Cost-effective implementation of order-statistics-based vector filters using minimax approximations

Vector operators based on robust order statistics have proved successful in digital multichannel imaging applications, particularly color image filtering and enhancement, in dealing with impulsive noise while preserving edges and fine image details. These operators often have very high computational requirements, which limits their use in time-critical applications. This paper introduces techniques to speed up vector filters using the minimax approximation theory. Extensive experiments on a large and diverse set of color images show that proposed approximations achieve an excellent balance among ease of implementation, accuracy, and computational speed.