A proposed Wavelet-Neural Network Based Image Restoration

The essential process of collecting raw data called image reconstruction which can interpret and analyze several imaging modalities in order to form an understandable image format for humans. Thus, the goal of image reconstruction is to create a usable image with high quality allowing the inspection of internal for defects without damaging the item. It consists of several cascade process like, data acquisition, preprocessing, and the application of a reconstruction algorithm and post-processing to enhance its visual quality. This paper proposed an algorithm of combining wavelet transform with neural network for image reconstruction. Such combination offers several significant advantages by leveraging the strengths of both techniques. The proposed hybrid algorithm is very effective in several tasks like image super-resolution, denoising, and medical imaging reconstruction. The algorithm depends on an 8-level wavelet decomposition for feature extraction and Elman recurrent neural network for uncompleted image in small and big losing blocks. The reconstruction achieved 100% of accuracy even with a losing case of 75% from the original image.