Disparity Mapping Based Stereoscopic Image Quality Analysis for Efficient Mobile Object Navigation

Abstract

Occlusion is anything that obstructs vision. Removing occlusion is imperative for better perception and object recognition. The removed part of the image is needed to be reconstructed in such a way that looks natural for the eye. Thus, the image in-painting is designed to restore the occluded portion by taking information from surrounding areas. Stereo matching is an essential one to avoid the obstacles or occlusion by means of measuring disparity map for each pixel. Various methods are available now to carry out occlusion removal from images. But, the conventional techniques were failed to increase the image quality. Before removing the occlusion, the preprocessing is employed to take away the noise artifacts present in the image. Besides, many image Inpainting methods are designed to restore the occluded portion. But, the accuracy and computational complexity was not handled effectively. Visual occlusion may result in delayed or even no visualisation of traffic signs for drivers. In order to reduce the risk of traffic accidents, periodically examining occlusion of traffic signs is one of the maintenance tasks. Therefore, many traffic sign occlusion identification methods were designed but the elimination of occlusions from video is still a difficult area. In order to handle these issues, the research work is implemented with three proposed methods to remove the occlusion and thus attain better quality of images with maximum accuracy and minimal complexity. newlinePrewitt Texture Synthesis based Image Occlusion Removal (PTS-IOR) Mechanism is designed to eradicate the occlusion objects for visualizing the hidden objects with higher image quality. PTS-IOR contains preprocessing, occlusion detection, edge detection, and region filling processes. In PTS-IOR Mechanism, the noise artifacts in the input images are eliminated for improving the peak signal to noise ratio through the preprocessing using Wiener filtering. After this, the best first search algorithm is employed based on the Priority assignment in order to