CONTOUR DETECTING SYSTEM FOR MULTI-VALUE PICTURE

PURPOSE:To obtain a chain code even when the difference of luminance is made small by deciding that a border line exists when the value of a directional vector is >= a certain level and enlarging a range when the directional vector is obtained. CONSTITUTION:A picture is taken in by a camera 20, formed as a digital signal by an A/D converter 16 and stored to a multi-value picture memory 13. In order to obtain the contour of the picture stored to the multi-value picture memory 13, surrounding 3X3 picture elements are segmented with a reference point as a center and these values are stored to a memory 152. Then, the size of the directional vector, which defines difference between the sum of the luminance in a lowest row and the sum of the luminance in a highest row and difference between the sum of the luminance in a right-most column and the sum of the luminance in a left-most column as elements, is obtained. When the value of the directional vector is >= a prescribed threshold, it is decided that the contour exists. Then, until the value of the directional vector exceeds the threshold, the range of the direction vector is enlarged from 3X3 to 5X5, 7X7 and 9X9 and the chain code is obtained. When the value does not still exceed the threshold even in the case that the range is enlarged up to the 9X9, it is decided that the contour can not be detected.