JPS576964A - Picture extracting method - Google Patents

Abstract

PURPOSE:To ensure the simple and accurate extraction of a specific picture in an original, by calculating a pattern region having uniform characteristic from the rate of closing of black picture elements and then enclosing the pattern region with a frame. CONSTITUTION:The read-out picture information of nXn is binary coded to obtain the run length values RL(W) and RL(B) of the white and black picture elements each on a scanning line. In the case of RL(W)<=n, a picture at that area has an intermediate level and is decided to be at the variable density picture part. In case RL(W) RL(B), RL(W)<m and RL(B)<m (m=10 or so defined with sampling density of 8 picture elements/mm.), the picture at that area is decided to be at the binary picture part. Based on the results of above decisions, the continuous picture regions having an even property are extracted. Then the address [(x2n-1, y2n-1), (x2n, y2n)] of the frame enclosing the region is stored with the frame No. As a result, for instance, a specific picture can be simply extracted out of an orignal in which a number of picture elements such as a sentence photo and the like just by supplying the frame No. through a keyboard.