摘要：Surface defects detection is an important work to assure armor plate good quality, In this paper, image complexity is introduced to detect defects in armor plate surface image obtained by CCD camera from rolling site. Three descriptions of complexity, general set complexity, gray level spatial distribution complexity and RPVC (rate of gray values change) complexity are analyzed and compared to get the most effective method to detect defects. The result of calculations show that RPVC for image complexity description has distinct difference between backgrounds images and defects images, some parameters of gray level spatial distribution complexity also have same effect, but in the view of time saving, image complexity described by RPVC is selected to detect defects on armor plate surface images. Experiments on images from rolling site show the validity of RPVC complexity. Aiming at twelve kinds of typical defects images, the accurate rate reaches 94%
关键词：Surface Defects Detection of Armor Plate;Image Complexity;RPVC
