Image Retrieval And Data Clustering Based On Pulse-coupled Neural Network

Pulse coupled neural network (PCNN) is a new generation of artificial neural networks and is characterized with adaptive image separation and adaptive image feature extraction. This paper presents a novel image feature, Space Adaptive Histogram(SAH), which is obtained from the output of PCNN stimulated by the image, to depict a histogram-like feature of an image but with both spatial and intensity features of the image. Then PCNN is utilized for adaptive image separation, and for SAH feature extraction based on the separated parts of the image. The SAH feature is combined with the conventional texture feature of comatrix to realize image retrieval. A large number of experiments on image retrieval together with the comparison with the conventional comatrix method show the validity and effectiveness of the method presented here. On this basis further study is made on the features of grouping by similarity and capture. Then a new data clustering method based on PCNN is also presented, which is prospective and has a wide application. In this paper, we do a lot of experiments on artificial data and real world data, and make a comparison with classical data clustering methods. The experimental results prove its great advantage over others.