Search

CN-122016825-A - Nondestructive testing method for axle fracture defect

CN122016825ACN 122016825 ACN122016825 ACN 122016825ACN-122016825-A

Abstract

The invention relates to the technical field of defect detection and discloses a nondestructive testing method for axle fracture defects, which comprises the steps of acquiring multiband image data of an axle surface under the condition of full spectrum illumination; the method comprises the steps of determining the surface phase consistency of an axle according to multi-band image data with separated spectral features to obtain a structural saliency map of the axle, carrying out region segmentation on the structural saliency map while maintaining the edge integrity of the structural saliency map to obtain segmented regions of the axle, carrying out multi-feature voting on the segmented regions to obtain suspected defect regions of the axle, carrying out reliability assessment on the suspected defect regions according to the image features of the suspected defect regions and acoustic emission features of the axle to obtain target defect regions of the axle, and carrying out depth feature analysis on the target defect regions to obtain quantitative description and confidence assessment of the defects of the axle.

Inventors

  • Hui Junhao

Assignees

  • 宝鸡市聚和机械制造有限公司

Dates

Publication Date
20260512
Application Date
20260414

Claims (10)

  1. 1. A method for non-destructive inspection of an axle break defect, the method comprising: s1, acquiring multiband image data of an axle surface under a full spectrum illumination condition; S2, determining the surface phase consistency of the axle according to the multiband image data separated by the spectral features to obtain a structural significance map of the axle; S3, carrying out region segmentation on the structure significance map while maintaining the edge integrity in the structure significance map to obtain segmented regions of the axle; S4, carrying out multi-feature voting on the segmented region to obtain a suspected defect region of the axle; S5, evaluating the credibility of the suspected defect area according to the image characteristics of the suspected defect area and the acoustic emission characteristics of the axle to obtain a target defect area of the axle; s6, performing depth feature analysis on the target defect area to obtain quantitative description and confidence assessment of the defects of the axle.
  2. 2. The method of claim 1, wherein acquiring multi-band image data of the axle surface under full spectrum illumination comprises: Configuring full spectrum illumination parameters to obtain uniform illumination conditions of the surface of the axle; Under the uniform illumination condition, acquiring an original reflection image of the axle surface to obtain an original image set of the axle surface; and carrying out data integration and standardization processing on the original image set to obtain multi-band image data of the axle surface.
  3. 3. The method for non-destructive inspection of an axle fracture defect according to claim 1, wherein determining the surface phase consistency of an axle based on the separated multiband image data of spectral features to obtain a structural saliency map of the axle comprises: multidirectional filtering is carried out on the band image data with the separated optical characteristics, and filtered image data is obtained; extracting a phase consistency feature map of the axle based on the filtered image data; and normalizing and enhancing the phase consistency graph to obtain a structural significance graph of the axle.
  4. 4. The method for non-destructive inspection of an axle fracture defect according to claim 1, wherein said performing a region segmentation of said structural saliency map while maintaining edge integrity of said structural saliency map to obtain segmented regions of said axle comprises: performing edge guiding optimization processing on the structural significance map to obtain a structural feature map of the axle; Performing region growing on the structural feature map according to predefined dynamic constraint to obtain an initial segmentation region of the axle; performing multi-level confidence verification on the initial segmentation area to obtain a trusted segmentation area of the axle; And performing topological structure optimization on the trusted partition area to obtain the partition area of the axle.
  5. 5. The method of claim 4, wherein the performing region growing on the structural feature map according to the predefined dynamic constraint to obtain the initial segmented region of the axle comprises: selecting a plurality of seed points in the structural feature map based on a local maximum principle; defining a dynamic growth criterion of the axle based on the pixel intensity similarity and spatial continuity of the structural feature map; and under the dynamic growth criterion, performing region growth on the structural feature map to obtain an initial segmentation region of the axle.
  6. 6. The method for non-destructive testing of an axle fracture defect of claim 5, wherein said performing a multi-level confidence verification of said initial segmented region to obtain a trusted segmented region of said axle comprises: Evaluating texture consistency and gray distribution characteristics in the segmentation area to obtain a primary verification result of the initial segmentation area; generating a medium-level verification result of the initial segmentation region according to the boundary curvature continuity and gradient consistency of the edge pixel sequence in the initial segmentation region; determining an advanced verification result of the initial segmentation region according to the spatial relationship of adjacent regions in the initial segmentation region and the defect shape conformity of the initial segmentation region; And screening out the credible dividing area of the axle in the initial dividing area according to the fusion result of the primary verification result, the intermediate verification result and the advanced verification result.
  7. 7. The method for non-destructive inspection of an axle fracture defect according to claim 1, wherein said subjecting said segmented regions to multi-feature voting to obtain suspected defect regions of said axle comprises: Generating a region feature description of the segmented region according to the multi-dimensional features of the segmented region; According to different feature dimensions in the region feature description, carrying out multidimensional abnormal degree voting on the segmented region to obtain an initial voting result of the segmented region; Weighting and fusing voting results of different feature dimensions in the initial voting results according to the consistency of the initial voting results to obtain comprehensive voting results of the segmented areas; And generating a region screening standard of the axle according to the comprehensive voting result, and screening the segmented region according to the region screening standard to obtain a suspected defect region of the axle.
  8. 8. The method for non-destructive testing of an axle fracture defect according to claim 1, wherein said evaluating the reliability of the suspected defect area based on the image characteristics of the suspected defect area and the acoustic emission characteristics of the axle to obtain the target defect area of the axle comprises: Performing defect correlation analysis on the axle according to texture complexity and shape irregularity measurement of the image features in the suspected defect area to obtain image feature credibility of the axle; Performing defect indication analysis on the acoustic emission characteristics of the axle to obtain the reliability of the acoustic emission characteristics of the axle; Establishing the complementary evaluation of the image characteristic credibility and the acoustic emission characteristic credibility to obtain a characteristic complementarity index of the axle; Calculating a comprehensive credibility score of the axle based on the image feature credibility, the acoustic emission feature credibility and the feature complementarity index; And carrying out priority ranking on the suspected defect areas according to the comprehensive credibility scores to obtain target defect areas of the axles.
  9. 9. The method for non-destructive inspection of an axle break defect according to claim 8, wherein said integrated confidence score is calculated as follows: ; In the formula, For the scoring of the integrated trustworthiness, For the degree of certainty of the features of the image, For the reliability of the acoustic emission characteristics, And mapping values for the feature complementarity indexes.
  10. 10. The method of claim 1, wherein said performing depth profile analysis on said target defect region to obtain a quantitative description of defects and a confidence assessment of said axle comprises: performing cross-scale fusion on the local features and the global features of the target defect area to obtain multi-scale features of the axle; Performing multi-dimensional feature fusion on the multi-dimensional features to obtain fusion features of the axle; Determining the aggregate characteristics of the curves in the target defect area based on the fusion characteristics, and generating a defect quantification description of the axle according to the texture complexity and the intensity distribution of the target defect area; and scoring the internal and external coincidence degree of the defects of the axle according to the fusion characteristics to obtain the confidence degree evaluation of the axle.

Description

Nondestructive testing method for axle fracture defect Technical Field The invention relates to the technical field of defect detection, in particular to a nondestructive testing method for axle breakage defects. Background The prior art has obvious defects in the image acquisition and preprocessing links of nondestructive detection of axle fracture defects, does not acquire multi-band image data under the condition of full spectrum illumination, only relies on single-band or common illumination images for detection, cannot fully utilize effective information after spectral feature separation, leads to low distinguishing degree of axle surface defects and backgrounds, does not generate a structure significance map through multidirectional filtering and phase consistency analysis, only adopts simple edge detection or gray level threshold processing, is difficult to accurately highlight structural features of defect areas, is easily interfered by noise, causes the subsequent area segmentation to lack a reliable image basis, causes the problems of edge fracture, area confusion and the like in segmentation results, and cannot provide accurate segmentation areas for defect screening. The prior art has prominent defects in the defect identification and evaluation links. The method comprises the steps of carrying out multi-feature voting on a segmented region, screening suspected defect regions, carrying out simple defect morphology description on the target defect regions, and carrying out cross-scale depth feature analysis on the target defect regions, wherein the probability of defects is judged only according to single features, so that a large number of non-defect regions are mixed in the suspected defect regions, screening accuracy is poor, complementary reliability evaluation is carried out on uncombined image features and acoustic emission features, authenticity of defects is judged only by means of single type features, false defect signals are difficult to eliminate, reliability of target defect region identification is insufficient, accurate quantitative results and confidence evaluation cannot be generated, and detection results lack of scientificity and practicality, so that requirements of axle fracture defect accurate detection and quality evaluation are difficult to meet. Disclosure of Invention The invention provides a nondestructive testing method for axle fracture defects, which aims to solve the problems in the background technology. In order to achieve the above object, the present invention provides a nondestructive testing method for axle fracture defect, comprising: s1, acquiring multiband image data of an axle surface under a full spectrum illumination condition; S2, determining the surface phase consistency of the axle according to the multiband image data separated by the spectral features to obtain a structural significance map of the axle; S3, carrying out region segmentation on the structure significance map while maintaining the edge integrity in the structure significance map to obtain segmented regions of the axle; S4, carrying out multi-feature voting on the segmented region to obtain a suspected defect region of the axle; S5, evaluating the credibility of the suspected defect area according to the image characteristics of the suspected defect area and the acoustic emission characteristics of the axle to obtain a target defect area of the axle; s6, performing depth feature analysis on the target defect area to obtain quantitative description and confidence assessment of the defects of the axle. In a preferred embodiment, the acquiring multi-band image data of the axle surface under full spectrum illumination conditions comprises: Configuring full spectrum illumination parameters to obtain uniform illumination conditions of the surface of the axle; Under the uniform illumination condition, acquiring an original reflection image of the axle surface to obtain an original image set of the axle surface; and carrying out data integration and standardization processing on the original image set to obtain multi-band image data of the axle surface. In a preferred embodiment, the determining the surface phase consistency of the axle according to the multiband image data separated by the spectral features to obtain the structural saliency map of the axle includes: multidirectional filtering is carried out on the band image data with the separated optical characteristics, and filtered image data is obtained; extracting a phase consistency feature map of the axle based on the filtered image data; and normalizing and enhancing the phase consistency graph to obtain a structural significance graph of the axle. In a preferred embodiment, the region segmentation of the structural saliency map while maintaining the edge integrity of the structural saliency map, to obtain segmented regions of the axle, includes: performing edge guiding optimization processing on the structural significance map to obt