Image Denoising based on Multi-Scale Dictionary Learning and Cuckoo Search Algorithm

摘要

We present a new approach to solve image denoising problem by using a combination of sparse coding and swarm optimization algorithms. One of the most recent approaches to solve image denoising problem is sparse decomposition over redundant dictionaries. In sparse coding we represent signals as a linear combination of a redundant dictionary atoms. In this paper we propose an algorithm for image denoising based on Multi-Scale Dictionary Learning (MSDL) and Cuckoo Search (CS) algorithm. In this method the dictionary is learned first by using observed images, then it applies a sparse representation algorithm to reconstruct the target image by using the constructed dictionary. In the learning step we select special atoms from the dictionary, then use the CS algorithm to update the dictionary atoms. Experiments confirms that our proposed algorithm produce state-of-the-art denoising results.