Image feature extraction and recognition based on adaptive Unit-Linking Pulse Coupled Neural Networks

A novel image feature extraction and recognition algorithm, using adaptive Unit-Linking Pulse Coupled Neural Networks (AULPCNN), is put forward. Firstly, ULPCNN linking strength and threshold are improved based on take into account image local information, and then AULPCNN is formed. Secondly, the time matrix is come into being, which is a mapping from the spatial image information to time information by using AULPCNN processing; it can be regarded as an image. Finally, an invariable center feature of the time matrix is defined and used in image feature extraction and recognition. Experimental results show that center feature of AULPCNN time matrix have the ability of anti-geometric distortions (TRS), and anti-noise disturbance, the novel method have characteristics of simple extraction approach, little extraction parameter, easy implementation, higher accurate recognition ratio and strong robustness.