基于曲率滤波增益和主动轮廓分割的陶瓷基复合材料微损伤特征提取方法OA

[Background and purposes]To address the challenge of accurately segmenting surface textures and micro-damage regions in ceramic matrix composites(CMCs),which are often intermingled and difficult to distinguish,an innovative method was proposed,by combining curvature filtering enhancement and active contour segmentation algorithms.CMCs are critical materials for advanced aerospace structures,due to their lightweight and high-load-bearing capabilities,directly influencing fuel efficiency and flight safety.However,during manufacturing processes,surface layering,micro-damages(such as excessive gaps)or overlaps between CMC layers can be formed.These micro-damages often exhibit high similarity in grayscale and morphology to background textures,making traditional visual detection methods ineffective for high-precision extraction.This study was aimed to develop a reliable and high-precision method for micro-damage detection in CMCs,by enhancing micro-damage edges and suppressing background textures and noise through curvature filtering,followed by precise segmentation using an active contour model. [Methods]A hybrid approach,integrating curvature filtering enhancement and active contour segmentation,was employed.Firstly,curvature filtering technology was applied to process CMC images,effectively suppressing noise generated by high-frequency textures,while preserving edge details.The curvature filtering enhancement method was used to precisely calculate local curvature values to selectively enhance micro-damage edges,while suppressing the background interference.A curvature-aware edge detection and curvature-constrained adaptive enhancement model were further utilized to optimize image contrast and reduce noise.Subsequently,an active contour segmentation algorithm was applied to the enhanced images.This algorithm drives contour evolution using an energy function and integrates multi-scale and multi-feature information to iteratively refine the contour,thus achieving precise segmentation without altering the micro-damage structure.Multiple evaluation metrics,including the Edge Preservation Index(EPI)and Normalized Cross-Correlation(NCC),were used to quantify enhancement performance,while accuracy and precision metrics were used to assess segmentation effectiveness.The OVITO visualization tool was employed to track atomic motion and microstructural evolution,enabling detailed observation of micro-damage features. [Results]It is demonstrated that,after curvature filtering enhancement,the EPI and NCC of micro-damage images reached 0.9874 and 0.9986,respectively,indicating excellent edge preservation and structural similarity.For micro-damage extraction using the active contour segmentation method,accuracy and precision reached 0.9301 and 0.7549,respectively,achieving high-precision extraction of micro-damages in CMCs.The proposed method effectively enhanced micro-damage edges,while suppressing background textures and noise and enabling precise segmentation of target regions.The integration of curvature filtering and active contour segmentation fully leverages the strengths of both algorithms,achieving high-precision micro-damage feature extraction even under complex texture backgrounds. [Conclusions]This study was used to present a novel method for high-precision micro-damage detection in CMCs,by combining curvature filtering enhancement with active contour segmentation.The approach can be used to effectively address the challenge of distinguishing micro-damages from background textures,providing a reliable and high-precision detection pathway for CMC structures.The method significantly improved the accuracy and stability of micro-damage identification,offering valuable theoretical support and practical solutions for structural health monitoring of CMCs in aerospace applications.The findings contributed to a deeper understanding of micro-damage mechanisms in CMCs and provided a foundation for optimizing material design and manufacturing processes to enhance overall structural reliability.