Reconstruction of images with RPCA and augmenting 3D model

Abstract

Crime scene photography is one of the most essential parts of documentation for all physical crime scenes in the field of forensic science. It is an organized and structured operation, having a goal of helping the investigators with the hidden information about the place of crime and finally recognizing the criminal. Photography and videography of the crime scene is a key procedure as this will further help to analyze and find viable indications about the crime. This obtained footages lead to the idea of creating a 3D model of the crime site. By using sophisticated methods in computer vision, the output obtained can be realistic reconstructions. The article further goes over a few essential designs such as crucial elements that should be included during a sophisticated Crime Scene Investigation (CSI) study. Our proposal here is to generate a realistic formation of the crime site with the multiple sophisticated designs. This also concerns the estimation of image properties such as pixel orientation, frequency, noise level etc. Moreover, we present a set of 3D pictures that have been generated and compared to state-of-the-art view synthesis methods. Furthermore, we demonstrate why using RPCA provides us with the best outcome possible.