Research of Background Segmentation Method in Sports Video

Abstract

Human motion image segmentation is the first step in the process of human motion analysis; it is low-level processing part of the visual analysis of human movement. The processing results of the stage of quality directly affects the progress of the follow-up work, the segmentation result has enormous influence on the final results human of movement analysis. Human motion image segmentation research is an important goal similar to the visual perception of the human, which make the computer feel the movement of human body, and make computers easier to understand the feelings of sports behavior. Deviation information based on Gaussian mixture model to join the chroma and brightness of background segmentation method is not only suitable for ordinary static scene, but also it is suitable for the special environment such as reflective of ice, fuzzy shadow, moving object reflection. According to experiments results, it is shown that the proposed algorithm has strong robustness.

DOI : http://dx.doi.org/10.11591/telkomnika.v12i6.5120