In this paper, a coring plumule positioning method of lotus seeds is proposed based on machine vision. A test platform is built to verify the coring plumule location. The lotus seeds with different sizes from four producing areas are collected and two vertical cameras are selected as image acquisition devices to capture lotus seed images fixed in the groove. After image clipping and gray processing, the orientation of lotus seed was identified by gray feature difference between head and tail of lotus seed. After binarization of the image of lotus seed along the tail of the long axis, the image is etched to eliminate edge impurities, and scaled to increase the curvature of feature points. The corner detection algorithm is used to determine the plumule coordinates of the lotus seeds which is positioned correctly. Then calculate the moving distance of manipulator by coordinate conversion to realize the lotus seed coring operation. The results show that the ideal coring position of lotus seed is the tail bump of lotus seed. The success rate of identifying whether lotus seeds are located correctly is about 97%, the accuracy rate of identifying tail bump coordinates is about 97%, the success rate of overall identification is about 94%, and the success rate of coring is about 93%. The average processing time of single lotus seed image was about 78 ms, and the average cored time was about 0.5 s. If the number of successful identification of the tail convex points is taken as the base number, the success rate of coring from the tail bump of lotus seed is 98.9%.