Abstract:

Segmentation method based on adaptive threshold is one of the most common detection methods of fabric defects and threshold valve directly affects the detection effect. This paper presented a new segmentation method of fabric defects based object rarity feature. First of all, the acquisition images are pretreatment using sliding average filtering; secondly, the fabric images after pretreatment are several pre-segmentation, and calculate for each segmentation obtained interest region number of object; on this basis, the largest difference of object interest region number about adjacent pre-segmentation determines the optimal threshold of fabric image segmentation; then, defects of fabric image are segmented based on optimal threshold; finally, using morphological filtering, noise elimination and false object are eliminated. The test results show that, compared with the common segmentation method, the recommended method can segment out the defects information integrity.