Uniform and Robust Modelling of Edge Based Text Region Detection with Image and Video Inpainting using Morphological Operation

The proposed system introduces a novel technique of text detection and removal from the input image as well as video called image inpainting using a cost effective computational technique. The paper discusses about inpainting problem empirically and proposes a new inpainting mechanism where the same algorithm can perform text removal from both image and video. Use of an effective morphological operator is also associated with the proposed technique. The algorithm deploys the use of edge based text detection that is in perspective of font sizes, styles, color, illumination, shadows etc. The experiment is performed in Matlab using real time images and video samples to show an efficient performance of the proposed system. Finally the proposed system is also compared with most of the related techniques used in past research work to show that the proposed model can effectively localize and inpaint text regions from real time images.