Performance Evaluation for Removing the Noise from the Source Data After Processing in Communication Transceiver Using Three Different Schemes

Abstract

Images are indispensable source of information used in different fields, but sometimes suffer from unwanted sources of noise. Last studies used some techniques to tackle the problem of corrupted images. These studies such as; the ordinary scheme of using blind source separation (BSS) based classical independent component analysis (ICA) without de-noising, and the traditional scheme of BSS followed by Curve let de-noising. Despite of the latter scheme provided better quality than the first one; the current manuscript suggests enhancing the performance of noise removal by performing Curve let de-noising first then followed by BSS based fast ICA. Three different experiments using matlab programming have been carried out for both the last studies and the proposed scheme. I have tested the performance of noise extraction of two examples of original source images such as Lena and Boat. This work aims to hold a comparison study among the pervious schemes and the proposed scheme in order to check the quality improvement for a given input signal to noise ratio (SNR) such as +10dB. The outcome of the simulation results revealed that; of all the addressed schemes, the proposed scheme has the best performance enhancement that consists in the key parameters of better (signal quality, output SNR, PSNR, and RMSE).