Ear rows is an important agricultural feature in maize (Zea mays L.). In order to examine the feasibility of ear rows counting method by machine vision, a new method was created. This method is based on edge marker and discrete curvature. Seventy-eight digital images of four maize cultivars were scanned from two sides of maize fragment face. Based on these images, we find that the number is mostly 12–18, the absolute error of detection is 0.103, and the relative error rate is 0.66%. The accuracy rate is above 90%. Machine vision has the advantages of low cost and high speed when compared to manual or biochemical detecting methods, and it is feasible to be applied to identification of numerous maize cultivars.