A multi-objective genetic algorithm for the design of biorthogonal filter banks for embedded image coding application is presented. To be effective, the filter bank would satisfy multiple requirements related to such application. Flexibility in the design is introduced by imposing Near Perfect Reconstruction (N-PR) condition instead of entire PR condition as in conventional designs. Especially for embedded coding purposes, the filter banks are designed to be near-orthogonal. This can only be made possible by minimizing the deviation from the orthogonality in the optimization process.The optimization problem is formulated as a constrained multi-objective problem and solved using a constrained Non-dominated sorting genetic algorithm (C-NSGA) by searching solutions that achieve the best compromise between the different objective criteria, these solutions are known as Pareto Optimal Solutions. Experiment results show that our designed filter banks lead to improved performances of image coding compared to those achieved by the 9/7 filter bank of JPEG2000.
In this paper, we present a global optimisation method based on a multi-objective Genetic Algorithm (GA) for the design of filter banks in a lossy image coding scheme. To be effective, the filter banks should satisfy a number of desirable criteria related to such scheme. We formulate the optimization problem as multi-objective and we use the Non-dominated Sorting Genetic Algorithm approach (NSGAII) to solve this problem by searching solutions that achieve the best compromise between the different objectives criteria, these solutions are known as Pareto Optimal Solutions. Flexibility in the design is introduced by relaxing Perfect Reconstruction (PR) condition and defining a PR violation measure as an objective criterion to maintain near perfect reconstruction (N-PR) solutions. Furthermore, the optimized filter banks are near-orthogonal. This can only be made possible by minimizing the deviation from the orthogonality in the optimization process. Our designed filter banks lead to a significant improvement in performance of coding with respect to the 9/7 filter bank of JPEG2000 at high compression ratios and offer a slight improvement at low compression ratios.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.