An efficient denoising technique for image restoration using hybrid optimization framework

Abstract

Digital images are very essential in everyday life for several applications, like medical process, astrology, satellite analysis, and so on. However, the images acquired from sensor devices are generally affected by various noises. Moreover, image processing is most effective field in various areas. Thus, image restoration techniques are developed in order to restore clear perfect images from corrupted images. Image restoration process is a vital role in various computer vision areas, like medical imaging, thermal analysis, graphic design, agricultural buildings and remote sensing. Although, the performance of existing methods has several issues, which directs to further development of enhanced image restoration approaches. The effectual image restoration approach is very essential to recover actual image from noisy input image. The digital image may degrade with the presence of different noises, such as salt and pepper noise, random noise, impulse noise and so on. This research mainly includes two contributions for image restoration process. In first contribution, JayaBat-based Deep Convolutional Neural Network (JayaBat-based DCNN) is developed for reconstructing the degraded image. At first, the input image is acquired, and it is processed to generate noisy pixel map. After that, the generated pixel map is utilized in order to identify noisy pixel map using DCNN classifier. Moreover, the DCNN classifier is trained by developing JayaBat optimization algorithm for effective restoration process. However, the developed JayaBat approach is the incorporation of Jaya optimization technique and Bat algorithm. Finally, the Deep Neuro fuzzy network is applied for performing pixel enhancement process. newline