Image Classification Method Based on Cellular Automata Transforms
Yanhui Bi, Yunjie Zhang, Ying Chen · 2006
We should give a objective description of the intrinsic characteristics of whole image according to the physical features of image before image processing, then categorize images appropriately in order to understand whether a given image suit to the processing. The cellular automata transform coefficients exhibit many features of an image. In the same scale space, significant transform coefficients congregate on a mutation gray region, where the degree of the "congregativeness" exactly correspond with the standard of image classification in term of the physical features properly. Based on the capability of detecting the mutation of the cellular automata transform, a novel image classification method is proposed. And then the description of intrinsic characteristics, which is achieved by using the appropriate orthogonal bases and energy value presented by the cellular automata transform coefficients, lead to the success in image classification based on the physical feature of image