In the process of producing products such as bicycle tires, there exist the phenomenon of large amount of manual identification and sorting workload, and the process is tedious and inefficient. Aimed to solve this problem, a tire rapid recognition and sorting method is researched based on machine vision. The method was relied on machine vision system to grab the production line work area image, by using the characteristic of normalized product correlation (NCC) product template matching accurate fast matching to the tire fixed number, and in sector fixed area scanned area number. By bilinear interpolation, the annular region was located and converted to a rectangular area, to achieve the rapid identification of tire number information, and ultimately driven the robot to complete the sorting work. Experimental results prove that the method is safe and reliable, and has good practical application value.