Evolvable Hardware Image Filters with Discriminations of Noise Patterns

In the recent years, evolutionary design of image filters that are adaptive to noises and hardware implement able is an emerging research topic. This study deals with the design of multiple evolvable hardware (EHW) based image filters with discriminations on noise patterns. Two indicators, similarity and divergence, are defined for describing the relations of pixels contained in a sliding window. in the proposed method, each pixel to be recovered is discriminated by similarity and divergence as one of the four noise patterns. Four EHW-based image filters, each of which is trained supervisedly and independently by the pixels belonging to a specific noise pattern, are built simultaneously. Because each image filter is dedicated to a specific noise pattern, it can recover pixels of the noise pattern more accurately. with the proposed method, the efficiency of training EHW models and accuracy of image filtering are both improved.