Machine vision-based v-groove cutting precision detection method and system

Abstract

The present application relates to the technical field of machine vision detection, in particular to a V-groove cutting precision detection method and system based on machine vision; the method comprises the following steps: obtaining V-groove design data, establishing a workpiece coordinate system and a V-groove segment coordinate axis, collecting a cutting transient image sequence and a grooving result image and generating a detection segment sequence; extracting cutting action data and grooving residual data, forming opposite records in the same detection segment and generating process residual difference segments; performing directional perturbation on the object to be verified, collecting a perturbation response image sequence and generating directional perturbation response data; generating difference segment verification status and detection results based on the opposite record type, the object to be verified and the directional perturbation response data. The present application realizes the segmental positioning of V-groove cutting abnormalities, process result correspondence and defect property verification.

Description

Technical Field

[0001] This invention relates to the field of machine vision inspection technology, and specifically to a machine vision-based method and system for detecting the accuracy of V-groove cutting. Background Technology

[0002] Machine vision inspection technology has been widely used in industrial processing quality inspection scenarios. Through image acquisition, coordinate calibration, image segmentation, boundary recognition, and feature matching, it identifies the shape, size, edge contour, and local defects of the workpiece surface. For V-groove cutting, V-grooves are usually characterized by narrow groove openings, inclined groove walls, indistinct groove bottom features, and continuous extension along the groove length. During the inspection process, it is necessary to identify the groove opening position, groove shoulder boundary, groove bottom formation state, local chipping, residues, and cutting offset to determine the groove quality and processing accuracy after V-groove cutting. In the actual process of detecting the cutting accuracy of V-groove based on machine vision, there are still problems such as difficulty in distinguishing the abnormal nature of the V-groove grooving result image and difficulty in accurately corresponding the abnormality of the cutting process with the abnormality of the grooving result. Specifically, after the V-groove is cut, phenomena such as discontinuity at the bottom of the groove, outward expansion of the groove shoulder, burr residue, chip adhesion, and surface reflection are easily presented as similar local anomalies in the grooving result image, making it difficult for the detection system to distinguish between real grooving defects, loose residues, and image pseudo-responses, which can easily lead to misjudgment of abnormality types such as shallow cutting, edge chipping, and residual occlusion. On the other hand, the V-groove is cut continuously along the groove length direction. Tool deviation, repeated cutting, insufficient cutting in some areas, or local residue usually only occur in a few segments. It is difficult to confirm whether a defect is ultimately formed based solely on the cutting process image, and it is difficult to trace the corresponding cutting action state based solely on the grooving result image, resulting in inaccurate judgment of the positional correspondence and abnormality nature of local anomalies. Summary of the Invention

[0003] The purpose of this invention is to provide a machine vision-based method and system for detecting the accuracy of V-groove cutting, in order to solve the problems mentioned in the background art, such as the difficulty in distinguishing abnormal properties in the V-groove grooving result image and the difficulty in accurately corresponding abnormalities in the cutting process with abnormalities in the grooving result.

[0004] To achieve the above objectives, the technical solution of the present invention is: a machine vision-based method for detecting the accuracy of V-groove cutting, comprising: S1. Obtain the V-groove design data of the workpiece to be tested, and establish the workpiece coordinate system and V-groove segment coordinate axis based on the V-groove design data; during the V-groove cutting process, acquire the instantaneous image sequence of cutting, and after the V-groove cutting is completed, acquire the groove result image, map the instantaneous image sequence of cutting and the groove result image to the workpiece coordinate system, and divide the segments along the V-groove segment coordinate axis to generate the detection segment sequence. The V-groove design data includes the V-groove design centerline and V-groove design tolerance; the detection segment sequence is a set of fragmented image data formed by dividing along the V-groove segment coordinate axis according to a preset segment length, including detection segments arranged according to the V-groove extension direction; S2. Based on the detection segment sequence, extract cutting action data from the instantaneous cutting image sequence and extract grooving residual data from the grooving result image; perform segment correspondence between the cutting action data and grooving residual data within the same detection segment; generate process residual difference segment when the cutting action data and grooving residual data form opposite records; and write the segment number, opposite record type, and object to be verified into the process residual difference segment. Among them, the opposite record is a record in which the cutting action data and the grooved residual data within the same detection segment are inconsistent for the same groove-shaped part; the process residual difference segment is a data segment that records the opposite record type and its corresponding detection segment position; the object to be verified is the perturbation target area determined by the opposite record type; S3. Determine the perturbation application location and perturbation direction based on the object to be verified, and determine the perturbation execution parameters based on the V-groove design tolerance; perform directional perturbation on the object to be verified and collect the perturbation response image sequence, extract the response verification data corresponding to the object to be verified from the perturbation response image sequence, and generate directional perturbation response data; Among them, directional perturbation is a controlled external action applied to the object to be verified; response verification data is data that records the boundary preservation state, position preservation state, residue movement state, and local response difference state of the object to be verified before and after directional perturbation; directional perturbation response data is data that records the image response state of the object to be verified after directional perturbation. S4. Generate the difference segment verification status based on the opposite record type, the object to be verified, and the directional perturbation response data in the process residual difference segment, and generate the V-groove cutting accuracy detection result based on the difference segment verification status; Among them, the difference fragment verification state is the fragment state after the residual difference fragments in the process have been verified by the directional perturbation response data.

[0005] Preferably, in step S1, the workpiece coordinate system is a spatial reference coordinate system established based on the positioning boundary of the workpiece to be measured, used to uniformly represent the positions of the V-groove design data, the cutting instantaneous image sequence, and the grooving result image on the workpiece to be measured; the V-groove segment coordinate axis is an axial coordinate established based on the position of the V-groove design centerline in the workpiece coordinate system, used to determine the V-groove extension direction and the arrangement direction of the detection segments; the cutting instantaneous image sequence is process image data acquired in chronological order during the V-groove cutting process, and the grooving result image is the result image data acquired after the V-groove cutting is completed; the cutting instantaneous image sequence and the grooving result image are mapped to the workpiece coordinate system, and then along the V-groove segment coordinate axis... The segment division specifically includes: mapping the V-groove cutting area in the instantaneous cutting image sequence to the workpiece coordinate system to obtain the process image coordinate region; mapping the V-groove grooving area in the grooving result image to the workpiece coordinate system to obtain the result image coordinate region; dividing the process image coordinate region and the result image coordinate region into co-position segments based on the V-groove segment coordinate axis to generate a detection segment sequence; the detection segment sequence is used to establish the segment correspondence between the instantaneous cutting image sequence and the grooving result image, and each detection segment in the detection segment sequence includes a segment number, a longitudinal segment range along the V-groove segment coordinate axis, a transverse coordinate range perpendicular to the V-groove segment coordinate axis, and a segment sorting identifier.

[0006] Preferably, in step S2, the cutting action data is process visual data extracted from the instantaneous cutting image sequence within the corresponding detection segment, used to characterize the action state of the tool relative to the detection segment during V-groove cutting. The cutting action data includes tool action data, cutting response data, and action timing data. The grooving residue data is result visual data extracted from the grooving result image within the corresponding detection segment, used to characterize the grooving residue state within the detection segment after V-groove cutting is completed. The grooving residue data includes groove opening residue data, groove shoulder residue data, groove bottom residue data, and defect residue data.

[0007] Preferably, in step S2, segment correspondence is based on the segment number, longitudinal segment range, and lateral coordinate range of the detection segment. This involves classifying the cutting action data and grooving residue data into the same detection segment for comparison within the same grooving region. Specifically, segment correspondence between the cutting action data and grooving residue data within the same detection segment includes: reading the segment number, longitudinal segment range, and lateral coordinate range of the detection segment; classifying the cutting action data within the longitudinal segment range and lateral coordinate range into the corresponding detection segment; and classifying the grooving residue data within the same longitudinal segment range and lateral coordinate range... Data is assigned to the corresponding detection segment; cutting action data and grooving residue data assigned to the same detection segment are matched according to the groove opening, groove shoulder, and groove bottom positions; the opposite record is used to mark the inconsistency between the cutting action data and grooving residue data formed in the same groove shape position, and the opposite record includes action-not-manifested record, residue-expanded record, repeated action record, and biased action record; the process residue difference segment is used to record the position of the opposite record in the detection segment sequence, the type of opposite record, and the object direction of subsequent directional perturbation, and the process residue difference segment is consistent with the segment number, vertical segment range, and horizontal coordinate range of the corresponding detection segment.

[0008] Preferably, in S2, the segment number is the identification data of the detection segment in the detection segment sequence, used to establish the segment reference relationship between the cutting action data, the trenching residue data, the process residue difference segment, and the directional perturbation response data; the opposite record type is a field characterizing the category to which the opposite record belongs, used to determine the object to be verified; the object to be verified is the perturbation target area corresponding to the opposite record type, used to determine the application position and perturbation direction of subsequent directional perturbations; the correspondence between the opposite record type and the object to be verified is as follows: the object to be verified corresponding to the non-manifest action type record is the trench bottom formation area, the object to be verified corresponding to the residual expansion type record is the trench shoulder expansion area, the object to be verified corresponding to the repeated action type record is the bottom line candidate area, and the object to be verified corresponding to the biased action type record is the left and right trench shoulder response areas; the area range of the object to be verified is jointly determined by the position range of the corresponding detection segment and the opposite record type.

[0009] Preferably, in S3, the perturbation application position is the target action position of the object to be verified in the workpiece coordinate system; the perturbation direction is the action direction determined relative to the groove part where the object to be verified is located; the perturbation execution parameters are a set of parameters that control the range and degree of action of the directional perturbation, used to make the directional perturbation act on the object to be verified within the perturbation execution upper limit. The perturbation execution parameters include the perturbation application position, perturbation direction, perturbation amplitude, perturbation duration, perturbation number, and perturbation range; determining the perturbation execution parameters based on the V-groove design tolerance specifically includes: determining the perturbation execution upper limit according to the V-groove design tolerance, constraining the perturbation amplitude, perturbation duration, perturbation number, and perturbation range based on the perturbation execution upper limit, and generating perturbation execution parameters corresponding to the object to be verified.

[0010] Preferably, in step S3, directional perturbation is a process of applying a controlled external force to the object to be verified and acquiring image responses, used to create visual response differences in the trench structure state, residue state, and image response state corresponding to the object to be verified; directional perturbation of the object to be verified specifically includes: perturbing the trench bottom formation area in the trench normal direction according to perturbation execution parameters; perturbing the trench shoulder expansion area in the trench shoulder lateral direction according to perturbation execution parameters; perturbing the bottom line candidate area in the trench bottom direction according to perturbation execution parameters; and perturbing the left and right trench shoulder response areas in the transverse alternating direction according to perturbation execution parameters; the perturbation response image sequence is an image sequence acquired during the directional perturbation process, used to record the image response state of the object to be verified before and after the directional perturbation; the perturbation response image sequence is consistent with the segment number of the corresponding process residue difference segment.

[0011] Preferably, in step S3, the response verification data corresponding to the object to be verified is the state data of the object to be verified extracted from the perturbation response image sequence, used to characterize the visual response change of the object to be verified before and after directional perturbation; extracting the response verification data corresponding to the object to be verified from the perturbation response image sequence specifically includes: extracting the boundary preservation state and position preservation state from the perturbation response image sequence corresponding to the trench bottom formation area; extracting the boundary preservation state and residue movement state from the perturbation response image sequence corresponding to the trench shoulder expansion area; extracting the position preservation state from the perturbation response image sequence corresponding to the bottom line candidate area; extracting the local response difference state from the perturbation response image sequence corresponding to the left and right trench shoulder response areas; the directional perturbation response data is used to record the correspondence between the response verification data and the object to be verified, and is mapped to the process residue difference segment according to the segment number; the directional perturbation response data includes the segment number, the object to be verified, the response verification data, and the response data identifier.

[0012] Preferably, in step S4, the difference segment verification status is a status identifier characterizing the opposite record type in the residual difference segment of the process after verification by directional perturbation response data. It is used to determine the validity of the opposite record in the corresponding detection segment. The difference segment verification status includes shallow cut confirmation status, edge expansion confirmation status, loose residue reversal status, repeated cutting confirmation status, off-cut confirmation status, and difference exclusion status. The difference segment verification status is generated based on the opposite record type in the residual difference segment of the process, the object to be verified, and the directional perturbation response data. Specifically, this includes: matching the opposite record type, the object to be verified, and the directional perturbation response data; determining the validity status, exclusion status, or reversal status of the opposite record type based on the matching result; and generating the corresponding difference segment verification status based on the validity status, exclusion status, or reversal status. The V-groove cutting accuracy detection result is cutting accuracy status data formed based on the difference segment verification status of the corresponding detection segment of the residual difference segment of the process. It includes the segment number of the corresponding detection segment in the detection segment sequence, the difference segment verification status, the longitudinal segment range, and the transverse coordinate range, and is arranged according to the segment sorting identifier in the detection segment sequence.

[0013] On the other hand, the present invention provides a machine vision-based V-groove cutting accuracy detection system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the machine vision-based V-groove cutting accuracy detection method described above.

[0014] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects: 1. In this invention, based on the segment correspondence between the instantaneous image sequence of cutting and the resulting image of grooving, the position of the cutting tool and the state of the groove after grooving can be compared in the same position, thereby more accurately determining the segment where the V-groove local anomaly is located and its source of anomaly. 2. In this invention, the object to be verified is determined by the opposite recording type, and the object to be verified is subjected to directional perturbation, which can distinguish between real trenching defects, loose residues and image pseudo-responses, thereby reducing the risk of misjudgment of anomalies such as shallow cutting, edge breakage, and residual occlusion. Attached Figure Description

[0015] Figure 1 This is a flowchart of one embodiment of the present invention. Detailed Implementation

[0016] Example 1, as Figure 1 As shown, the specific implementation steps of the V-groove cutting accuracy detection method based on machine vision proposed in this invention are as follows: S1. Obtain the V-groove design data of the workpiece to be tested, and establish the workpiece coordinate system and V-groove segment coordinate axis based on the V-groove design data; during the V-groove cutting process, acquire the instantaneous image sequence of cutting, and after the V-groove cutting is completed, acquire the groove result image, map the instantaneous image sequence of cutting and the groove result image to the workpiece coordinate system, and divide the segments along the V-groove segment coordinate axis to generate the detection segment sequence. The V-groove design data includes the V-groove design centerline and V-groove design tolerance; the detection segment sequence is a set of fragmented image data formed by dividing along the V-groove segment coordinate axis according to a preset segment length, including detection segments arranged according to the V-groove extension direction; S2. Based on the detection segment sequence, extract cutting action data from the instantaneous cutting image sequence and extract grooving residual data from the grooving result image; perform segment correspondence between the cutting action data and grooving residual data within the same detection segment; generate process residual difference segment when the cutting action data and grooving residual data form opposite records; and write the segment number, opposite record type, and object to be verified into the process residual difference segment. Among them, the opposite record is a record in which the cutting action data and the grooved residual data within the same detection segment are inconsistent for the same groove-shaped part; the process residual difference segment is a data segment that records the opposite record type and its corresponding detection segment position; the object to be verified is the perturbation target area determined by the opposite record type; S3. Determine the perturbation application location and perturbation direction based on the object to be verified, and determine the perturbation execution parameters based on the V-groove design tolerance; perform directional perturbation on the object to be verified and collect the perturbation response image sequence, extract the response verification data corresponding to the object to be verified from the perturbation response image sequence, and generate directional perturbation response data; Among them, directional perturbation is a controlled external action applied to the object to be verified; response verification data is data that records the boundary preservation state, position preservation state, residue movement state, and local response difference state of the object to be verified before and after directional perturbation; directional perturbation response data is data that records the image response state of the object to be verified after directional perturbation. S4. Generate the difference segment verification status based on the opposite record type, the object to be verified, and the directional perturbation response data in the process residual difference segment, and generate the V-groove cutting accuracy detection result based on the difference segment verification status; Among them, the difference fragment verification state is the fragment state after the residual difference fragments in the process have been verified by the directional perturbation response data.

[0017] In this embodiment S1, the workpiece coordinate system is a spatial reference coordinate system established based on the positioning boundary of the workpiece to be measured, used to uniformly represent the positions of the V-groove design data, the cutting instantaneous image sequence, and the grooving result image on the workpiece to be measured; the V-groove segment coordinate axis is an axial coordinate established based on the position of the V-groove design center line in the workpiece coordinate system, used to determine the V-groove extension direction and the arrangement direction of the detection segments; the cutting instantaneous image sequence is process image data collected in chronological order during the V-groove cutting process, and the grooving result image is the result image data collected after the V-groove cutting is completed; mapping the cutting instantaneous image sequence and the grooving result image to the workpiece coordinate system, and dividing the segments along the V-groove segment coordinate axis, specifically includes: mapping the V-groove cutting area in the cutting instantaneous image sequence to the workpiece coordinate system to obtain the process image coordinate area; and dividing the segment into segments along the V-groove segment coordinate axis. The V-groove grooving region in the grooving result image is mapped to the workpiece coordinate system to obtain the result image coordinate region. Based on the V-groove segment coordinate axis, the process image coordinate region and the result image coordinate region are divided into corresponding segments to generate a detection segment sequence. The detection segment sequence is used to establish the segment correspondence between the cutting instantaneous image sequence and the grooving result image. Each detection segment in the detection segment sequence includes a segment number, a longitudinal segment range along the V-groove segment coordinate axis, a transverse coordinate range perpendicular to the V-groove segment coordinate axis, and a segment sorting identifier. Among them, the segment number is used to identify the detection segment, the longitudinal segment range is used to determine the position of the detection segment in the V-groove extension direction, the transverse coordinate range is used to determine the coverage area of ​​the detection segment in the V-groove width direction, and the segment sorting identifier is used to determine the arrangement order of the detection segments in the detection segment sequence.

[0018] In this embodiment S1, a V-groove refers to a groove structure with a V-shaped cross-section formed by cutting along a preset extension direction on the surface of the workpiece to be tested. The workpiece to be tested can be a metal sheet, a composite material sheet, or a workpiece with a surface to be cut. The V-groove design data is a set of data characterizing the theoretical machining position and allowable machining deviation of the V-groove. When acquiring the V-groove design data of the workpiece to be tested, the V-groove parameters in the corresponding machining task file, CNC machining program, workpiece design model, and inspection configuration file are read, and the data related to the theoretical direction, machining position, and accuracy requirements of the V-groove are extracted as V-groove design data; V-groove design center The line represents the theoretical extension position of the V-groove on the surface of the workpiece to be measured. It is determined by the starting position, ending position, and intermediate path points of the V-groove and is used to determine the extension direction of the V-groove on the workpiece to be measured. The V-groove design tolerance is the allowable range of dimensional deviation, positional deviation, and morphological deviation after the V-groove is cut. It is used as a process constraint for subsequent groove formation state determination and perturbation execution parameter determination. After obtaining the V-groove design data, the coordinate analysis of the V-groove design centerline is performed to obtain the centerline position of the V-groove in the design coordinates of the workpiece to be measured. The V-groove design tolerance is associated with and stored with the corresponding groove opening, groove shoulder, groove bottom, and groove length direction position.

[0019] In this embodiment S1, when establishing the workpiece coordinate system based on the V-groove design data, the positioning boundary, positioning hole, fixture datum, and machining datum of the workpiece to be tested are selected as the basis for coordinate establishment. The reference position of the workpiece to be tested in the image acquisition field of view is determined through visual recognition, boundary fitting, feature point positioning, and workpiece registration processing, and a spatial reference coordinate system is established based on this reference position. The workpiece coordinate system is used to uniformly represent the position of the V-groove design data, the instantaneous image sequence of cutting, and the grooving result image on the workpiece to be tested. Its coordinate axis direction is consistent with the length direction, width direction, and surface normal direction of the workpiece to be tested. The direction corresponds; when establishing the V-groove segment coordinate axis based on the position of the V-groove design center line in the workpiece coordinate system, the V-groove design center line is projected onto the workpiece coordinate system, and an axial coordinate is established along the extension direction of the V-groove design center line. The direction perpendicular to this axial coordinate and covering the width direction of the V-groove is used as the lateral coordinate, thereby forming the V-groove segment coordinate axis for segment division; the establishment of the workpiece coordinate system and the V-groove segment coordinate axis can be achieved through camera calibration, workpiece positioning, coordinate transformation, image registration, and design model matching processing, so that the design datum, process image, and result image have a unified positional expression.

[0020] In this embodiment S1, when acquiring the instantaneous image sequence during V-groove cutting, the acquisition area of ​​the image acquisition device is set to cover the V-groove cutting path and its adjacent area. Multiple frames of process images are acquired according to the preset acquisition sequence, cutting trigger signal, and tool position trigger signal. The multiple frames of process images arranged in the acquisition time sequence constitute the instantaneous image sequence of cutting. The instantaneous image sequence of cutting is used to record the action position of the tool on the surface of the workpiece to be tested during V-groove cutting, the visual response when the material is cut, and the time sequence of the cutting action passing through different detection positions. Each frame of the instantaneous image sequence of cutting carries an acquisition time identifier, a frame number, and a corresponding acquisition position identifier. When acquiring the grooving result image after V-groove cutting is completed, the image acquisition device is aligned with the V-groove grooving area to acquire result image data including the groove opening, groove shoulder, groove bottom, and adjacent workpiece surfaces. The grooving result image is used to represent the groove shape state, boundary state, and surface residue state after V-groove cutting is completed. The grooving result image can be formed by registering and fusing single acquisition images, multi-angle acquisition images, and multi-light acquisition images, and the grooving area in the workpiece coordinate system is used as the basis for its position expression.

[0021] In this embodiment S1, when mapping the instantaneous cutting image sequence and the grooving result image to the workpiece coordinate system, the correspondence between the image coordinates and the workpiece coordinates is first determined based on the calibration relationship of the image acquisition device, the positioning boundary of the workpiece to be measured, and the workpiece reference features. Then, the V-groove cutting area containing the tool action position and cutting path in the instantaneous cutting image sequence is converted into the process image coordinate area in the workpiece coordinate system, and the V-groove grooving area containing the groove opening, groove shoulder, groove bottom, and adjacent boundaries in the grooving result image is converted into the result image coordinate area in the workpiece coordinate system. The process image coordinate area is the coordinate coverage range of the V-groove cutting area in the instantaneous cutting image sequence in the workpiece coordinate system, and the result image coordinate area is the coordinate coverage range of the V-groove grooving area in the grooving result image in the workpiece coordinate system. During the mapping process, the relative positions of the V-groove design center line, the process image coordinate area, and the result image coordinate area in the workpiece coordinate system are kept consistent, so that the process image position formed during the cutting process and the grooving result position after the cutting are assigned according to the same workpiece position. The mapping process can be achieved through reference point registration, boundary registration, image calibration parameter conversion, and design center line matching.

[0022] In this embodiment S1, when segmenting along the V-groove segment coordinate axis, continuous segment boundaries are set in the axial direction of the V-groove segment coordinate axis based on a preset segment length, and the lateral coordinate range of each segment in the direction perpendicular to the V-groove segment coordinate axis is determined according to the lateral coverage range corresponding to the V-groove design center line; the preset segment length is determined based on the V-groove design length, image resolution, detection accuracy requirements, and cutting cycle time; when performing co-position segmentation of the process image coordinate region and the result image coordinate region, the same set of axial segment boundaries and lateral coordinate ranges are used to divide the process image coordinate region and the result image coordinate region respectively, so that the process image data and result image data corresponding to the same segment number are... They have the same vertical segment range and horizontal coordinate range; the detection segment sequence formed by the segmentation of the same position segment is a fragmented image data set, which is used to establish the segment correspondence between the instantaneous image sequence of cutting and the resulting image of the groove; each detection segment in the detection segment sequence includes a segment number, a vertical segment range along the V-groove segment coordinate axis, a horizontal coordinate range perpendicular to the V-groove segment coordinate axis, and a segment sorting identifier, wherein the segment number is used to identify the detection segment, the vertical segment range is used to determine the position of the detection segment in the V-groove extension direction, the horizontal coordinate range is used to determine the coverage area of ​​the detection segment in the V-groove width direction, and the segment sorting identifier is used to determine the arrangement order of the detection segments in the detection segment sequence.

[0023] In this embodiment S2, the cutting action data is the process visual data extracted from the instantaneous cutting image sequence within the corresponding detection segment, used to characterize the action state of the tool relative to the detection segment during V-groove cutting. The cutting action data includes tool action data, cutting response data, and action timing data. The grooving residue data is the result visual data extracted from the grooving result image within the corresponding detection segment, used to characterize the grooving residue state within the detection segment after V-groove cutting. The grooving residue data includes groove opening residue data, groove shoulder residue data, groove bottom residue data, and defect residue data. Among them, the tool action data is used to characterize the action range of the tool within the detection segment, the cutting response data is used to characterize the visual response state formed when the material is cut within the detection segment, and the action timing data is used to characterize the time sequence of the tool action within the detection segment. The groove opening residue data is used to characterize the residue state at the groove opening position of the V-groove, the groove shoulder residue data is used to characterize the residue state at the groove shoulder positions on both sides of the V-groove, the groove bottom residue data is used to characterize the formation state at the groove bottom position of the V-groove, and the defect residue data is used to indicate the defect state outside the boundary of the V-groove.

[0024] In this embodiment S2, when extracting cutting action data based on the detection segment sequence, the segment number, longitudinal segment range, and transverse coordinate range of each detection segment in the detection segment sequence are read sequentially, and the process image region falling within the range of the detection segment is determined in the instantaneous cutting image sequence. Target region localization, tool proximity region identification, cutting path boundary identification, frame sequence tracking, and visual response extraction processing are performed on the process image region to obtain cutting action data associated with the corresponding detection segment. The cutting action data is the process visual data extracted from the instantaneous cutting image sequence within the corresponding detection segment, including tool action data, cutting response data, and action timing data. Tool action data is used to characterize the tool's action range, action position, and action trajectory within the detection segment. Cutting response data is used to characterize the brightness change, texture change, cutting boundary change, and chip movement response formed by the workpiece material under the action of the tool. Action timing data is used to characterize the time sequence of the tool entering, passing through, and leaving the detection segment, which can be determined by the image frame number, acquisition time identifier, tool position trigger identifier, and cutting execution order.

[0025] In this embodiment S2, when extracting residual data of grooving based on the detection segment sequence, the coordinate region of the grooving result image in the workpiece coordinate system is read, and the grooving result image region within the corresponding detection segment is determined according to the longitudinal segment range and the transverse coordinate range of the detection segment; the grooving result image region is processed for groove shape region positioning, groove opening boundary recognition, groove shoulder region recognition, groove bottom region recognition, residue region recognition, and outer boundary defect region recognition to obtain the residual data of grooving associated with the corresponding detection segment; the residual data of grooving is obtained by the grooving result image in the corresponding detection segment. The extracted visual data within the segment includes residual data at the groove opening, shoulder, bottom, and defects. The residual data at the groove opening is used to characterize the boundary continuity, local residual state, and groove offset state of the groove opening. The residual data at the shoulder is used to characterize the residual area, residual width, and boundary state of the shoulder positions on both sides of the groove. The residual data at the bottom is used to characterize the visible state, discontinuous state, bottom line candidate state, and groove bottom formation state of the bottom position of the groove. The residual data at the defects is used to characterize the chipped edge area, material missing area, and outward defect area outside the boundary of the groove.

[0026] In this embodiment S2, segment correspondence is based on the segment number, longitudinal segment range, and lateral coordinate range of the detection segment. It involves classifying cutting action data and grooving residue data into the same detection segment for comparison within the same grooving region. Segment correspondence of cutting action data and grooving residue data within the same detection segment specifically includes: reading the segment number, longitudinal segment range, and lateral coordinate range of the detection segment; classifying the cutting action data within the longitudinal segment range and lateral coordinate range into the corresponding detection segment; classifying the grooving residue data within the same longitudinal segment range and lateral coordinate range into the corresponding detection segment; and classifying the cutting action data and grooving residue data classified into the same detection segment according to the grooving opening, grooving shoulder, and grooving bottom positions. The opposite record is used to mark inconsistencies between the cutting action data and grooving residue data within the same grooving region. The opposite record includes records of unmanifested action, records of outward expansion of residue, records of repeated action, and records of biased action. The recording of opposite records for cutting action data and grooving residue data refers to the following: when the cutting action data indicates that the corresponding detection segment has a cutting action and the grooving residue data indicates that the grooving bottom formation state of the corresponding detection segment does not meet the grooving bottom forming requirements corresponding to the V-groove design data, an action-not-manifested record is formed; when the grooving residue data indicates that the corresponding detection segment has a grooving shoulder residue state or a defect residue state that exceeds the cutting action range, a residue expansion type record is formed; when the cutting action data indicates that the corresponding detection segment has repeated cutting action and the grooving residue data indicates that the corresponding detection segment has more than two bottom line candidate areas, a repeated action type record is formed; when the cutting action data indicates that the corresponding detection segment has a cutting bias and the grooving residue data indicates that the grooving shoulder residue states on both sides of the corresponding detection segment are inconsistent, a bias action type record is formed; the process residue difference segment is used to record the position of the opposite record in the detection segment sequence, the type of opposite record, and the object direction of subsequent directional perturbation. The process residue difference segment is consistent with the segment number, longitudinal segment range, and lateral coordinate range of the corresponding detection segment.

[0027] In this embodiment S2, when matching the cutting action data and grooving residue data within the same detection segment, the segment number, longitudinal segment range, and transverse coordinate range of the detection segment are read. Data in the cutting action data whose corresponding workpiece coordinate position, action area boundary, or action center position falls within the longitudinal segment range and transverse coordinate range are assigned to the corresponding detection segment. Similarly, data in the grooving residue data whose corresponding grooving area position, residual area boundary, or defect area position falls within the same longitudinal segment range and transverse coordinate range are assigned to the corresponding detection segment. After the detection segment assignment is completed, the cutting action data and grooving residue data assigned to the same detection segment are matched according to the groove opening, groove shoulder, and groove bottom position. The groove opening corresponds to the V-groove opening boundary area, the groove shoulder corresponds to the two sides of the V-groove and their adjacent areas, the groove bottom position corresponds to the theoretical position of the V-groove bottom and its adjacent areas, and the defect residue data is assigned according to its adjacent groove shoulder or the outer area of ​​the V-groove boundary. Segment matching enables the cutting action data and grooving residue data to form a comparable data combination under the same detection segment and the same groove shape.

[0028] In this embodiment S2, when the cutting action data and the residual grooving data for the same detection segment and the same groove shape are inconsistent, a reverse record is generated. The reverse record includes an inconspicuous action record, a residual expansion record, a repeated action record, and a biased action record. The inconspicuous action record is formed based on the cutting action data indicating that the corresponding detection segment has a cutting action and the residual grooving data indicating that the groove bottom formation state of the corresponding detection segment does not meet the groove bottom forming requirements corresponding to the V-groove design data. The residual expansion record is formed based on the residual grooving data indicating that the corresponding detection segment has a groove shoulder residual state or a defect residual state that exceeds the cutting action range. The repeated action record is formed based on the cutting action data indicating that the corresponding detection segment has repeated cutting action and the residual grooving data indicating that the corresponding detection segment has two or more bottom line candidate areas. The biased action record is formed based on the cutting action data indicating that the corresponding detection segment has a cutting bias and the residual grooving data indicating that the residual states of the groove shoulders on both sides of the corresponding detection segment are inconsistent.

[0029] In this embodiment S2, when generating process residual difference segments, the detection segment that forms the opposite record is used as the difference segment generation object. The segment number, vertical segment range, and horizontal coordinate range of the corresponding detection segment are read, and the opposite record formed within the detection segment and its opposite record type are read. The association between the process residual difference segment and the detection segment is established based on the segment number. The position range of the process residual difference segment in the detection segment sequence is determined based on the vertical segment range and the horizontal coordinate range. The object pointing to the process residual difference segment is determined based on the opposite record type. The process residual difference segment is used to record the position of the opposite record in the detection segment sequence, the opposite record type, and the object pointing to the subsequent directional perturbation. It is consistent with the segment number, vertical segment range, and horizontal coordinate range of the corresponding detection segment.

[0030] In this embodiment S2, the segment number is the identification data of the detection segment in the detection segment sequence, used to establish the segment reference relationship between cutting action data, trenching residual data, process residual difference segments, and directional perturbation response data; the opposite record type is a field characterizing the category to which the opposite record belongs, used to determine the object to be verified; the object to be verified is the perturbation target area corresponding to the opposite record type, used to determine the application position and perturbation direction of subsequent directional perturbations; the correspondence between the opposite record type and the object to be verified is as follows: the object to be verified corresponding to the action-insignificant type record is the trench bottom formation area, the object to be verified corresponding to the residual expansion type record is the trench shoulder expansion area, the object to be verified corresponding to the repetitive action type record is the bottom line candidate area, and the object to be verified corresponding to the biased action type record is the left and right trench shoulder response areas; the area range of the object to be verified is jointly determined by the position range of the corresponding detection segment and the opposite record type; when the same detection segment meets the formation conditions of two or more opposite record types, the opposite record type of the process residual difference segment is determined in the order of repetitive action type record, residual expansion type record, action-insignificant type record, and biased action type record, and the object to be verified is determined according to the opposite record type of the process residual difference segment.

[0031] In this embodiment S2, when writing the segment number, opposite record type, and object to be verified into the process residual difference segment, the segment number of the detection segment forming the opposite record is first written into the process residual difference segment. This segment number is used as a segment reference field between the cutting action data, the trenching residual data, the process residual difference segment, and the directional perturbation response data. Then, the category to which the opposite record belongs is written into the opposite record type field. The opposite record type field is used to characterize whether the opposite record belongs to an action-invisible type record, a residual expansion type record, a repeated action type record, or a biased action type record. Subsequently, the object to be verified is determined according to the opposite record type, and the object to be verified is written into the target area field in the process residual difference segment, so that the process residual difference segment simultaneously has a location reference, a type identifier, and a perturbation target area. The opposite record type and the object to be verified... A fixed correspondence is established between the following: the object to be verified for the unmanifested action record is the groove bottom formation area, which is the theoretical groove bottom position of the V-groove within the corresponding detection segment and its adjacent area; the object to be verified for the residual expansion type record is the groove shoulder expansion area, which is the area where the groove shoulder position or the outer side of the V-groove boundary in the corresponding detection segment shows residual, missing, or expanded visual states; the object to be verified for the repeated action type record is the bottom line candidate area, which is the candidate visual area in the groove formation result image that can be used as the bottom line of the V-groove; the object to be verified for the biased action type record is the left and right groove shoulder response areas, which are the groove shoulder areas located on both sides of the V-groove design center line within the corresponding detection segment; the area range of the object to be verified is jointly determined by the longitudinal segment range, the lateral coordinate range, and the opposite record type of the corresponding detection segment.

[0032] In this embodiment S2, when the same detection segment meets the formation conditions of two or more opposite record types, the opposite record type of the residual difference segment in the writing process is determined in the order of repeated action type record, residual expansion type record, action insignificant type record, and biased action type record. After determining the opposite record type, the object to be verified is determined according to the opposite record type of the residual difference segment in the writing process. The repeated action type record corresponds to the bottom line candidate area, the residual expansion type record corresponds to the slot shoulder expansion area, the action insignificant type record corresponds to the slot bottom formation area, and the biased action type record corresponds to the left and right slot shoulder response areas. Through this sequential processing, the same residual difference segment in the process is written with one opposite record type and one object to be verified, and a definite data correspondence is formed between the segment number, the opposite record type, and the object to be verified.

[0033] In this embodiment S3, the perturbation application position is the target action position of the object to be verified in the workpiece coordinate system; the perturbation direction is the action direction determined relative to the groove part where the object to be verified is located; the perturbation execution parameters are a set of parameters that control the range and degree of action of the directional perturbation, used to make the directional perturbation act on the object to be verified within the upper limit of the perturbation execution. The perturbation execution parameters include the perturbation application position, perturbation direction, perturbation amplitude, perturbation duration, perturbation number, and perturbation range; determining the perturbation application position and perturbation direction according to the object to be verified specifically includes: when the object to be verified is the groove bottom forming area, determining the perturbation of the groove opening normal perturbation. The application location and perturbation direction are determined as follows: When the object to be verified is the outer expansion area of ​​the groove shoulder, the application location and perturbation direction of the lateral perturbation of the groove shoulder are determined; when the object to be verified is the candidate area of ​​the bottom line, the application location and perturbation direction of the vibration perturbation are determined; when the object to be verified is the response area of ​​the left and right groove shoulders, the application location and perturbation direction of the transverse alternating perturbation are determined; the perturbation execution parameters are determined based on the V-groove design tolerance, specifically including: determining the upper limit of perturbation execution according to the V-groove design tolerance, constraining the perturbation amplitude, perturbation duration, perturbation number and perturbation range based on the upper limit of perturbation execution, and generating the perturbation execution parameters corresponding to the object to be verified.

[0034] In this embodiment S3, when determining the perturbation application position and perturbation direction based on the object to be verified, the segment number, opposite record type, and object to be verified are read from the residual difference segment in the reading process. Based on the segment number, the longitudinal segment range and lateral coordinate range of the corresponding detection segment are retrieved, and the target area of ​​the object to be verified is determined within the position range of the detection segment. The groove bottom forming area is the theoretical groove bottom position of the V-groove in the corresponding detection segment and its adjacent forming area. The groove shoulder expansion area is the groove shoulder position in the corresponding detection segment and the area where the outward expansion visual state appears outside the V-groove boundary. The bottom line candidate area is the candidate visual area in the groove forming result image that can be used as the bottom line of the V-groove. The left and right groove shoulder response areas are the corresponding detection segment positions. The perturbation is applied to the shoulder areas on both sides of the V-groove design centerline. When the object to be verified is the groove bottom formation area, the perturbation application position is determined at the theoretical groove bottom position and its adjacent area, and the perturbation direction is determined as the groove opening normal. When the object to be verified is the groove shoulder expansion area, the perturbation application position is determined at the boundary area or expansion area of ​​the groove shoulder expansion area, and the perturbation direction is determined as the groove shoulder lateral direction. When the object to be verified is the bottom line candidate area, the perturbation application position is determined at the location of the bottom line candidate area, and the perturbation direction is determined as the vibration action direction. When the object to be verified is the left and right groove shoulder response areas, the perturbation application positions are determined at the left and right groove shoulder areas respectively, and the perturbation direction is determined as the lateral alternating action direction.

[0035] In this embodiment S3, when determining the perturbation execution parameters based on the V-groove design tolerance, the V-groove design tolerance in the V-groove design data is read, and the perturbation execution upper limit is determined according to the dimensional allowable deviation, positional allowable deviation, and groove shape allowable deviation corresponding to the V-groove design tolerance. The perturbation execution upper limit is a parameter boundary that constrains the degree of directional perturbation, used to limit the perturbation amplitude, perturbation duration, perturbation number, and perturbation range. After determining the perturbation execution upper limit, a perturbation parameter corresponding to the object to be verified is generated based on the region size, image resolution, and lateral coordinate range of the detection segment. The perturbation execution parameters include the perturbation application location, perturbation direction, perturbation amplitude, perturbation duration, perturbation frequency, and perturbation range. The perturbation amplitude is used to limit the displacement, airflow intensity, vibration intensity, or contact action of the directional perturbation. The perturbation duration is used to limit the duration of a single directional perturbation. The perturbation frequency is used to limit the number of times the same object to be verified is repeated. The perturbation range is used to limit the boundary of the local area covered by the directional perturbation. The perturbation application location and perturbation direction are used to limit the spatial location and orientation of the directional perturbation acting on the object to be verified.

[0036] In this embodiment S3, when constraining the perturbation amplitude, perturbation duration, perturbation count, and perturbation range, the perturbation amplitude is configured within the amplitude range limited by the perturbation execution upper limit, the perturbation duration is configured within the time range limited by the perturbation execution upper limit, the perturbation count is configured within the count range limited by the perturbation execution upper limit, and the perturbation range is configured within the lateral coordinate range of the detection segment where the object to be verified is located. For the slot bottom formation area, the perturbation range covers the theoretical slot bottom position and its adjacent area. For the slot shoulder expansion area, the perturbation range covers the slot shoulder expansion area and its boundary area. For the bottom line candidate area, the perturbation range covers the bottom line candidate visual area. For the left and right slot shoulder response areas, the perturbation range covers the left slot shoulder response area and the right slot shoulder response area, respectively. The generated perturbation execution parameters are associated and stored with the corresponding object to be verified, segment number, and opposite record type, so that the object to be verified in the residual difference segment of the same process has a definite perturbation execution basis.

[0037] In this embodiment S3, directional perturbation is a process of applying a controlled external force to the object to be verified and acquiring image responses. It is used to create visual response differences in the trench structure state, residue state, and image response state corresponding to the object to be verified. The directional perturbation of the object to be verified specifically includes: performing trench mouth normal perturbation on the trench bottom forming area according to the perturbation execution parameters; performing trench shoulder lateral perturbation on the trench shoulder expansion area according to the perturbation execution parameters; performing vibration perturbation on the bottom line candidate area according to the perturbation execution parameters; and performing lateral alternating perturbation on the left and right trench shoulder response areas according to the perturbation execution parameters. The perturbation response image sequence is an image sequence acquired during the directional perturbation process, used to record the image response state of the object to be verified before and after the directional perturbation. The perturbation response image sequence is consistent with the segment number of the corresponding process residue difference segment.

[0038] In this embodiment S3, directional perturbation is a process of applying a controlled external action to the object to be verified according to the perturbation execution parameters and collecting its image response. The controlled external action can be achieved through micro airflow components, flexible contact components, micro vibration components, pulse airflow components, and controllable displacement actuators. When performing directional perturbation on the object to be verified, the corresponding perturbation method is selected according to the type of the object to be verified, and the perturbation application position, perturbation direction, perturbation amplitude, perturbation duration, perturbation number, and perturbation range are controlled according to the perturbation execution parameters. Directional perturbation is used to create visual response differences in the trough structure state, residue state, and image response state corresponding to the object to be verified. The trough structure state exhibits a stable response in terms of boundary and position before and after directional perturbation. The residue state exhibits a local movement, dispersion, or position change response before and after directional perturbation. The image response state exhibits changes in local grayscale, texture, reflection, or contour response before and after directional perturbation.

[0039] In this embodiment S3, when performing groove-mouth normal perturbation on the groove bottom forming area according to the perturbation execution parameters, a controlled action is applied at the perturbation application position corresponding to the groove bottom forming area, either along the groove opening towards the groove bottom or along the V-groove depth direction, and the boundary and position states of the groove bottom forming area before and after the perturbation are collected; when performing groove shoulder lateral perturbation on the groove shoulder expansion area according to the perturbation execution parameters, a controlled action is applied inward or outward along the groove shoulder boundary in the boundary region or expansion region of the groove shoulder expansion area, and the boundary state and residual position state of the groove shoulder expansion area before and after the perturbation are collected; when performing vibration perturbation on the bottom line candidate area according to the perturbation execution parameters, a small-amplitude periodic controlled action is applied in the bottom line candidate area, and the candidate line position state of the bottom line candidate area before and after the perturbation is collected; when performing lateral alternating perturbation on the left and right groove shoulder response areas according to the perturbation execution parameters, alternating perturbation is applied in the left and right groove shoulder areas along the lateral direction perpendicular to the V-groove segment coordinate axis. The perturbation process involves a controlled action and the acquisition of local response differences in the left and right shoulder areas before and after the perturbation. When acquiring the perturbation response image sequence, the acquisition area of ​​the image acquisition device is aligned with the object to be verified and its adjacent areas. Multiple frames of response images are acquired according to a preset acquisition sequence before, during, and after the directional perturbation is performed. These multiple frames of response images arranged in the order of acquisition time constitute the perturbation response image sequence. The perturbation response image sequence is process image data recording the image response state of the object to be verified before and after the directional perturbation, used to characterize the boundary position, region contour, candidate line position, residue position, local grayscale response, and local texture response changes of the object to be verified. When acquiring the perturbation response image sequence, the segment number of the corresponding process residual difference segment, the object to be verified, and the response acquisition sequence are written into the image acquisition record, ensuring that the perturbation response image sequence is consistent with the segment number of the corresponding process residual difference segment.

[0040] In this embodiment S3, the response verification data corresponding to the object to be verified is the state data of the object to be verified extracted from the perturbation response image sequence, which is used to characterize the visual response change of the object to be verified before and after directional perturbation. The extraction of response verification data corresponding to the object to be verified from the perturbation response image sequence specifically includes: extracting the boundary preservation state and position preservation state from the perturbation response image sequence corresponding to the trench bottom formation area; extracting the boundary preservation state and residue movement state from the perturbation response image sequence corresponding to the trench shoulder expansion area; extracting the position preservation state from the perturbation response image sequence corresponding to the bottom line candidate area; and extracting the local response difference state from the perturbation response image sequence corresponding to the left and right trench shoulder response areas. The directional perturbation response data is used to record the correspondence between the response verification data and the object to be verified, and is mapped to the process residue difference segment according to the segment number. The directional perturbation response data includes the segment number, the object to be verified, the response verification data, and the response data identifier.

[0041] In this embodiment S3, when extracting response verification data corresponding to the object to be verified from the perturbation response image sequence, the image region to be processed is first determined according to the segment number carried by the perturbation response image sequence and the object to be verified. Then, the image frames before, during and after the execution of the directional perturbation are registered to extract the boundary, contour, candidate line, residual area and local visual response of the object to be verified in each image frame. The response verification data is the state data of the object to be verified extracted from the perturbation response image sequence, which is used to characterize the changes in the visual response of the object to be verified before and after the directional perturbation. The response verification data corresponds to the object to be verified because the acquisition area, perturbation application position and extraction area of ​​the perturbation response image sequence are all determined by the same object to be verified, and are associated with the corresponding residual difference segment through the segment number.

[0042] In this embodiment S3, when the object to be verified is the bottom formation area of ​​the trench, the boundary preservation state and the position preservation state are extracted from the perturbation response image sequence corresponding to the bottom formation area of ​​the trench. The boundary preservation state is the state data of the boundary contour of the object to be verified maintaining the same boundary shape or the same boundary position before and after the directional perturbation. The position preservation state is the state data of the target area, candidate line position or forming position of the object to be verified maintaining within the corresponding detection segment range before and after the directional perturbation. When the object to be verified is the outer expansion area of ​​the trench shoulder, the boundary preservation state and the residual movement state are extracted from the perturbation response image sequence corresponding to the outer expansion area of ​​the trench shoulder. The residual movement state refers to the state data of the positional changes of the visible residual area, connected area, or local texture area within the outer expansion area of ​​the slot shoulder before and after directional perturbation; when the object to be verified is the bottom line candidate area, the position preservation state is extracted from the perturbation response image sequence corresponding to the bottom line candidate area; when the object to be verified is the left and right slot shoulder response areas, the local response difference state is extracted from the perturbation response image sequence corresponding to the left and right slot shoulder response areas. The local response difference state is the boundary displacement difference, regional change difference, grayscale response difference, or texture response difference formed by the left and right slot shoulder response areas under the lateral alternating perturbation.

[0043] In this embodiment S3, when generating directional perturbation response data, the response verification data extracted from the perturbation response image sequence is associated with the corresponding object to be verified, and the segment number of the corresponding process residual difference segment is read. The segment number, the object to be verified, the response verification data, and the response data identifier are written into the same data record to form directional perturbation response data. The response data identifier is used to identify the acquisition batch, response type, and data sequence number of this response data. The directional perturbation response data is used to record the correspondence between the response verification data and the object to be verified, and is mapped to the process residual difference segment according to the segment number. Through the directional perturbation response data, the image response state of the trench bottom formation area, trench shoulder expansion area, bottom line candidate area, and left and right trench shoulder response areas after the directional perturbation is structured and recorded as data that can be read and matched.

[0044] In this embodiment S4, the difference segment verification status is a status identifier characterizing the opposite record type in the residual difference segment after verification by directional perturbation response data. It is used to determine the validity of the opposite record in the corresponding detection segment. The difference segment verification status includes shallow cut confirmation status, edge expansion confirmation status, loose residue reversal status, repeated cutting confirmation status, off-cut confirmation status, and difference exclusion status. The difference segment verification status is generated based on the opposite record type in the residual difference segment, the object to be verified, and the directional perturbation response data. Specifically, this includes: matching the opposite record type, the object to be verified, and the directional perturbation response data; determining the validity status, exclusion status, or reversal status of the opposite record type based on the matching result; and generating the corresponding difference segment verification status based on the validity status, exclusion status, or reversal status. The V-groove cutting accuracy detection result is cutting accuracy status data formed based on the difference segment verification status of the corresponding detection segment of the residual difference segment. It includes the segment number of the corresponding detection segment in the detection segment sequence, the difference segment verification status, the longitudinal segment range, and the transverse coordinate range, and is arranged according to the segment sorting identifier in the detection segment sequence.

[0045] In this embodiment S4, when generating the difference segment verification status based on the opposite record type, the object to be verified, and the directional perturbation response data in the process residual difference segment, the segment number, opposite record type, and object to be verified in the process residual difference segment are read, and the directional perturbation response data corresponding to the process residual difference segment is searched according to the segment number; after finding the corresponding directional perturbation response data, the object to be verified and the response verification data in the directional perturbation response data are read, and the object to be verified in the process residual difference segment is matched with the object to be verified in the directional perturbation response data, and the opposite record type is matched with the response verification data; the object consistency matching is used to determine that the directional perturbation response data originates from the same source. For an object to be verified, type response matching is used to determine whether the response verification data corresponds to the required verification state of the opposite record type. The difference fragment verification state is the status identifier of the opposite record type in the residual difference fragment of the characterization process after verification by the directional perturbation response data, which is used to determine the establishment result of the opposite record in the corresponding detection fragment. The establishment result of the opposite record includes the establishment state, the exclusion state, and the reversal state. The establishment state is used to indicate that the opposite record type and the corresponding response verification data form a confirmation relationship. The exclusion state is used to indicate that the opposite record type forms a difference exclusion relationship after matching with the corresponding response verification data. The reversal state is used to indicate that the appearance anomaly corresponding to the opposite record type forms a loose residual reversal relationship after matching with the corresponding response verification data.

[0046] In this embodiment S4, a verification state for the difference segment is generated based on the opposite record type, the object to be verified, and the directional perturbation response data in the residual difference segment. Specifically, when the opposite record type is an insignificant action record, and the response verification data of the corresponding trench bottom formation area in the directional perturbation response data represents both boundary preservation and position preservation states, a shallow-cut verification state is generated; when the opposite record type is a residual outward expansion record, and the response verification data of the corresponding trench shoulder outward expansion area in the directional perturbation response data represents a boundary preservation state, a collapse-edge expansion verification state is generated; when the opposite record type is a residual outward expansion record, and... When the response verification data of the corresponding shoulder expansion area in the directional perturbation response data represents the movement state of the residue, a loose residue reclassification state is generated; conversely, when the record type is a repeating action record and the response verification data of the corresponding bottom line candidate area in the directional perturbation response data represents the position holding state, a repeating cutting confirmation state is generated; when the record type is a biased action record and the response verification data of the corresponding left and right shoulder response areas in the directional perturbation response data represents the local response difference state, a biased shearing confirmation state is generated; when the opposite record type is not verified by the corresponding directional perturbation response data, a difference exclusion state is generated.

[0047] In this embodiment S4, when matching the opposite record type, the object to be verified, and the directional perturbation response data, the matching is performed in the order of segment number, object to be verified, and response verification data. When the opposite record type is an inactive record, the object to be verified is the trench bottom formation area, and the response verification data in the directional perturbation response data includes the boundary preservation state and position preservation state of the trench bottom formation area, the opposite record type is determined to be established, and a shallow-cut verification state is generated. When the opposite record type is a residual outward expansion record, the object to be verified is the trench shoulder outward expansion area, and the response verification data in the directional perturbation response data includes the boundary preservation state of the trench shoulder outward expansion area, the opposite record type is determined to be established, and a collapse-edge expansion verification state is generated. When the opposite record type is a residual outward expansion record, the object to be verified is the trench shoulder outward expansion area, and... Furthermore, when the response verification data in the directional perturbation response data includes the residual movement state of the outer expansion area of ​​the shoulder, the opposite record type is determined to form a reversal state, and a loose residual reversal state is generated; when the opposite record type is a repeating action type record, the object to be verified is the bottom line candidate area, and the response verification data in the directional perturbation response data includes the position holding state of the bottom line candidate area, the opposite record type is determined to form a valid state, and a repeating cutting verification state is generated; when the opposite record type is a biased action type record, the object to be verified is the left and right shoulder response areas, and the response verification data in the directional perturbation response data includes the local response difference state of the left and right shoulder response areas, the opposite record type is determined to form a valid state, and a biased cutting verification state is generated; when the opposite record type and the corresponding response verification data form a difference exclusion relationship, a difference exclusion state is generated.

[0048] In this embodiment S4, when generating the V-groove cutting accuracy detection result based on the difference segment verification status, the segment number, longitudinal segment range, lateral coordinate range, and segment sorting identifier of the detection segment corresponding to the residual difference segment in the process are read, and the difference segment verification status is written into the detection result field of the corresponding detection segment. The V-groove cutting accuracy detection result is cutting accuracy status data formed based on the difference segment verification status of the detection segment corresponding to the residual difference segment in the process, including the segment number, difference segment verification status, longitudinal segment range, and lateral coordinate range of the detection segment corresponding to the residual difference segment in the process. The segment number is used to identify the detection segment where the difference segment is located, the difference segment verification status is used to indicate the state of the opposite record in the detection segment after verification by the directional perturbation response data, the longitudinal segment range is used to indicate the position of the detection segment in the V-groove extension direction, and the lateral coordinate range is used to indicate the coverage area of ​​the detection segment in the V-groove width direction. The detection result fields are arranged according to the segment sorting identifier of the detection segment to generate the V-groove cutting accuracy detection result arranged along the V-groove extension direction.

[0049] Example 2: The machine vision-based V-groove cutting accuracy detection system proposed in this invention is applied to the machine vision-based V-groove cutting accuracy detection method proposed in Example 1. It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the machine vision-based V-groove cutting accuracy detection method in Example 1.

[0050] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A machine vision-based method for detecting the accuracy of V-groove cutting, characterized in that, Includes the following steps: S1. Obtain the V-groove design data of the workpiece to be tested, and establish the workpiece coordinate system and V-groove segment coordinate axis based on the V-groove design data; during the V-groove cutting process, acquire the instantaneous image sequence of cutting, and after the V-groove cutting is completed, acquire the groove result image, map the instantaneous image sequence of cutting and the groove result image to the workpiece coordinate system, and divide the segments along the V-groove segment coordinate axis to generate the detection segment sequence. The V-groove design data includes the V-groove design centerline and V-groove design tolerance; the detection segment sequence is a set of fragmented image data formed by dividing along the V-groove segment coordinate axis according to a preset segment length, including detection segments arranged according to the V-groove extension direction; S2. Based on the detection segment sequence, extract cutting action data from the instantaneous cutting image sequence and extract grooving residual data from the grooving result image; perform segment correspondence between the cutting action data and grooving residual data within the same detection segment; generate process residual difference segment when the cutting action data and grooving residual data form opposite records; and write the segment number, opposite record type, and object to be verified into the process residual difference segment. Among them, the opposite record is a record in which the cutting action data and the grooved residual data within the same detection segment are inconsistent for the same groove-shaped part; the process residual difference segment is a data segment that records the opposite record type and its corresponding detection segment position; the object to be verified is the perturbation target area determined by the opposite record type; S3. Determine the perturbation application location and perturbation direction based on the object to be verified, and determine the perturbation execution parameters based on the V-groove design tolerance; perform directional perturbation on the object to be verified and collect the perturbation response image sequence, extract the response verification data corresponding to the object to be verified from the perturbation response image sequence, and generate directional perturbation response data; Among them, directional perturbation is a controlled external action applied to the object to be verified; response verification data is data that records the boundary preservation state, position preservation state, residue movement state, and local response difference state of the object to be verified before and after directional perturbation; directional perturbation response data is data that records the image response state of the object to be verified after directional perturbation. S4. Generate the difference segment verification status based on the opposite record type, the object to be verified, and the directional perturbation response data in the process residual difference segment, and generate the V-groove cutting accuracy detection result based on the difference segment verification status; Among them, the difference fragment verification state is the fragment state after the residual difference fragments in the process have been verified by the directional perturbation response data.

2. The machine vision-based V-groove cutting accuracy detection method according to claim 1, characterized in that: In step S1, the workpiece coordinate system is a spatial reference coordinate system established based on the positioning boundary of the workpiece to be measured. It is used to uniformly represent the positions of the V-groove design data, the cutting instantaneous image sequence, and the grooving result image on the workpiece to be measured. The V-groove segment coordinate axis is an axial coordinate established based on the position of the V-groove design centerline in the workpiece coordinate system. It is used to determine the V-groove extension direction and the arrangement direction of the detection segments. The cutting instantaneous image sequence is process image data acquired in chronological order during the V-groove cutting process, and the grooving result image is the result image data acquired after the V-groove cutting is completed. The cutting instantaneous image sequence and the grooving result image are mapped to the workpiece coordinate system, and segments are generated along the V-groove segment coordinate axis. The division specifically includes: mapping the V-groove cutting area in the instantaneous cutting image sequence to the workpiece coordinate system to obtain the process image coordinate region; mapping the V-groove grooving area in the grooving result image to the workpiece coordinate system to obtain the result image coordinate region; dividing the process image coordinate region and the result image coordinate region into co-position segments based on the V-groove segment coordinate axis to generate a detection segment sequence; the detection segment sequence is used to establish the segment correspondence between the instantaneous cutting image sequence and the grooving result image, and each detection segment in the detection segment sequence includes a segment number, a longitudinal segment range along the V-groove segment coordinate axis, a transverse coordinate range perpendicular to the V-groove segment coordinate axis, and a segment sorting identifier.

3. The machine vision-based V-groove cutting accuracy detection method according to claim 2, characterized in that: In S2, the cutting action data is the process visual data extracted from the instantaneous cutting image sequence within the corresponding detection segment, used to characterize the action state of the tool relative to the detection segment during V-groove cutting. The cutting action data includes tool action data, cutting response data, and action timing data. The grooving residue data is the result visual data extracted from the grooving result image within the corresponding detection segment, used to characterize the grooving residue state within the detection segment after V-groove cutting is completed. The grooving residue data includes groove opening residue data, groove shoulder residue data, groove bottom residue data, and defect residue data.

4. The machine vision-based V-groove cutting accuracy detection method according to claim 3, characterized in that: In step S2, segment correspondence is based on the segment number, longitudinal segment range, and lateral coordinate range of the detection segment. This process groups the cutting action data and grooving residue data into the same detection segment for comparison within the same grooving region. Specifically, segment correspondence between the cutting action data and grooving residue data within the same detection segment includes: reading the segment number, longitudinal segment range, and lateral coordinate range of the detection segment; assigning the cutting action data within the longitudinal segment range and lateral coordinate range to the corresponding detection segment; and assigning the grooving residue data within the same longitudinal segment range and lateral coordinate range to the corresponding detection segment. The data are assigned to the corresponding detection segment; the cutting action data and grooving residue data assigned to the same detection segment are matched according to the groove opening, groove shoulder, and groove bottom positions; the opposite record is used to mark the inconsistency between the cutting action data and grooving residue data formed in the same groove shape position. The opposite record includes action-not-manifested type record, residue-expanded type record, repeated action type record, and biased action type record; the process residue difference segment is used to record the position of the opposite record in the detection segment sequence, the type of opposite record, and the object direction of subsequent directional perturbation. The process residue difference segment is consistent with the segment number, vertical segment range, and horizontal coordinate range of the corresponding detection segment.

5. The machine vision-based V-groove cutting accuracy detection method according to claim 4, characterized in that: In S2, the segment number is the identification data of the detection segment in the detection segment sequence, used to establish the segment reference relationship between cutting action data, trenching residue data, process residue difference segments, and directional perturbation response data; the opposite record type is a field characterizing the category to which the opposite record belongs, used to determine the object to be verified; the object to be verified is the perturbation target area corresponding to the opposite record type, used to determine the application position and perturbation direction of subsequent directional perturbations; the correspondence between the opposite record type and the object to be verified is as follows: the object to be verified corresponding to the unmanifested action record is the trench bottom formation area, the object to be verified corresponding to the residual expansion type record is the trench shoulder expansion area, the object to be verified corresponding to the repeated action type record is the bottom line candidate area, and the object to be verified corresponding to the biased action type record is the left and right trench shoulder response areas; the area range of the object to be verified is jointly determined by the position range of the corresponding detection segment and the opposite record type.

6. The machine vision-based V-groove cutting accuracy detection method according to claim 5, characterized in that: In S3, the perturbation application position is the target action position of the object to be verified in the workpiece coordinate system; The perturbation direction is the direction of action determined relative to the groove-shaped part where the object to be verified is located. The perturbation execution parameters are a set of parameters that control the range and degree of action of directional perturbations. They are used to ensure that the directional perturbation acts on the object to be verified within the upper limit of the perturbation execution. The perturbation execution parameters include the perturbation application location, perturbation direction, perturbation amplitude, perturbation duration, perturbation number, and perturbation range. The perturbation execution parameters are determined based on the V-groove design tolerance. Specifically, this includes: determining the upper limit of perturbation execution based on the V-groove design tolerance; constraining the perturbation amplitude, duration, number of perturbations, and range of action based on the upper limit of perturbation execution; and generating perturbation execution parameters corresponding to the object to be verified.

7. The machine vision-based V-groove cutting accuracy detection method according to claim 6, characterized in that: In S3, directional perturbation is a process of applying controlled external action to the object to be verified and acquiring image response, which is used to make the grooved structure state, residue state and image response state corresponding to the object to be verified form visual response differences. The targeted perturbation of the object to be verified specifically includes: perturbing the groove normal of the groove bottom formation area according to the perturbation execution parameters; perturbing the groove shoulder lateral area according to the perturbation execution parameters; perturbing the bottom line candidate area according to the perturbation execution parameters; and perturbing the left and right groove shoulder response areas alternately laterally according to the perturbation execution parameters. The perturbation response image sequence is an image sequence acquired during the targeted perturbation process, used to record the image response state of the object to be verified before and after the targeted perturbation. The perturbation response image sequence is consistent with the segment number of the corresponding residual difference segment.

8. The machine vision-based V-groove cutting accuracy detection method according to claim 7, characterized in that: In S3, the response verification data corresponding to the object to be verified is the state data of the object to be verified extracted from the perturbation response image sequence, which is used to characterize the visual response change of the object to be verified before and after the directional perturbation. The response verification data corresponding to the object to be verified is extracted from the perturbation response image sequence. Specifically, this includes: extracting the boundary preservation state and position preservation state from the perturbation response image sequence corresponding to the trench bottom formation area; extracting the boundary preservation state and residue movement state from the perturbation response image sequence corresponding to the trench shoulder expansion area; extracting the position preservation state from the perturbation response image sequence corresponding to the bottom line candidate area; and extracting the local response difference state from the perturbation response image sequence corresponding to the left and right trench shoulder response areas. The directional perturbation response data is used to record the correspondence between the response verification data and the object to be verified, and is mapped to the process residue difference segment according to the segment number. The directional perturbation response data includes the fragment number, the object to be verified, the response verification data, and the response data identifier.

9. The machine vision-based V-groove cutting accuracy detection method according to claim 8, characterized in that: In S4, the difference segment verification status is a status identifier after the opposite record type in the residual difference segment of the characterization process is verified by the directional perturbation response data. It is used to determine the result of the opposite record in the corresponding detection segment. The difference segment verification status includes shallow cutting verification status, edge collapse expansion verification status, loose residue re-judgment status, repeated cutting verification status, off-cutting verification status, and difference exclusion status. The verification status of the difference fragment is generated based on the opposite record type, the object to be verified, and the directional perturbation response data in the residual difference fragment of the process. Specifically, this includes matching the opposite record type, the object to be verified, and the directional perturbation response data. The establishment, exclusion, or reversal status of the opposite record type is determined based on the corresponding matching result; the corresponding difference segment verification status is generated based on the establishment, exclusion, or reversal status; the V-groove cutting accuracy detection result is cutting accuracy status data formed by the difference segment verification status of the detection segment corresponding to the process residual difference segment, including the segment number of the corresponding detection segment in the detection segment sequence, the difference segment verification status, the vertical segment range, and the horizontal coordinate range, and is arranged according to the segment sorting identifier in the detection segment sequence.

10. A machine vision-based V-groove cutting accuracy detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes a computer program to implement the machine vision-based V-groove cutting accuracy detection method as described in any one of claims 1-9.

Images