Hybridization of Social Spider Optimization (SSO) Algorithm with Differential Evolution (DE) Using Super-Resolution Reconstruction of Video Images

Abstract

AbstractIn video super-resolution reconstruction, the pixel correlation of continuous multi-frame image sequences is used to establish an effective video image super-resolution reconstruction model. Optimization problem is formed by transforming it to pixel sequence of high-resolution image from pixel sequence of low-resolution image. To effectively reconstruct the super-resolution video image, two techniques are proposed by this research. It involves the soft computing technique application as an intelligent hybridization. Differential evolution (DE) is hybridized with social spider optimization (SSO) algorithm to form a SSO-DE algorithm. Effective balance between exploitation and exploration can be achieved by this. The search space of SSO algorithm is controlled as well as adjusted dynamically by introducing a weighting factor which changes with iteration. The trapping on local optimum value is avoided by introducing a mutation operator, after the completion of social spider search. The global search’s ability is strengthened by this. In evolution for super-resolution reconstruction of video images, efficiency of proposed method is demonstrated by the experimentation on set of standard benchmark functions. Effectiveness and feasibility of the proposed algorithm are verified by a set of video super-resolution reconstruction example. KeywordsSocial spider optimization (SSO) algorithmImage/videoGlowworm swarm optimization (GSO)Differential evolution (DE)Super-resolution reconstruction