| 摘要 |
PURPOSE:To detect the position of an object at whatever position in an image it is located. CONSTITUTION:Horizontal scan line positions are set for an image 31 obtained by picking up an object 32 by an image pickup means, a plurality of horizontal segment patterns 42 to 46 are extracted by scanning these positions in the horizontal direction and compared with a reference dimension H in the horizontal direction and thereby a first evaluation value is determined. Determination of a second evaluation value corresponding to the first evaluation value and extraction of a plurality of vertical segment patterns 47 to 54 are conducted also for the vertical direction of the image from which the horizontal segment patterns 42 to 46 of which the first evaluation value is within allowable limits are extracted. The plurality of horizontal segment patterns and the plurality of vertical segment patterns are selected one by one in a pairing manner and all the segment patterns are combined together in this way. Segment pattern pairs 42 and 48, and 44 and 48, having points of intersection on the horizontal segment patterns 42 and 44 and the vertical segment pattern 48 which are selected, are extracted, a third evaluation value is computed from the first and second evaluation values, the segment pattern pair 42 and 48 having the best evaluation value is selected from the pairs, and the position 35 of the center of the object 32 is determined. |
