Multilevel Image Compression on Remote Sensing Images Using Biorthogonal Wavelet Transform and Hybrid Genetic based Directional Median Filter

Abstract

Remote sensing photos are images that are collected and interpreted without coming into direct contact with an object, place, or event. The strategies for image compression with computational complexity are introduced in the preceding research. This work is focused on 2D Modified Lapped Biorthogonal Wavelet Transform (2D ML BWT) with Hybrid Genetic Based Directional Median Filter (HGMF) using Genetic based Artificial Nero-Fuzzy Inference System (GANFIS). In this research, Lapped Biorthogonal Transform (LBT) is used to lap the information when transforming the images. By using this technique, the quality of the reconstructed images is improved. Then BWT is applied by employing the WAPDF which enhances the symmetrical coefficients. Biorthogonal wavelet filter banks with equal even lengths are a subset of bi-orthogonal wavelet filter banks. Hence, the noise rates are reduced but to improve the optimal coefficient selection, GANFIS is used in this research. HGMF is used to increase the correlation co- efficiency during the image compression process. Incorporating genetic algorithms into the design of such ANFIS (Artificial Nero-Fuzzy Inference System) models are focused to eliminate impulse noise from the specified images more proficiently. The genetic algorithm generates the objective function by computing the better fitness value and selects the fuzzy rules optimally. The significant aim of ANFIS algorithm is used to attain more precise outcomes along with enhanced image quality. This technique is applied for training and learning neural networks to decide the optimal factors of fuzzy system grounded on neuron model. Thus, the proposed method achieves robust image compression performance by means of various performance metrics using 2D-MLBWT with GANFIS approach.