Adaptive weighted median filter using local entropy for ultrasonic image denoising

In this paper we present an information theory based adaptive weighted median filter. The weights of the filter are set based on localized image entropy measurements. The lower the local entropy is the smoother becomes the behavior of the proposed filter. In contrast, when the localized entropy is high, the filter's signal preservation mechanism becomes more dominant. The paper presents experimental data to compare the performance of the proposed filter with other well-known filters on a set of simulated images as well as B-mode ultrasonic images. It is shown that the proposed adaptive entropy weighted median (AEWM) filter has a superior performance in both the speckle reduction and edge preservation.