Comparative Analysis of Inpainting Algorithms
Balasaheb H. Patil, Predeep M. Patil · 2018 International Conference On Advances in Communication and Computing Technology (ICACCT) · 2018
Alternating Direction Methods are classified in three major types as ADM-A, ADM-B and ADM-S. ADM-A algorithm is used for the analysis based approach which is effectively useful for reconstruction of images. The advantage of ADM-A is that it gives better result for uniform background. Experimental result shows that ADM -S gives artifacts around edges and also image contains noise residuals. ADM -B algorithm is efficient and robust but it takes slightly more time than other two methods. ADM-B gives best visual quality and removes blurring effects and noise. Reconstructed images have sharper image edges than other two methods. In wavelet based image inpainting technique the selected object is being replaced by a visually conceivable background that imitates the exterior of the basis region. Different wavelet based schemes like Haar, Daubechies, Symlet and Coiflet are compared with different performance parameters like PSNR, MSE and CPU time. Haar wavelet has performed better for image inpainting than other wavelets.