DAI Weijun

(Heyuan Polytechnic College, Heyuan, 517000, Guangdong, China)

Abstract: In order to solve the hard detection of such surface quality problems in honeycomb ceramics as irregular shape, subtle surface cracks and tiny side defects, the machine vision was applied to surface defects detection. The side image was captured by a CCD line scan camera with a turntable unit. Median filter was used to remove noise. Canny operator was used to extract edges. Neighboring weighted segmentation was used to extract defects and get the binary image. Region grow was used to mark a detected region. Area characteristic value was used to judge defects. The results indicate that the algorithm can rapidly detect side defects and accurately determine unqualifi ed samples; it lays the foundation for surface quality inspection of honeycomb ceramics based on on-line visual.

Key words: machine vision; honeycomb ceramics; side defects; region grow