Improved view interpolation for side information in multiview distributed video coding

Distributed video coding (DVC) is an attractive coding scheme for distribute camera networks because it has the potential to achieve low-complexity encoding and to exploit inter-view correlation without communication among cameras during encoding. It is well-known that the quality of the side information strongly impacts the coding performance of DVC. Therefore, this paper proposes an adaptive filtering method that yields improved view interpolation side information (VISI). The proposed adaptive filter compensates the inter-view mismatches caused by heterogeneous cameras. The filter coefficients are estimated to minimize the differences between the view-interpolated frame and the decoded frame on the key frame just prior to the WZF decoding step. Simulations show that the proposed method improves VISI quality by about 2 dB on breakdancers and about 3 dB on ballet sequences.