Abstract:Aiming at the current aircraft maintenance inspection with manual visual inspection,low efficiency and human factors,a nondestructive inspection system of aircraft parts surface based on image recognition and machine deep learning was designed.The images of aircraft fuselage and engine parts taken by the first-line aircraft maintenance personnel of an airline company were collected and sorted out.The image set was preprocessed,including channel extraction,Sobel filtering and binarization.Finally,Blob analysis was used to make the features extraction and analyze for the processed images.The system runs fast,has high accuracy and can recognize the image continuously automatically.Using machine vision technology to carry out nondestructive testing on the surface of aircraft parts can not only improve the production efficiency,but also remove the influence of human factors on aircraft flight safety,so as to further improve the flight safety of aircraft.The practice shows that the system is stable and reliable,and has high application value.