Visual Information Extraction Model, Based On Wavelet Transform

The abstraction of the vision information is a complex process that involves signal processing, image analysis and physiology. It is revealed by the related research that human vision system is not only a photoelectric transformer, but also of some ability of preprocess of visual information. The distinct channels with discrete frequency exist in human vision system, which have different sensibility for different range of both spatial and frequency respectively. The visual image is the result from the synthesizing of the information snatched by these channels. And wavelet transform takes on the features of multi-resolution and variable scale. There are same characteristics between wavelet transform and the process of human vision system. In this paper a model of human vision information abstracting based on wavelet transform is presented. In this model an image is decomposed into components with wavelet and then proper transformation is made upon the components, and finally a new image with similar features with human vision information is synthesized using wavelet inverse transform, and detailed discussion and analysis about this model is made. And also this process is simulated with computer. Considering the abstraction of human visual information is an active process, a feedback policy exists in vision channels, a channel model of human vision information abstraction is put forward after the physical undersenses of the parameters of the model are analysed.