Abstract: Defects edge is the most important characteristic in an electrode image. The Canny operator is used to extract the defect edge, and its principles and implementation are discussed. To overcome the problems of detecting the false edge as well as losing the local edge with slow variation of value while detecting the electrode edge with traditional Canny operator, a new method is proposed to calculate image gradient by 8-neighbor pixels, also to adaptively determine the high and low thresholds of Canny operator using 2D maximum entropy by the edge gradient feature. Experimental results show that the Canny operator is a very effective way to detect the defect edge, and suitable for the electrode defect-detecting process based on computer vision.

Key words: Canny operator, adaptive threshold, entropy, edge detection