In response to the manual inspection of mobile phone battery surface quality in the electronics industry, a non-destructive defect detection software for battery surfaces has been developed. Initially, the battery surface undergoes preprocessing operations such as tilt correction, feature area extraction, and character grayscale value modification. The feature areas are then traversed using sub-images based on grayscale density distribution and grayscale difference with an adaptive threshold brightness method, merging defect sub-images with overlapping areas and filtering out regions without obvious defects. Subsequently, a support vector machine multi-class classification method is used to extract the distribution patterns of binary image pixels as training features to identify types of surface defects on the battery.