformation. Finally, the control system performs curve fitting on all weighted coordinate points using the least square method to obtain a mathematical equation describing the spatial trajectory of the weld, and unitizes the tangent vector of the equation as the initial weld direction.

[0111] In some embodiments, the spatial distribution analysis and weighted fitting of feature points can be implemented in various ways: optionally, the control system can adopt a distribution analysis method based on kernel density estimation, first select an appropriate kernel function to calculate the local density distribution of feature points, then obtain the edge distribution by integral projection, determine the weight coefficient based on the statistical characteristics of the edge distribution, and finally use the weighted RANSAC algorithm for robust fitting; optionally, the control system can adopt a direction estimation method based on principal component analysis, first perform principal component analysis on the feature point set to obtain the principal direction, then calculate the contribution of each feature point to the principal direction as the weight basis, adjust the weight distribution through iterative optimization, and finally use the weighted orthogonal distance regression method for accurate fitting. It can be understood that other ways can also be used to implement distribution analysis and curve fitting, which are not limited here.

[0112] S209, based on the initial weld direction, a dynamic error band is constructed, coordinate points outside the dynamic error band are taken as candidate abnormal points, the continuity criterion of adjacent feature points is combined to determine the final abnormal points in the candidate abnormal points, the remaining coordinate points outside the final abnormal points are re-fitted to obtain the target weld direction.

[0113] Referring to step S105, the control system will screen abnormal points through the dynamic error band and re-fit to determine the target direction.

[0114] S210, compare the target weld direction with the reference direction, and generate a hierarchical correction control signal according to the deviation degree.

[0115] Referring to step S106, the control system will generate a hierarchical control signal according to the deviation from the reference direction.

[0116] S211, obtain the contact pressure data of the robot adsorption device, and generate a wall surface contact state distribution map.

[0117] Wherein, the adsorption device represents the mechanism used by the underwater robot for wall surface adhesion; the contact pressure data is used to represent the pressure value of the contact point of the adsorption device and the wall surface; the wall surface contact state distribution map refers to a two-dimensional distribution map reflecting the contact condition of the robot and the wall surface; the pressure data acquisition point represents the arrangement position of the pressure sensor.

[0118] During the welding seam tracking process, the control system needs to monitor the contact state between the robot and the wall surface in real time. Specifically, the control system first acquires the pressure data collected by each pressure sensor of the adsorption device, processes these discrete pressure values through time averaging and spatial interpolation, and constructs a continuous pressure distribution field. Then, the control system normalizes the pressure distribution to eliminate system errors between sensors. Based on the normalized pressure data, the control system generates a pressure distribution cloud chart using the contour method and visually displays the pressure changes combined with color mapping. Finally, the control system converts the pressure distribution into discrete contact state levels through threshold segmentation and generates a wall surface contact state distribution chart.

[0119] S212, identifying the welding seam contour abnormal area according to the contact state distribution chart, and determining the welding seam structure type of the welding seam contour abnormal area.

[0120] Among them, the contact state distribution chart represents the two-dimensional distribution of the contact state between the robot and the wall surface; the welding seam contour abnormal area is used to represent the area where the contact pressure distribution changes significantly; and the welding seam structure type refers to the welding seam shape with different geometric features and pressure distribution characteristics, such as flat weld, corner weld, and lap weld.

[0121] After obtaining the contact state distribution chart, the control system needs to analyze the welding seam structure features. Specifically, the control system first performs morphological processing on the contact state distribution chart to extract the abnormal change area of the pressure distribution. Then, the control system analyzes the geometric features of the abnormal area, including area, shape, direction, and other parameters, and combines the pressure gradient distribution features in the area. Based on these feature parameters, the control system constructs a feature vector and classifies the abnormal area into different welding seam structure types through pattern recognition algorithms. For complex abnormal areas, the control system also analyzes the pressure distribution pattern of the surrounding area to improve the accuracy of structure type identification.

[0122] In some embodiments, the identification of the welding seam structure type can be achieved in various ways: optionally, the control system can use a deep learning-based structure identification method, use a pre-trained convolutional neural network to extract deep features of the pressure distribution chart, identify the welding seam structure type through a multi-layer classifier, and continuously optimize the identification model through online learning; optionally, the control system can use a rule-based reasoning structure analysis method, establish a welding seam structure feature knowledge base, determine the structure type through feature matching and logical reasoning, and verify it combined with historical data. It can be understood that other ways can also be used to implement structure type identification, which are not limited here.

[0123] In practical applications, complex weld structures can have multiple types of features simultaneously, leading to ambiguous structure type identification. To address this, the control system can employ a multi-modal feature fusion strategy: first, features are extracted from multiple dimensions such as pressure distribution, geometric shape, and temporal changes. Then, a feature importance evaluation model is established to adjust the weights based on the contribution of different features in structure identification. When there is a conflict between features, the system analyzes the reliability of each feature and prioritizes the use of high-reliability features for judgment. For example, for the weld corner area, the system considers both the pressure mutation feature and the geometric inflection point feature to comprehensively determine the structure type.

[0124] The feature importance evaluation model is used for feature fusion in weld structure type identification. In the training phase, the system collects a labeled dataset containing multi-dimensional features such as pressure distribution, geometric shape, and temporal changes, where each sample is labeled with the corresponding weld structure type. By minimizing the objective function of structure type identification error, a feature weight network based on attention mechanism is trained. This model uses a multi-channel input structure to receive different types of feature vectors, calculates the correlation between features through self-attention layers, and then dynamically allocates weight coefficients based on the recognition contribution of features. In practical applications, when new feature data is input, the model first evaluates the reliability of each feature, gives higher weights to high-reliability features, and combines the weighted results of different features to obtain the final structure type judgment. For example, at the weld corner, when both the pressure mutation feature and the geometric inflection point feature are obvious, the model will give higher weights to these two types of features, while reducing the weight of the possibly disturbed temporal feature, thereby improving the accuracy of structure identification.

[0125] S213、When the weld structure type changes, update the detection region division scheme.

[0126] wherein the weld structure type represents the current recognized weld shape feature; the detection region division scheme is used to represent the arrangement strategy of the image feature extraction region; and the update refers to adjusting the position, size, and shape of the detection region parameters according to the structure change.

[0127] When the weld structure type is identified to change, the control system needs to adjust the feature extraction strategy in a timely manner. Specifically, the control system first analyzes the impact of the new structure type on feature extraction, including weld direction change, width change, and local feature distribution. Then, the control system adjusts the parameters of the detection region according to these changes: for weld direction change, adjust the direction and arrangement of the detection region; for width change, modify the size and spacing of the detection region; for local feature distribution change, optimize the shape and density of the detection region. At the same time, the control system also establishes the association between the detection regions to ensure the continuity and integrity of the region division.

[0128] In some embodiments, dynamic updating of the detection area can be achieved through a variety of methods: Optionally, the control system can use a prediction-based regional adjustment method to predict the optimal layout of subsequent areas based on the trend of weld structure changes, and achieve gradual adjustment of the detection area through smooth transition while maintaining detection continuity; Optionally, the control system can use a feedback-based adaptive adjustment method to monitor the feature extraction effect in real time, automatically optimize regional parameters when a decrease in feature quality is detected, and find the optimal partitioning solution through iterative optimization. It is understandable that other methods can also be used to achieve detection area updates, which are not limited here.

[0129] S214. Select a corresponding correction execution strategy according to the weld structure type.

[0130] Among them, the weld structure type represents the geometric characteristics of the weld; the correction execution strategy is used to represent the control scheme for the underwater robot's posture adjustment; the corresponding relationship refers to the matching rules between different structure types and corresponding correction strategies; and the control parameters represent the specific execution parameters required to implement the correction action.

[0131] After determining the weld structure type, the control system needs to select an appropriate correction strategy. Specifically, the control system first retrieves a basic strategy template that matches the current structure type from the strategy library. Then, the control system optimizes the parameters of the basic strategy based on real-time pressure distribution data and motion status, including adjusting dynamic parameters such as correction speed, acceleration, and transition time. For complex structures, the control system decomposes the correction process into multiple sub-stages, configures an independent control strategy for each stage, and designs a smooth transition mechanism for strategy switching. Finally, the control system converts the optimized strategy into a specific execution instruction sequence and sends it to the robot's motion control unit.

[0132] In some embodiments, the control system will select an appropriate correction strategy based on the structural characteristics of the weld, that is, the control system will classify the weld structure types into straight welds, fillet welds and lap welds based on the geometric characteristics and pressure distribution characteristics of the abnormal weld contour area; when the weld structure type is a straight weld, a unidirectional progressive correction strategy is adopted, and the robot posture is gradually adjusted according to the deviation direction until it returns to the reference direction; when the weld structure type is a fillet weld, a bidirectional coordinated correction strategy is adopted, and the lateral position and steering angle of the robot are adjusted at the same time to keep the relative position of the detection area and the weld contour stable; when the weld structure type is a lap weld, a step transition correction strategy is adopted, and the correction action is suspended at the weld step, and the correction reference is re-established after the robot crosses the step.

[0133] wherein the weld contour abnormal region represents a position where the pressure distribution changes; the geometric feature is used to represent the weld shape and size parameters; the pressure distribution feature refers to a numerical feature reflecting the contact state; the weld structure type includes three basic forms of flat weld, fillet weld, and lap weld; the one-way progressive correction strategy represents a control mode of step-by-step adjustment in a single direction; the two-way coordinated correction strategy is used to represent a control mode of adjusting two degrees of freedom simultaneously; the step transition correction strategy refers to a control mode of pausing and then rebuilding the reference at a specific position; and the reference direction represents an ideal weld tracking trajectory.

[0134] Upon detecting a change in the weld structure, the control system needs to select an appropriate correction strategy. Specifically, the control system first comprehensively analyzes the geometric features and pressure distribution features of the contour abnormal region, including parameters such as region shape, area, pressure gradient, and distribution symmetry. Based on these feature parameters, the control system classifies the weld structure into three basic types. For flat welds, the control system uses a one-way progressive strategy to gradually eliminate deviations through small-step pose adjustments, while maintaining the smoothness of movement. For fillet welds, the control system enables a two-way coordinated strategy, establishing a coupled control model of lateral position and turning angle to coordinate adjustments of the two degrees of freedom to maintain the relative position stability of the detection region. For lap welds, the control system executes a step transition strategy, reducing the movement speed when approaching the step position, pausing the correction action until the step is crossed, and then re-acquiring reference data to start a new correction process.

[0135] In some embodiments, structure recognition and correction control can be implemented in various ways: optionally, the control system can use a deep reinforcement learning-based control strategy generation method to establish a state space model containing weld features, robot state, and environmental factors, output the optimal control action through a policy network, introduce a reward mechanism to optimize control effect, and continuously update the policy network according to the execution results; optionally, the control system can use an adaptive control method based on model prediction to build a dynamics model for weld tracking, optimize the control sequence in a rolling time domain, and continuously update the model parameters through online parameter identification to achieve adaptive control of different structure types. It can be understood that other ways can also be used to implement structure adaptive control, which are not limited here.

[0136] In practical applications, the transition region of weld structures may contain complex geometric features, leading to instability in structural type determination and strategy switching. To address this, the control system can adopt a hybrid control architecture: first, a state-space representation of multiple basic control strategies is established, and a smooth transition mechanism for strategy switching is designed. When fuzzy structural features or the coexistence of multiple features is detected, the system simultaneously activates multiple related control strategies and generates the final control output through a weighted combination. For example, in the region where a straight weld transitions to a fillet weld, the system gradually reduces the weight of the unidirectional strategy while increasing the weight of the bidirectional strategy to ensure the continuity and stability of the control action. At the same time, the system adjusts the strategy weights based on the execution results to optimize control performance.

[0137] In the embodiments of the present application, due to the use of dynamic filtering parameter adjustment based on local noise intensity, double-layer threshold segmentation, independent feature extraction of multiple detection areas, weighted fitting based on distribution density, dynamic error band outlier screening, as well as posture compensation, water flow analysis and structural adaptation and other technical means, it is possible to achieve adaptive optimization of image processing parameters, complete extraction of weld features, effective isolation of local interference, outstanding contribution of key areas, accurate identification of outliers, active compensation of environmental disturbances and intelligent adaptation of complex structures, effectively solving the problems existing in the prior art such as excessive smoothing caused by fixed parameters, feature loss caused by a single threshold, susceptibility of global features to interference, poor adaptability to environmental changes, and insufficient ability to handle complex structures, thereby achieving high-precision, high-reliability and strong adaptability of weld tracking control in underwater environments, and improving the operation quality and efficiency of underwater welding robots.

[0138] The following describes the control system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the structure of a physical device of the control system in an embodiment of the present application.

[0139] It should be noted that Figure 3 The structure of the control system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0140] like Figure 3 As shown, the control system includes a CPU 301, which can perform various appropriate actions and processes according to the programs stored in the ROM 302 or the programs loaded from the storage unit 308 into the RAM 303, such as executing the methods described in the above embodiments. The RAM 303 also stores various programs and data required for system operation. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An I / O interface 305 is also connected to the bus 304.

[0141] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a push button switch, and the like; an output section 307 including a Liquid Crystal Display (LCD), an audio output device, an indicator lamp, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as necessary. A removable medium 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is attached to the drive 310 as necessary, so that a computer program read out from it is installed in the storage section 308 as necessary.

[0142] In particular, the processes described above with reference to the flow charts can be implemented as a computer software program according to embodiments of the present application. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing computer programs for executing the methods shown in the flow charts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by the CPU 301, various functions defined in the present application are performed.

[0143] The flow charts and block diagrams in the drawings are schematic illustrations of possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flow charts or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures.

[0144] In particular, the control system of the embodiment includes a processor and a memory, and the memory stores a computer program which, when executed by the processor, implements the automatic deviation correction method for the underwater robot provided by the above-described embodiment.

[0145] As another aspect, the present application also provides a computer readable storage medium which can be included in the control system described in the above-described embodiment, or can exist separately without being assembled into the control system. The storage medium carries one or more computer programs which, when executed by a processor of the control system, cause the control system to implement the automatic deviation correction method for the underwater robot provided by the above-described embodiment.

[0146] The above-described examples are merely intended to illustrate the technical solutions of the present application, but not to limit the same; even though the present application has been described in detail with reference to the foregoing examples, those of ordinary skill in the art should understand that modifications can still be made to the technical solutions recorded in the foregoing examples, or equivalent replacements can be made to some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the examples of the present application.

[0147] In the above-described examples, the term "when" can be interpreted to mean "if" or "after" or "in response to determining" or "in response to detecting" depending on the context. Similarly, the phrase "on determining" or "if detecting (a stated condition or event)" can be interpreted to mean "if determining" or "in response to determining" or "on detecting (a stated condition or event)" or "in response to detecting (a stated condition or event)" depending on the context.

Claims

1. A method for automatically correcting the deviation of an underwater robot, characterized in that: Applied to a control system, the method includes: Acquire a weld image captured by an underwater robot camera, and adjust a filter parameter according to the noise intensity of a local image in the weld image to obtain a de-noised image; Performing double-layer threshold segmentation on the denoised image to obtain a binary image; the first layer threshold of the double-layer threshold is used to extract the main contour of the weld, and the second layer threshold of the double-layer threshold is used to retain the characteristics of the transition area of ​​the weld edge; Setting a plurality of detection areas in the binary image, independently extracting edge feature points in each detection area, determining an effective detection area based on the number distribution of feature points in each detection area, and converting the edge feature points in the effective detection area into coordinate data; Determine the weight coefficient according to the distribution density of the edge feature points in the detection area, perform partition weighted fitting on the coordinate data, and obtain the initial weld direction; the step of determining the weight coefficient according to the distribution density of the edge feature points in the detection area, performing partition weighted fitting on the coordinate data, and obtaining the initial weld direction specifically includes: calculating the projection distribution of the edge feature points in the horizontal and vertical directions in each detection area to obtain a two-dimensional density distribution matrix; calculating the clustering degree of the feature points in each area according to the two-dimensional density distribution matrix, and determining the weight coefficient according to the clustering degree of the feature points; dividing the detection area into several sub-areas of equal length along the direction of the weld, and applying the corresponding weight coefficient to weight the feature point coordinates in each sub-area; performing least squares fitting on all weighted feature point coordinates to obtain the spatial equation of the weld, and determining the direction vector of the spatial equation as the initial weld direction; A dynamic error band is constructed based on the initial weld direction, coordinate points outside the dynamic error band are used as candidate outlier points, a final outlier point among the candidate outlier points is determined in combination with a continuity criterion of adjacent feature points, and the remaining coordinate points outside the final outlier point are refitted to obtain a target weld direction; The target weld direction is compared with the reference direction, and a graded deviation correction control signal is generated according to the degree of deviation.

2. The method according to claim 1, characterized in that The step of performing double-layer threshold segmentation on the denoised image to obtain a binary image specifically includes: Performing grayscale normalization processing on the denoised image to obtain a standard grayscale image; Calculating a grayscale histogram and a cumulative distribution function of the standard grayscale image to obtain grayscale distribution characteristics; Determining a first layer threshold based on the grayscale distribution characteristics, performing a primary segmentation to obtain a weld candidate area; Calculating a second layer threshold according to the grayscale distribution of the weld candidate area, performing secondary segmentation and retaining the edge transition area; The area with a value higher than the first layer threshold is marked as a foreground pixel value, and the area with a value lower than the second layer threshold is marked as a background pixel value, thereby obtaining a binary image.

3. The method according to claim 1, characterized in that After the step of obtaining a weld image captured by a camera of an underwater robot and adjusting a filter parameter according to the noise intensity of a local image in the weld image to obtain a de-noised image, the method further includes: Obtain the posture data collected by the underwater robot's posture sensor and establish the spatial transformation matrix between the image plane and the weld plane; Performing perspective correction on the weld image according to the spatial transformation matrix to eliminate viewing angle deviation; Extracting the moving bubble trajectory in the weld image, establishing the bubble motion vector field, and determining the direction of water flow disturbance; The image sampling timing is adjusted according to the direction of the water flow disturbance to avoid the peak moment of the water flow disturbance.

4. The method according to claim 3, characterized in that The steps of extracting the moving bubble trajectory in the weld image, establishing the bubble motion vector field, and determining the direction of the water flow disturbance specifically include: Perform inter-frame difference calculation on the continuously acquired multi-frame images to extract the position change of the moving target, and filter the target according to its area and grayscale distribution characteristics to obtain the bubble target; Establishing a time-series tracking sequence for the bubble targets, calculating the motion speed and acceleration of each bubble target, and constructing a local bubble motion vector; Perform spatial interpolation on the image plane for each local bubble motion vector to obtain a continuous bubble motion vector field distribution; Cluster analysis is performed on the bubble motion vector field to extract the direction of water flow disturbance, and the periodic parameters of the water flow disturbance are identified based on the gradient change of the vector field.

5. The method according to claim 1, wherein After the step of comparing the target weld direction with the reference direction and generating a graded deviation correction control signal according to the degree of deviation, the method further includes: Obtain contact pressure data of the robot adsorption device and generate a wall contact state distribution map; Identify the weld contour abnormal area according to the contact state distribution map, and determine the weld structure type of the weld contour abnormal area; When the weld structure type changes, updating the detection area division scheme; Select the corresponding correction execution strategy according to the weld structure type.

6. The method according to claim 5, characterized in that The step of selecting a corresponding correction execution strategy according to the weld structure type specifically includes: According to the geometric characteristics and pressure distribution characteristics of the abnormal weld contour area, the weld structure type is classified into straight weld, fillet weld and lap weld; When the weld structure type is a straight weld, a one-way progressive deviation correction strategy is adopted to gradually adjust the robot posture according to the deviation direction until it returns to the reference direction; When the weld structure type is a fillet weld, a two-way coordinated correction strategy is adopted to simultaneously adjust the lateral position and steering angle of the robot to maintain a stable relative position between the detection area and the weld contour; When the weld structure type is a lap weld, a step transition correction strategy is adopted, and the correction action is suspended at the weld step, and the correction reference is re-established after the robot crosses the step.

7. A control system, characterized in that: The control system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the control system to execute the method as described in any one of claims 1-6.

8. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a control system, the control system is caused to execute the method according to any one of claims 1 to 6.

9. A computer program product, characterized in that When the computer program product is run on a control system, the control system is caused to execute the method according to any one of claims 1 to 6.

Images