An improved image inpainting technique using fuzzy hard c-means algorithm

Abstract

A novel algorithm which can repair video sequence deprived of any artefacts that are present in numerous such prevailing technique is proposed in this paper. Specifically, a video inpainting is proposed that detects moving object in a background where water is falling from a fountain. The input video is converted to frames. Using the Canny edge detection, the edges of the objects in the frames are found out. These edges are classified into different clusters using fuzzy hard C-means algorithm so that inpainting can be done effectively. The classified edges are patch-matched with the most similar pixels. The resultant inpainted video when viewed gives a very good result with minimum time frame.