Research on Image Recognition Method of Aircraft Outer Surface Damage After Lightning Stroke
Defect detection has become one of the important research directions of machine vision.In the field of aeronautical re-manufacturing,machine vision technology is expected to speed up the detection task of lightning strike damage of aircraft,to re-duce labors and to shorten inspection time.In this study,the sobel_amp()operator is used to get edge images of outdoor images of aircraft's outer surface containing lightning strike damage.Based on defect characteristics of aircraft's outer surface lightning damages,a dilation edge gray difference method is proposed to identify lightning strike damages.With the proposed method,a lightning strike damage recognition system is proposed.It performs HSV component conversion,image clipping,edge extraction,improved Blob analysis,region selection and morphology processing on the image containing the lightning strike damage area of the aircraft,and finally determine the lightning strike damage area by manual recheck.Experiments show that,compared to trans-fer learning method proposed by previous work,the proposed method do not need manual annotation and model training,which can significantly improve recall rate of damage identification and avoid possibility of missing damage.Thus,it can improve detec-tion efficiency and significantly reduce human time without sacrificing flight safety.
万方数据
维普
