Based on machine vision, the wet fruit flesh of jelling with different kind and size can be inspected automatically for multi-type foreign bodies detection before filled into containers. According to the difference degree in color and brightness between fruit flesh and foreign bodies, two approaches of image processing for foreign bodies reorganization were put forwarded. For high-saturation color fruit flesh, a segmentation algorithm based on the independence of HSI three components was adopted. For low-saturation color fruit flesh, an edge detection algorithm on mathematical morphology was taken as the main means. Then the result images were divided into several regions according to the size of fruit flesh, and the judgment whether foreign bodies existed in each region was worked out. As for these approaches, the specific processes and programs were given and verified by experiments. The results show that the method can detect a variety of foreign bodies on wet fruit flesh with a high accuracy.