| 摘要 |
<p>PURPOSE:To always obtain an optimum object recognition result even in the case when a misalignment is generated in a recognition object image by setting again a monitoring area to a monitoring point having brightness information being easily decidable at the time of image pickup of a recognition object. CONSTITUTION:A monitoring area Kj is allocated, for instance, to a correction range AR of X=+ or -2 and Y=+ or -2 centering around a monitoring position Pjmn of the monitoring area Kj. Subsequently, when the monitoring area Kj is moved to each monitoring point Pjmn contained in the correction area AR, integral values Vjmn which are obtained, respectively are derived from a first and a second frame memories, respectively. Next, the monitoring position in which an absolute value of its difference is the largest is determined as an optimum monitoring position. In such a way, even in the case when a misalignment is generated in an image in a display screen, an object can always be recognized under an optimum condition.</p> |
