A Modified NLM Model for Despeckling Ultrasound Images Considering Rapid Wavelet Fragmentation and Spit

Abstract

Speckle noise, an intrinsic ultrasound property, makes it difficult for computer-aided diagnostic (CAD) systems to appropriately diagnose patients. Using a modified non-local means (NLM) filter, ultrasound pictures may be despeckled. In the proposed NLM model, feature vectors are generated from the input picture and clustered using the fuzzy c means (FCM) technique. Individual clusters of blocks can be matched using the rotationally invariant block matching (RIBM) approach rather than the complete image. No pixels are lost in the NLM process because of the intra-cluster block matching. Images compressed using the FFWT were examined for their compressive power in this study. Biorthogonal filter banks may be created by combining FFWT with Set Partitioning in Hierarchical Tree to investigate compression performance in terms of subjective quality metrics (SPIHT). Other topic quantity measurements, including as PSNR and MSE, were used to compare the results of this study.