Abstract: Niche technology was used to enhance the identification ability and clustering ability of the algorithm for local optimal solution, 
improve the local search ability of the artificial bee colony algorithm, and improve the characteristics of rapid search for global optimal value of the scout bee. We proposed an artificial bee colony algorithm based on niche technology, and applied it to extract the character image edge on the bogie wheelset axle end. This algorithm improved the disadvantage of show 
recognition speed and poor recognition effect of character edge image caused by the large amount of operation. Neighborhood gray gradient value was used as the fitness function value of the algorithm, and the niche radius was used to maintain the diversity of the population, so as to improve the local search ability of the characters, and finally a high quality edge image of the character image was obtained. The experimental results show that the number of local character edge points can be increased 1.66 times by using niche algorithm.

Key words: train wheel set, character recognition, niche radius, artificial bee colony algorithm, image edge