Vision-Based Object Fuzzy Recognition in Plain Industrial Environment by Contour Matching
M. Sang, W. Wei, Y. Peng
The vision-based object recognition and location (VORL)
technology can greatly increase the flexibility of production automation. To simplify some very complicated VORL systems while keeping the capacity of recogniting and locating many kinds of objects, the contour matching technology is introduced to the industrial VORL system consisting of object region detection, contour detection, contour matching, pose estimation and object location. To obtain the global
optimal contour matching and decrease the computation complexity as much as possible, the dynamic programming
with ordered contour points is adopted.
Besides, based on the result of contour matching, the rotation angle can be estimated at a very little cost of computation complexity