The work aims to realize automatic detection of flour bag stitch breakage. A inverse P-M diffusion segmentation method was designed to detect stitch breakage. First, the stitch image was collected for gray-scale opening operation to eliminate the interference of complex background and the end lines on both sides of the flour bag. In view of the effects of environmental factors such as on-site illumination change, uneven reflection and flour noise on the extraction of stitches, the gray-scale features and gradient features of the surface of flour bags and stitches were analyzed, and the inverse P-M diffusion factor was designed. First, the image was inverse P-M diffused, and the diffused image was differentiated from the image after opening operation, so as to suppress complex background texture and enhance the differentiation between stitches and flour bags.The stitch segmented by the threshold value was compared with the stitch at the top of the flour bag in terms of length to judge whether the stitch was broken. The experimental results showed that, when the algorithm was used for stitch detection experiment, the accuracy of stitch breakage detection was 96% and the processing time of each stitch image was only 120ms. The stitch breakage detection method based on inverse P-M diffusion segmentation is a new non-contact stitch breakage detection method, which has the advantages of high accuracy and fast speed, and meets the requirements of automatic detection of stitch breakage in flour production enterprises.