A New Approach of Hybrid Switching Median Filter for Very Low Level Impulse Noise Reduction in Digital Images

摘要

Most of the image processing applications are greatly affected by the quality of images. Unfortunately, noise often contaminates images; yielding degradation of the quality of images. In order to overcome this flaw, a new filtering technique called the adroit hybrid switching median (AHSM) filter is presented in this work. The proposed filter is the integration of rank-ordered filtering and directional correlation-dependent filtering concepts. At first, the proposed algorithm will use a rank order based on the impulse detector to classify any potential noise pixels. Subsequently, the detected noise pixels are replaced by the estimated median value in the filtering stage. As a complement, a directional correlation-dependent filtering concept is adopted in the last iteration to clear the remainder noise pixels which are not attained in the previous filtering stage. This method is able to minimise the effects of random-valued impulse noise without degrading the details of the images. Both quantitative and qualitative analyses favor the proposed AHSM method, which consistently outperforms the conventional filtering techniques. Furthermore, the relatively easy-to-implement algorithm suggests the AHSM filter’s applicability in electronic imaging products.