Image Target Recognition Using Pulse Coupled Neural Networks Time Matrix
Qing Liu, Yide Ma, Zhang Shaogang, HE Wan-sheng · 2006
A novel method for image target recognition based on the Pulse Coupled Neural Networks (PCNN) time matrix is presented in this paper. We describe the PCNN and put forward the concept of PCNN time matrix. The time matrix contains useful information related to spatial information of the image that is under processing. According to some physical concepts, a kind of new invariable feature of image histogram vector, histogram vector center, is defined, which is used in image target recognition. The results of computer simulations are that histogram vector center of PCNN time matrix has the ability of anti-geometric distortions (translation, rotation and scaling, TRS), and its extraction method is simple. Moreover, little parameters are extracted, and the presented target recognition approach is more efficient which compared with the traditional method.