Research of paper surface defects detection system based on blob algorithm

Effective recognition and localization of paper defect based on machine vision is the key issue for paper defect detection system. This paper proposed an improved algorithm by combination with Blob analysis algorithm and image preprocessing approach to detect the paper defects which exist in captured images by a linear charge coupled device (CCD) camera. First, the defected images are preprocessed, such as image denoising, image segmentation, connectivity analysis, and then extract effective paper textures: defect amount, regional area, long axis, short axis, central position and so on, meanwhile draw the minimum bounding rectangles. Compared with the traditional morphology algorithm and threshold segmentation and fractal feature algorithm, the improved algorithm is validated by a great deal of experimental results with high detection efficiency and defects localization accuracy.