摘要 |
PURPOSE:To make it possible to accurately read a color mark, by dividing color mark data into meshes and providing a plurality of threshold values to remove data assumed as error. CONSTITUTION:A color mark is taken in an image RAM through a TV camera and divided into meshes M (x, y) through CPU to judge a color part or a colorless part to perform binarization. When the y-component is integrated with respect to the x-component in binarized two-dimensional data M (x, y), a distorted graph A is formed because of ink omission, dust or blur. When a threshold value BTH is determined on the basis of the length ratio of the mark and dust with respect to the color part of the max. value of the graph A to perform binarization, a noise component such as small ink omission except dust and blur components is removed. Subsequently, the color part and a colorless part are binarized by a threshold value CTH corresponding to the dimensional set ratio of the continuous colorless part and the dust to remove blur and, if treatment is performed by a threshold value ATH due to the continuous color part, the dust is removed to accurately read the color mark and color mark omission inspection is performed well. |