Speckle Denoising by the Combination of Spatial and Transform Domain Filters

Abstract: Ultrasound (US) imaging a technique used for non-invasive detection of flaws inside the human body. The quality of such images has to be at par for the use of doctors. US images are corrupted by speckle noise which arises due to the back scattering of the signal. In this work a new method for image denoising is presented. In this method firstly Approximation and detail bands are obtained from 2 level wavelet decomposition and fast bilateral filter is applied on the approximation band. Secondly the thresholding of the detail bands are done. Then filtered approximation and detail bands combined to get final denoised image. For the performance evaluation of proposed method parameters such as Peak signal to noise ratio (PSNR) and Edge Keeping Index (EKI) are used. The results are obtained for a synthetic kidney image generated by Field II program.

Keywords: Bilateral filter, DWT, speckle reduction, stick filter.