A comparative analysis of various speckle reduction techniques for ultrasound images

Abstract

In this work, a comparison of existing speckle reduction techniques have been done. Median and wiener filters, wavelet based methods and SRAD methods have been compared. The methods are first tested on synthetic images for comparison of the results and then applied on real ultrasound images. It is observed that SRAD method outperforms the other techniques for low noise variance. For higher noise variance, exponential thresholding outperforms the remaining techniques. Also, it is observed that wiener filter have a comparable performance and hence performs well over range of noise variance. Wavelet denoising based on the soft thresholding is reported to be least performing.