Abstract: To solve the problem of automatic teliospores counting of Tilletia controversa in wheat in inspectionand quarantine, this paper put forward a method of image processing. Firstly, the paper introduced the approach ofacquiring microscopic image of teliospores of T. controversa in wheat. Then under the condition of converting the color images of teliospores into grayscale images, the noise was suppressed by median filter and image segmentationn was accomplished by metric and mathematical morphology. In order to extract teliospores district accurately,background noise and holes of teliospores in binary image were eliminated by image morphology. Furthermore, \incomplete teliospore districts on the boundary were eliminated as well. At last, the method of region label was adopted to acquire the number of teliospores. The experiment results demonstrated that the method could improvedthe image qualityeffectively and acquire teliospore districts automatically and accurately, which made it easy and convenient for recognition works in further studies.