A fast object recognition method based on improved template matching
Hang Yi-qing · Electronic Design Engineering · 2013
At first,the template matching database with different kinds of object model were constructed in this paper;then constructing approximate polygon of the main object contour by using minimum perimeter polygon,so that the approximate polygon can simplify the main object contour and reduce the amount of calculation;next,describing the object to be recognized by extracting three affine invariants which were the number of vertices of polygons,the number of vertices of both sides of the longest line of polygons and triangle area ratio of the same hemline respectively,then matching the three affine invariants hierarchically by using traversal search in the template matching database.At last,The results of experiment show that the object recognition algorithm based on template matching in this paper can recognize object rapidly,improving the real-time performance of object recognition and getting the posture of the object to be recognized at the same time.