Noise Reduction of Images using BFO Cascaded with Wiener Filter and Framelet Transform

Abstract

Image noise removal is necessary for image processing systems. In this paper, a hybrid denoising that combines Bacterial Foraging Optimization (BFO) cascaded with spatial domain Wiener filter and thresholding function in the Framelet domain is done. Three algorithms are proposed in this study. The first one stands for hybrid denoising algorithm that employs Wiener filter with 2-level Discrete Wavelet Transform (DWT). The second algorithm uses Wiener filter with 2-level Framelet Transform (FLT). The last one is hybrid denoising algorithm that combines Wiener filter with 1-level WT, and then applies FLT on LL of WT. After that, the BFO algorithm is applied to minify the error quantities between thea noisy image and the produced image. The adopted procedure has been tested on gray and color images compared with MAX, MINMAX algorithms, and FUZZY filters in the wavelet domain. Simulation results based on MATLAB simulator for the first proposed algorithm with DWT (db5 type) is superior to the second, third, and the conventional denoising approaches for most test noisy images with Gaussian, Salt and Pepper noises. The third proposed algorithm with hybrid Wavelet & FLT is superior to others for noisy images with speckle noise; this algorithm has given good results with medical images.