期刊名称：International Journal of Signal Processing, Image Processing and Pattern Recognition
印刷版ISSN：2005-4254
出版年度：2016
卷号：9
期号：1
页码：281-288
DOI：10.14257/ijsip.2016.9.1.27
出版社：SERSC
摘要：Image-based crack detection methods are increasing in practical applications, because they are automatically nondestructive inspections and provide the quantitative information. To investigate the tiny and unclear cracks on the surface of plastic, we employ a novel method based on image processing and optimization to identify the cracks. Cracks are discerned in the gradient image reconstructed with four different angular orientations. Circularity is used as the description of region shape to measure the quality of binarization. After noise reductions, cracks are apparent in the binary image. Experimental results show the proposed method is an effective way to inspect micro cracks in plastic. It could be found that our method is robust to the number of sub-images. Hence, it is very suitable for the micro crack inspections in practical applications.
