This paper presents a distributed coding scheme for the representation of 3D scenes captured by omnidirectional cameras. We consider a scenario with a pair of similar cameras that benefit from equivalent bandwidth and computational resources. The images are captured at different viewpoints and encoded independently, while a joint decoder exploits the correlation between images for improved decoding quality. The distributed coding is built on the multiresolution representation of spherical images, whose information is split into two partitions. The encoder then transmits one partition after entropy coding, as well as the syndrome bits resulting from the channel encoding of the other partition. The joint decoder exploits the intra-view correlation by predicting one partition from the other partition. At the same time, it exploits the inter-view correlation by using motion estimation between images from different cameras. Experiments demonstrate that the distributed coding solution performs better than a scheme where images are handled independently. Furthermore, the coding rate stays balanced between the different cameras, which interestingly permits to avoid hierarchical relations between vision sensors in camera networks.

The Laplacian pyramid (LP) provides a frame expansion. Thus, there exist infinitely many synthesis operators which achieve perfect reconstruction in the absence of quantization. However, if the subbands are quantized in the open-loop mode then the dual frame synthesis operator, which is the pseudo-inverse of the analysis operator, minimizes the mean squared error (MSE) in the reconstruction. Note that this requires modification of the conventional simple synthesis scheme. For the open-loop mode, we propose novel quantization noise processing at the encoder that allows us to achieve the same performance as dual frame reconstruction and yet retain the simple synthesis scheme at the decoder. This has the advantage that the decoder can be simple in structure as well as be agnostic of whether the encoder was open-loop or closed-loop and achieves minimum MSE reconstruction for both cases. Experimental results show a gain of around 1 dB with the dual frame reconstruction compared to the simple synthesis operator. Furthermore, experiments confirm that this gain can also be obtained by retaining the simple synthesis operator and performing the proposed quantization noise processing at the encoder. Index Terms — Laplacian pyramid, dual frame, framed pyramid.

Scalability is one of the essential requirements in the compression of visual data for present-day multimedia communications and storage. The basic building block for providing the spatial scalability in the scalable video coding (SVC) standard is the well-known Laplacian pyramid (LP). An LP achieves the multiscale representation of the video as a base-layer signal at lower resolution together with several enhancement-layer signals at successive higher resolutions. In this paper, we propose to improve the coding performance of the enhancement layers through efficient interlayer decorrelation techniques. We first show that, with nonbiorthogonal upsampling and downsampling filters, the base layer and the enhancement layers are correlated. We investigate two structures to reduce this correlation. The first structure updates the base-layer signal by subtracting from it the low-frequency component of the enhancement layer signal. The second structure modifies the prediction in order that the low-frequency component in the new enhancement layer is diminished. The second structure is integrated in the JSVM 4.0 codec with suitable modifications in the prediction modes. Experimental results with some standard test sequences demonstrate coding gains up to 1 dB for I pictures and up to 0.7 dB for both I and P pictures. Copyright © 2007 Wenxian Yang et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1.

Abstract—Burt and Adelson [2] proposed a multiscale representation of images for image encoding called the Laplacian Pyramid (LP). LP is an overcomplete representation and hence results in lower compression ratios. In this paper we revisit the critical representation of the Laplacian Pyramid proposed by [1]. We investigated the (R, D) performance of the critical LP for different filters and various decomposition levels. Furthermore we explored the result of using different quantization levels for the coarse and detail signals and finally we compared the performance of the critical LP and the original LP by entropy and arithmetic coding them. I.