Automatic video background replacement using shape-based probabilistic spatio-temporal object segmentation

While video background replacement has an extensive range of applications, the chroma-keying process is still the most widely applied despite it having to maintain specific lighting conditions, a blue (or green) background, and the requirement of painstaking manual handling. This paper addresses these issues by presenting a novel fully automatic video background replacement technique that does not require any specific recording constraints. This paper employs a generic shape-based probabilistic spatio-temporal (PST) video object segmentation algorithm employing a Gaussian mixture model (GMM) to achieve a long-desired outcome towards fully automated video background replacement techniques. Experimental results using a number of standard video test sequences reveal the merits of the proposed technique.