Graph-based transform for data decorrelation

Research output: Chapters, Conference Papers, Creative and Literary Works (RGC: 12, 32, 41, 45)32_Refereed conference paper (with ISBN/ISSN)peer-review

View graph of relations

Author(s)

Detail(s)

Original languageEnglish
Title of host publicationInternational Conference on Digital Signal Processing, DSP
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages177-180
ISBN (Print)9781509041657
Publication statusPublished - Oct 2016
Externally publishedYes

Conference

Title2016 IEEE International Conference on Digital Signal Processing, DSP 2016
PlaceChina
CityBeijing
Period16 - 18 October 2016

Abstract

Transform coding can decorrelate data, and is widely used for data compression. The recent graph-based signal processing has been attracting an increasing amount of interest. In this paper, we investigate how to effectively explore the intercorrelation of a set of images as well as the spatial correlation of human motion capture data using graph-based transform (GT). Specifically, the graph structure (or matrix is first estimated by an optimization algorithm, and then the data is projected onto an orthogonal matrix consisting of eigenvectors of the estimated graph matrix, leading to sparse coefficients. Experimental results demonstrate that the GT-based method can decorrelate much better than DCT at an almost negligible price of overhead for the extremely sparse graph matrix.

Research Area(s)

  • Graph signal processing, motion capture, transform coding

Citation Format(s)

Graph-based transform for data decorrelation. / Hou, Junhui; Liu, Hui; Chau, Lap-Pui.

International Conference on Digital Signal Processing, DSP. Institute of Electrical and Electronics Engineers Inc., 2016. p. 177-180 7868540.

Research output: Chapters, Conference Papers, Creative and Literary Works (RGC: 12, 32, 41, 45)32_Refereed conference paper (with ISBN/ISSN)peer-review