Abstract: Under complex meteorological conditions, as the direct pressure layer on the aerostat surface, the surface of aerostat envelope is required to be smooth and have no scratch in order to minimize the friction with air. In this paper the surface defect detection system of aerostat envelope material is designed based on machine vision. Firstly, in order to reduce the effect of background gray changes on defect detection, a preprocessing algorithm with noise filtering and image enhancement functions is studied. Secondly, the image binarization and median filtering are used to realize the feature image preprocessing and then characteristic parameters of different defect type images of envelope material is extracted by simulation experiments, combined with the texture feature extraction technology (based on gray co-occurrence matrix). The envelope material surface defect types are judged by analysing different characteristic parameters. The analysis result of 200 envelope material defect samples shows that this method can accurately identify the size, location and type of the typical envelope material surface defects of 93.6%. This system has high recognition rate and accuracy and can realize rapid detection of surface defects of aerostat envelope material.