Numerical Simulation Study on Compression of Speckle-Degraded Images Using Daubechies Wavelet

A numerical simulation study was performed in MATLAB to compress speckle-degraded images using the db10 wavelet from the Daubechies wavelet family. The results indicate that image compression quality is influenced by the selection of threshold values and decomposition levels. During the experiments, multiple comparative tests were conducted with different datasets to analyze how varying thresholds and decomposition depths affect compressed image quality. The implementation typically involves three key steps: First, performing multi-level wavelet decomposition using MATLAB's wavedec2() function to obtain approximation and detail coefficients. Second, applying thresholding techniques (hard or soft thresholding) to the detail coefficients using wthresh() function to eliminate noise while preserving important features. Third, reconstructing the compressed image through waverec2() function. Through comparative experimental results, several interesting findings emerged that provide significant guidance for optimizing image compression algorithms. The study reveals that optimal compression requires balancing threshold values (affecting data elimination) and decomposition levels (controlling frequency resolution), with higher decomposition levels capturing more detail but potentially amplifying artifacts when combined with inappropriate thresholds.