Edge-adaptive depth map coding with lifting transform on graphs

Chao, Yung‐Hsuan; Ortega, Antonio; Hu, Wei; Cheung, Gene

doi:10.1109/pcs.2015.7170047

Cited by 20 publications

(20 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unlike [7], GFTs derived from structured sparse graphs lead to efficient implementation via lifting [4]. However, while edge weights can be selected to reflect the similarities of connecting pixels, using an underlying 4-connected graph structure with only vertical and horizontal edges restricts the degree of freedom when adapting to the target block's signal structure.…”

Section: Related Workmentioning

confidence: 99%

“…Our proposal differs in the following regards. First, while [9] considered up to 24-connected graphs (each pixel is connected to its 24 nearest pixels), all our graph templates remain 4-connected and thus are sparser, which can lead to faster implementation via lifting [4]. Second, we cluster blocks based solely on principal gradients derived from computed structure tensor, which leads to fast identification of blocks to appropriate clusters.…”

Section: Related Workmentioning

confidence: 99%

“…Writing Laplacian L in the form (4) shows that the eigenvectors of L depend only on the relative size of w 1 to w 2 rather than the absolute values of w 1 and w 2 .…”

Section: Edge Weight Computationmentioning

confidence: 99%

“…(Sparse graphs are preferable, since fast implementation of GFTs becomes possible via lifting [4].) Specifically, we first cluster blocks via the Lloyd-Max algorithm based on their principal gradients; for each code block its structure tensor is first computed, and the eigenvector corresponding to the larger eigenvalue is the principal gradient.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Designing sparse graphs via structure tensor for block transform coding of images

Rotondo

Cheung

Ortega

et al. 2015

2015 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA)

Self Cite

View full text Add to dashboard Cite

The graph Fourier transform (GFT)-adaptive to the signal structures of local pixel blocks-has recently been shown to be a good alternative to fixed transforms, e.g., the Discrete Cosine Transform (DCT), for image coding. However, the majority of proposed GFTs assume an underlying 4-connected graph structure with vertical and horizontal edges only. In this paper, we propose a design methodology to select more general sparse graph structures and edge weights, on which GFTs are defined for block-based coding. Specifically, we first cluster blocks via the Lloyd-Max algorithm based on their principal gradients, which are eigenvectors of the computed structure tensors. For each cluster a graph template with edges orthogonal to the principal gradient is designed. Finally, optimal edge weights are computed assuming each template is a graph describing the inter-pixel correlation in a Gaussian Markov Random Field (GMRF). Experimental results show that GFTs derived from our graph templates lead to sparser signal representations and fewer encoding bits than DCT for a set of natural test images.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

“…Writing Laplacian L in the form (4) shows that the eigenvectors of L depend only on the relative size of w 1 to w 2 rather than the absolute values of w 1 and w 2 .…”

Section: Edge Weight Computationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Designing sparse graphs via structure tensor for block transform coding of images

Rotondo

Cheung

Ortega

et al. 2015

2015 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Lifting on graphs has been applied to image denoising and video compression [13,10,9], but these were global transforms, applied to the whole image (or video frame). In [1], lifting was applied for block-based coding of depth maps, with results that outperformed DCT-based coding and were comparable to using GFT.…”

Section: Introductionmentioning

confidence: 99%

Graph-based lifting transform for intra-predicted video coding

Chao

Ortega

Yea

2016

2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

In this paper, we propose a graph-based lifting transform for intrapredicted video sequences. The transform can approximate the performance of a Graph Fourier Transform (GFT) for a given graph, but does not require computing eigenvectors. A predict-update bipartition is designed based on a Gaussian Markov Random Field (GMRF) model with the goal to minimize the energy in the prediction set. Additionally, a novel re-connection method is applied for multi-level graphs, leading to significant gain for the proposed bipartition method and for the conventional MaxCut based bipartition. Experiments on intra-predicted video sequences show that the proposed method, even considering the extra overhead for edge information, outperforms the Discrete Cosine Transform (DCT) and approximates the performance of the higher complexity GFT.

show abstract

Graph Spectral Image Processing

Cheung

Magli

Tanaka

et al. 2021

View full text Add to dashboard Cite

Edge-adaptive depth map coding with lifting transform on graphs

Cited by 20 publications

References 8 publications

Designing sparse graphs via structure tensor for block transform coding of images

Designing sparse graphs via structure tensor for block transform coding of images

Graph-based lifting transform for intra-predicted video coding

Graph Spectral Image Processing

Contact Info

Product

Resources

About