To determine neighborhoods in ultrasound image denoising

Ultrasound images are effected by serious noises. In this paper, we propose an adaptive method by determining a variable neighborhood of each pixel in ultrasound image denoising. The real neighbors of a pixel can be chosen according to the local energy of the pixel, the distance between the pixel and its close pixels, the requirement of reliable estimate. All of the real neighbors of a pixel form a neighborhood, named the real neighborhood (RN) of the pixel. The RNs can preserve edges and obtain reliable estimates simultaneously, which can not be met at the same time and becomes a tradeoff problem in existing methods. The experiments also show better performance both in denoising and edges preserving than the state-of-art techniques.