A Dual-Population GA Based Visual Identification with Model Matching
Aiming at increasing speed of analysis and understanding of image and improving identification accuracy and efficiency, a novel model matching method with imaged object feature is presented, by which the grey value of image information collected from vision sensor is utilized directly and the finished image recognized result is achieved without pre-processing, so it saves the time for identification. In model matching process, a dual populations GA (DPGA) method is used for searching the best matched object, in which one is the global population served for detecting the possible area existing the optimal position, the other is the local population served for searching in the possible area for the optimal position carefully and quickly. The effectiveness of the method has been verified using real image of workpiece with rectangular feature.
Dongye Sun, Wen-Pei Sung and Ran Chen
Y. Li et al., "A Dual-Population GA Based Visual Identification with Model Matching", Applied Mechanics and Materials, Vols. 121-126, pp. 3708-3713, 2012