Real time Tarp-linear estimator for image noise reduction

Abstract

Many efficient wavelet domain estimation algorithms for noise reduction have been proposed in the literature. All of them are based on variance estimation, which require all wavelet coefficients to be saved in the memory thus causing constrains towards the development of real time system. Tarp filter has been used successfully for online/real-time variance estimation of the multilevel wavelets and has shown exceptional performance in image compression. This paper proposes the utilization of the Tarp filter for the estimation of wavelet coefficient variance without saving them in memory. The paper presents the results for Tarp-linear estimator (TLE) and compares it with simple linear estimator (LE). It is found that TLE, while having an advantage of memory free noise reduction method, compromise on SNR