Image defect inspection based on human visual characteristics

This paper presents a new image defects inspection algorithm based on image morphology analysis and human visual characteristics. In this algorithm, the difference image is obtained by subtracting of the standard gray image and the image to be detected, referred to as noised image. The noises´ characteristics are analyzed by considering three parameters, namely, the area of noises region and perimeter, the degree of roundness, and the ratio of the noises in the region. Finally, determine those analyzed noises whether or not can be accepted by human eyes according to the human eyes´ sensitivity on background brightness and textures´ changes. Experimental results clearly prove that the algorithm can effectively distinguish image defects. This algorithm retains a great potential in image defects inspection such as printing related areas.