Abstract :
The research of this paper depends on digital image processing technology, electromechanical integration technology, nondestructive testing, evaluation technology and computer technology to further study machine vision image-based real-time processing algorithm and completes key technology research, as well as system realization of ceramic defects recognition. At first, we establish median filter-based preprocessing algorithm and reduces in homogeneity of image gray. Then, it selects edge detection operator appropriate for ceramic tube image processing. Finally, it uses iteration formulation threshold separation to perform shape feature extraction on the separated surface crack target and adopts circularity index to judge quality. Through specific cases, the algorithm is verified to be feasible, stable, and has higher quality classification accuracy.
