Abstract: The smooth operation was used to reduce and eliminate the source image noise. The binary threshold digital image can be gotten by histogram threshold processing. The gradient information correction algorithm was firstly used for edge detection, and then the stomatal edge character information was highlighted by corrosion and expansion operating through detailed algorithm to extract circular defect images skeleton. The number of defects was statistically obtained by using circular center mark and the average size of defects was calculated. The results show that the circular defect contour based on Visual C++ programming algorithm is clear, and its statistics quantity is accurate.