Distributed Coding of Dynamic Scenes

We address the problem of compressing correlated distributed video signals that are captured from a dynamic scene. The correlated video signals originate from cameras that are not co-located or that cannot cooperate to directly exploit their correlation. However, the decoder is able to exploit the coded information from all cameras to achieve the best reconstruction of the correlated video signals. Our distributed coding scheme is based on a motioncompensated lifted wavelet transform to exploit the temporal correlation of the camera signals. The correlation among the video signals is considered by coset-encoding the quantized wavelet transform coefficients. The experimental results demonstrate that conditional decoding can reduce the bit-rate of one sequence by up to 20% when compared to independent decoding. Further, we consider theoretically the associated rate-distortion problem with side information and determine the optimal conditional KarhunenLoeve transform for video coding with side information and outline the performance bounds.