median filter method was applied to eliminate noise of collected hot stamping images. The color images were converted from RGB space into HSV space. H channel threshold was segmented. NCC algorithm was improved combining difference sum theorem and block template matching. Hot stamping area identification and extraction was realized. The algorithm was validated with a large number of examples. The results showed that the algorithm can quickly and effectively identify hot stamping area, which is significant for application of machine vision in hot stamping quality inspection.