XU Shu 1, CHEN Xiao 1, ZHANG Chengwei 1, HUANG Haipeng 1, FANG Yongzhen 2
(1. Shenzhen Power Supply Bureau Co., LTD., China Southern Power Grid, Shenzhen 518000, Guangdong, China; 2. Jingdezhen Ceramic University, Jingdezhen 333403, Jiangxi, China)
Abstract: In view of the issues regarding to surface defects of piezoelectric ceramics in high voltage network, such as difficult detection, different artificial judgment standards and low detection accuracy, an improved EPF-Wastershod coupling algorithm is proposed to detect and identify piezoelectric ceramics. A local adaptive contrast enhancement model is added to improve the distinction between defects and surrounding pixels. The mean transfer filter matrix is used to replace the Gaussian filter matrix to enhance the integrity of edge retention filter. An adaptive threshold segmentation function is used to replace the global threshold segmentation function to ensure the accuracy of threshold segmentation in different regions of the image. Through local adaptive contrast enhancement, mean shift filtering and gray scale transformation, the image pretreatment process of piezoelectric ceramic surface defects in high voltage grid is completed. Based on adaptive thresholding segmentation morphology transform and watershed transform, the surface defect image segmentation process of piezoelectric ceramic is realized. The experimental results show that the detection accuracy of piezoelectric ceramic based on the improved EPF-Watershod coupling algorithm is 96.25%. The improved EPF-Watershod coupling algorithm effectively increases the detection accuracy of piezoelectric ceramic surface defect as a feasible method.
Key words: piezoelectric ceramic; surface defect; EPF-Watershod coupling algorithm; local adaptive contrast enhancement; gray scale transformation