GPU acceleration of predictiion-based lower triangular transform for lossless compression

Shih-Chieh Wei; Bormin Huang

doi:10.1117/12.931311

19 October 2012 GPU acceleration of predictiion-based lower triangular transform for lossless compression

Shih-Chieh Wei, Bormin Huang

Proceedings Volume 8514, Satellite Data Compression, Communications, and Processing VIII; 851404 (2012) https://doi.org/10.1117/12.931311
Event: SPIE Optical Engineering + Applications, 2012, San Diego, California, United States

Abstract

The prediction-based lower triangular transform (PLT) features the same de-correlation and coding gain properties as the Karhunen-Loeve transform (KLT), but with a lower design and implementational cost. Unlike KLT, PLT has the perfect reconstruction property which allows its direct use for lossless compression. Our previous work has shown that PLT is good for lossless compression of ultraspectral sounder data with several thousands of channels. As the computation involves many operations on large matrices, this work will exploit the parallel compute power of graphics processing unit (GPU) to speed up the PLT encoding scheme. The CUDA (Compute Unified Device Architecture) platform by NVidia will be used for comparison with a single threaded CPU core. The experimental result reveals that our GPU implementation of the PLT encoding scheme shows a speedup of 95x compared to its original Matlab implementation on CPU. Thus it is promising to apply the GPU-based PLT encoding scheme for ultraspectral sounder data compression.

Citation Download Citation

Shih-Chieh Wei and Bormin Huang "GPU acceleration of predictiion-based lower triangular transform for lossless compression", Proc. SPIE 8514, Satellite Data Compression, Communications, and Processing VIII, 851404 (19 October 2012); https://doi.org/10.1117/12.931311

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available