Image enhancement and blur pixel identification with optimization-enabled deep learning for image restoration

Image enhancement is the process of enhancing specific aspects of an image, such as its borders or contrast. The procedure of restoring a destroyed image is known as image restoration. A multitude of factors, such as low camera resolution, motion blur, noise, and others, can cause images to degrade throughout the acquisition process. Although image restoration techniques can remove haze from a degraded image, they are problematic for use in a real-time system since they necessitate numerous photographs from the same location. The suggested fractional Jaya Bat algorithm (FJBA) provides picture enhancement and blur pixel identification to address this issue. Firstly, the blur pixel identification is done using a deep residual network (DRN) trained with FJBA considering blurry image. FJBA is created by combining the Jaya Bat algorithm (JBA) and fractional notion (FC). Furthermore, a blurred image is deblurred using a fusion convolutional neural network (CNN) approach tuned through Pelican hunter optimization (PHO). PHO stands for Pelican optimization (PO) and hunter prey optimization (HPO). Lastly, the image is enhanced using the neural fuzzy system (NFS) and the image enhancement conditional generative adversarial network (IE-CGAN), which has been fine-tuned using FJBA. The proposed FJBA-NFS-IE-CGAN provided enhanced performance with the highest PSNR of 50.536 dB, SDME of 60.724 dB, and SSIM of 0.963, respectively.