Stereo Image Transformation Employing Novel View Synthesis

Today’s computer graphics (CGI) and artificial intelligence world is in great demand for developing a new virtual world from accessible data, but the CPUs and monitors utilised for this type of job are too high-end and cost inefficient. Our goal is to present a method for generating a fresh viewpoint of a scene or an item from accessible data, which can then be processed for future use. In this paper, we propose a computer vision-based algorithm to generate novel views for a given pair of stereo images. We present a strategy in which we apply traditional picture warping based on camera parameters, and we primarily extract the camera’s intrinsic properties after calibrating the camera. We next build a disparity map for the provided scene whose new perspective is to be generated using the retrieved intrinsic matrix of the supplied camera. After creating the disparity map, we render it to a point cloud, transform it, and then project the altered point cloud to a 2D picture. Of course, this projected picture will have gaps, which we will patch after creating a bespoke mask. We employ inpainting techniques to cover the voids. To be more particular, the Navier–Stokes and Telea techniques.