基于混合视觉的封严蜂窝复杂磨痕量化评估方法

A hybrid vision-based method for adaptive processing and quantitative evaluation of sealing honeycomb scratch measurement data is proposed. Taking into account the characteristics of edge blurring and complex cross-sections in honeycomb scratches，the method integrates point cloud and image data through fusion analysis to achieve adaptive scratch identification，feature extraction，and quantitative evaluation. In point cloud data analysis， the measured point cloud is realigned， and the primary plane is extracted based on the prior geometric features of the honeycomb structure，enabling the alignment of the measurement coordinate system with the theoretical coordinate system. In image data analysis， image morphology algorithms are employed to address challenges such as the abundance of honeycomb cavities and deep holes，ensuring the precise extraction of honeycomb regions and the quantitative computation of scratch geometry. Testing on multiple sets of measured data from worn and ablated honeycomb surfaces demonstrates the ability of the proposed method to accurately identify all scratches， with deviations in scratch width and depth measurements being less than 5% compared to manual measurements. Moreover，the identification speed improved by more than sevenfold. The results indicate that the proposed method effectively utilizes surface morphology point cloud data for the adaptive identification and quantitative evaluation of honeycomb scratches caused by wear and ablation.