An improved image denoising technique using differential evolution-based salp swarm algorithm

This paper proposes an improved denoising method based on the cascaded arrangement of filters. The different combinations of filters are obtained optimally through the improved performance of the salp swarm algorithm and cascading four filters out of twelve different types of filters. The searching ability of standard salp swarm algorithm is enhanced following the strategies in differential evolution, and hence the algorithm is named as differential evolution-based salp swarm algorithm (DESSA). Most of the existing image denoising algorithms are suitable to remove either Gaussian, Salt & Pepper, or Speckle noise. Alternatively, due to the optimal combination of filters in the cascaded arrangement, the proposed denoising method exhibits its effectiveness in the removal of all three noises and the denoised images are better in terms of both quantitative analysis and visual quality. The denoising performance of the proposed method is also tested on the mixed noise which demonstrates the significant improvements compared to state-of-the-art algorithms. Further, the experiment on CEC 2014 benchmark functions indicates that the proposed DESSA achieves better optimal solutions than existing algorithms.