Lightweight real-time error-resilient encoding of visual sensor data
 
Lightweight real-time error-resilient encoding of visual sensor data 
 
Jan Hanca, Geert Braeckman, Adrian Munteanu, Wilfried Philips
 
Abstract 

Extremely low-resolution video still proves suitable for many video interpretation methods and therefore may be used in small, low-cost and low-power visual sensors. Although some image analysis algorithms can be performed at the sensor node, collecting multiple video streams at the server side is necessary to execute many advanced video-based applications. Thus, an error-resilient video codec is a key component of every wireless visual sensor network. This paper introduces a novel end-to-end video compression and transmission system for such sensor networks. In the proposed framework, high-performance video coding techniques are employed, while maintaining the complexity of the encoder at the minimum. The produced bitstream is protected against wireless network errors by forward error correction codes and a row-column bit interleaver. Experimental results show that proposed codec performs close to H.264/AVC, at only a small fraction of its encoding time and offers robustness against transmission errors in various network conditions.