Image Noise Removal via Wavelet Transform and r/K-PSO

Image noise removal is a classic problem. In this paper, a novel scheme combining r/K-PSO and wavelet transform was introduced to removal image noise. By wavelet transform, image was decomposed into detail subbands and approximation subband. Every detail subband will be shrunk by a special threshold instead of a universe one. A novel optimization algorithm named r/K-PSO was introduced to optimize and determine the thresholds. The main idea of r/K-PSO is inspired by the r- and K-selection of Ecology. r-selection can be characterized as: quantitative, little parent care, large growth rate and rapid development and K-selection as: qualitative, much parent care, small growth rate and slow development. And experimental results also proved that the proposed noise removal scheme optimized by r/K-PSO was superior to 2-D Winner Filtering (WF), universal hard-thresholding (UHT) and universal soft-thresholding (UST) in terms of peak signal-to-noise ratio (PSNR).