Aiming at the problem that the background of welding seam on the inner wall of tank truck affects the defect segmentation,a method based on edge detection is proposed to extract the welding seam area. Firstly,the image of tank car inner wall is preprocessed by grayscale and Gaussian filtering.Then,Canny operator is used to obtain the edge details of the weld. A series of morphological processing methods are used to connect the edge of the weld into a region. The maximum area method is used to remove the surrounding noise and preliminarily extract the weld area. Then,the minimum outer rectangle of the weld area was obtained,and the four vertex coordinates of the minimum outer rectangle were applied to the original weld image of tank truck to extract the complete weld area. Finally,the "rotary method" is used to solve the extraction problem of inclined weld.