Genetic seam carving: A genetic algorithm approach for content-aware image retargeting

Abstract

Seam Carving is a method to retarget images by removal of pixels paths with minimal visual impact. The method acts by exhaustive searching of minimal cost paths according to a pixel relevance function. In the present paper, we explore optimal or suboptimal paths obtained by a new Genetic Algorithm method called Genetic Seam Carving. Besides the suboptimal character of this approach, we show in the experiments that, we achieve quality results similar to the original Seam Carving method, and in some cases we even obtain less degradation.