Abstract:
Edge-preserving noise reduction is an essential operation for computer-aided ultrasound image processing and understanding. The commonly used median filter can effectively reduce speckle noise and preserve the edges in ultrasound images but requires a long computation time. This report describes a filter which is a two-dimensional extension of the one-dimensional Savitzky-Golay filter. The described filter, referred to as the two-dimensional Savitzky-Golay filter, is based on the least squares fitting of a polynomial function to image intensities. The performance of the two-dimensional Savitzky-Golay filter has been compared with that of the commonly used median filter in reducing speckle noise on synthetic images and ultrasound thyroid images. Experimental results indicates that on these particular examples, the two-dimensional Savitzky-Golay filter can achieve at least the same level of noise reduction and edge preservation to that of the median filter, but with far less computation time. Since its complexity scales linearly with the problem size, the two-dimensional Savitzky-Golay filter is suitable for filtering problems with large windows.