Image denoising based on hierarchical wavelet thresholding and particle swarm optimization

Yunyi Yan,Baolong Guo
2007-01-01
Journal of Information and Computational Science
Abstract:Hierarchical wavelet thresholding (HWT) which filters each subband using special hierarchical threshold instead of a universal threshold is proposed to remove noise from noisy image in this paper. Considering the computation cost, HWT simplifies the thresholds as weighted universal threshold by hierarchical factors. As a global optimization problem, the determination of hierarchical factors is achieved by the Particle Swarm Optimization (PSO) algorithm. Experiments in wide noise range showed that hierarchical wavelet thresholding was superior in terms of MSE to universal thresholding and 2D Wiener filtering, and it could preserve more fine details and also obtain better subjective image quality.
What problem does this paper attempt to address?