| 摘要 |
PROBLEM TO BE SOLVED: To reliably and surely recognize an object by use of an extracted contour line even if the object has an arbitrary contour line. SOLUTION: When learning an object 2, areas T1 and R2 for which a contour line 3 to be learnt is extracted are instructed, the contour line is extracted in the instructed areas, and each extracted contour line is recorded with a code showing the area from which the contour line concerned is extracted. During object recognition, areas for which a contour line to be recognized is extracted are instructed, a code showing the area from which each contour line is extracted is assigned to the contour line extracted in this area, and the object is recognized using a contour line in which the code showing the contour line extraction area assigned in recognition coincides with the code assigned in learning. In such a structure, the contour lines are collated after confirming coincidence of the contour line extraction area code assigned in object recognition with that assigned in learning, so that the accuracy of object recognition is improved. COPYRIGHT: (C)2008,JPO&INPIT
|
