关键词: 磨粒, Canny 算子, 最大类间方差, 粒子群优化算法

Abstract: In order to effectively identify the abrasive grains on a diamond grinding wheel, the Canny
operator based on particle swarm optimization was applied on the grain edge detection. With
Maximum Classes Square Error (OTSU) as the objective function, the threshold of the Canny operator
was optimized for grains detection. The edges of a single grain and a group of grains were detected
using the method respectively. The results of the examples show that the algorithm can identify grains
of diamond grinding wheel effectively. The method can automatically set and optimize the threshold of
Canny operator which traditionally need to be set artificially. Finally, the threshold optimized Canny
operator was compared with the traditional Canny operator and OTSU by using the ratio of the pixels
number of four and eight connected areas to the total number of pixels. The results show that the
threshold optimized Canny operator can improve the accuracy of detection.

Key words: grains, Canny operator, maximum classes square error, particle swarm optimization