Abstract: The condition of stem is an important character to classify the pear. The pictures of Huanghua pears were analyzed with a computer vision system. The median filtering method was used to smooth the picture, and the local threshold algorithm was applied to segment the pear from the background. As the thinning and erosiondilation algorithm in judging presence of the stem is too slow, a new fast algorithm was put forward. Compared with other part of the pear, the stem is obviously thin and long, with the help of various sized templates,to judge the presence of stem was easily, meanwhile the stem head and the intersection point of stem bottom and pear body were labeled. Furthermore, after the coordinates of the stem head and bottom and the ratio of slope were calculated, the broken stem could be distinguished from the good stem. As the statistical result of 53 pictures of pears, the method accuracy to judge the stem presence is 100 %, and whether the stem is good reaches 93 %, and the classification rate was increased about 4～6 times. The algorithm is robustness and can be made invariant to translation and rotation.