Art2 Network with Neoteny Learning Law and and its application to color pixel analysis

Pixel analysis is the primary step for other image processing relative analyses and how to more correctly and effectively distinguish pixels with different color and luminance in an image or a video stream are of the most important aspect in these field. This paper firstly provides a reasonable mapping operation to solve the problem caused by normalization operation in ART2 network. Then it proposes the concept of "neoteny learning law" and adjustable vigilance value of ART2. Finally, it gives an application example in color pixels categorization. The processing steps and results not only demonstrate the action of "neoteny learning" but also illustrate that it is coherent with the human psychological and physiological process of observing an image and has strong adaptability for shadow noise suppression