Title: Hybrid image inpainting using reproducing kernel Hilbert space and dragonfly inspired wavelet transform

Authors: Balasaheb H. Patil; Pradeep M. Patil

Addresses: All India Shri Shivaji Memorial Society's Institute of Information Technology (AISSM's IOIT), Pune, India ' Jayawantrao Sawant College of Engineering, Pune, India

Abstract: This paper intends to propose a new inpainting model that based on Mumford Shah (MS) modelling, where the original image is gained accurately by doing inpainting process in the masked image. Here, discrete wavelet transform (DWT) is used for processing with the digital image. Further, to find the optimal filter coefficients from DWT, a renowned optimisation technique named dragonfly (DA) is used. Moreover, the smoothing of image is process via reproducing kernel Hilbert smoothing model. The proposed dragonfly optimised DWT kernel-MS (DODWTK-MS) model compares its performance with other conventional methods in terms of second derivative measure of enhancement (SDME), peak signal-to-noise ratio (PSNR), signal-to-noise ratio (SNR), mean squared error (MSE) and edge similarity and the efficiency of the developed model is explained.

Keywords: digital inpainting; Mumford Shah model; discrete wavelet transform; DWT; filter coefficient; dragonfly algorithm.

DOI: 10.1504/IJNBM.2019.104946

International Journal of Nano and Biomaterials, 2019 Vol.8 No.3/4, pp.301 - 320

Received: 09 Aug 2018
Accepted: 03 May 2019

Published online: 07 Feb 2020 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article