Reconstruction of 'Phi' in Thresholding Process for a Better Compressed Image Quality

In this paper, a new thresholding algorithm that can distinguish between significant and non-significant coefficient at each detail subbands using standard deviation-based wavelet coefficients threshold estimation is proposed. The proposed algorithm start with calculating the threshold value by using the proposed threshold value estimator at wavelet detail subbands (Diagonal, Vertical and Horizontal subband). This proposed algorithm will estimate the suitable threshold value for each individual subband. The calculated threshold values are then applied to its' respective subband. The coefficients with a lower value than the calculated threshold will be discarded while the rest are retained. The novelty of the proposed method is it use the principle of the standard deviation method in deriving the threshold estimator equation. Experiments show that the proposed method effectively remove a large amount of insignificant wavelet coefficient without compromising with the image quality.