Color image processing using a generalized directional distance-rational hybrid filters

This paper describes a new class of nonlinear filters called the Generalized Directional Distance Rational Hybrid Filters (GDDRHF) for multi-spectral image processing. The application at hand is color image filtering problems. The GDDRHF filter is a two-stage filter, which exploits the features of a new directional distance filter, where the temporal information related to the central sample is expressed by their weight, and those of the rational operator. This weight is incorporated to a sum of vector distances and vector angles, so that generalized directional distance rational hybrid filters will provide larger design possibility than standard vector directional distance rational hybrid filters VDDRHF. Thus, the VDDRHF is a special case of the GDDRHF. The GDDRHF includes other special cases such as the vector median rational hybrid filters (VMRHF), the vector directional distance rational hybrid filters (VDDRHF), the center weighted vector median rational hybrid filters (CWVMRHF), and the center weighted directional distance rational hybrid filters (CWDDRHF). Experimental results show that the new GDDRHF outperforms a number of widely known nonlinear filters for multi-spectral image processing with respect to all criteria used.