Abstract: Robotic harvesting system for mushroom production requires guidance and location of vision system. Mushroom images captured in the growing bed need to be coped with and analyzed. This paper describes an algorithm which can identify center position and contour of individual mushrooms in the growing bed. First, digital characteristics and the method for contour segmentation of mushroom image were analyzed. The region marking technique for mushroom image based on sequential scan was proposed in order that center region of individual mushrooms could be marked by a series of numbers, then could be distinguished. It is proved that searching mushroom contour along radius direction with different angles from center coordinates of center region is to segment each mushroom from others independently. Coordinates points of each contour which have been found were put into corresponding dynamics link, a kind of data structure, to distinguish each contour of individual mushroom. Finally, it is proved that the above algorithm is feasible by the software of Visual C++, which can establish the theoretical foundation for the development of the vision system of mushroom harvesting robot.