Multi-threshold-based frame segmentation for content-aware video coding in WMSN
Résumé
Wireless Visual Sensor Networks have gained relevant interest in the last years due to the many advantages they show. These advantages are characterized by the efficiency of WVSN to offer accurate monitoring of critical zones with a low budget. Nevertheless, WVSN still suffer from a high amount of data when dealing with images and videos that make them not feasible in terms of lifetime. Hence, the optimization of the transmission energy and bitrate has been always a challenging task for many recent works. We advance the state-of-the-art by proposing a low bitrate technique for video coding in WVSN. The proposed method aims to apply a Region-of-Interest Based video coding in an Analyze-Then-Compress Paradigm. We detect the regions of interest where high movement occurs. Then, the compression parameters are set based on region importance. We analyze the efficiency of the proposed method using several evaluation metrics. The proposed technique shows promoting results for very low bitrate WMSN systems.