Real-Time Video Fusion Implemented In GStreamer Framework

Abstract

In this paper we present real-time pixel level video fusion implemented in GStreamer framework as a plugin. The processing is performed in two steps. The first step is image registration, which is compensation of the effect of different viewing angles and perspective of two cameras that are observing the same scene. The second step is pixel-level fusion using Laplacian pyramid method or simple addition. The OpenVX standard is used to accelerate the algorithm. We also present experimental performance results of video fusion of thermal and visible light camera.