High bit depth infrared image compression via low bit depth codecs

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

759 Downloads (Pure)

Abstract

Future infrared remote sensing systems, such as monitoring of the Earth's environment by satellites, infrastructure inspection by unmanned airborne vehicles etc., will require 16 bit depth infrared images to be compressed and stored or transmitted for further analysis. Such systems are equipped with low power embedded platforms where image or video data is compressed by a hardware block called the video processing unit (VPU). However, in many cases using two 8-bit VPUs can provide advantages compared with using higher bit depth image compression directly. We propose to compress 16 bit depth images via 8 bit depth codecs in the following way. First, an input 16 bit depth image is mapped into 8 bit depth images, e.g., the first image contains only the most significant bytes (MSB image) and the second one contains only the least significant bytes (LSB image). Then each image is compressed by an image or video codec with 8 bits per pixel input format. We analyze how the compression parameters for both MSB and LSB images should be chosen to provide the maximum objective quality for a given compression ratio. Finally, we apply the proposed infrared image compression method utilizing JPEG and H.264/AVC codecs, which are usually available in efficient implementations, and compare their rate-distortion performance with JPEG2000, JPEG-XT and H.265/HEVC codecs supporting direct compression of infrared images in 16 bit depth format. A preliminary result shows that two 8 bit H.264/AVC codecs can achieve similar result as 16 bit HEVC codec.
Original languageEnglish
Title of host publicationProc. SPIE 10403, Infrared Remote Sensing and Instrumentation XXV
Number of pages9
PublisherSPIE - International Society for Optical Engineering
Publication date2017
Article number104030A
DOIs
Publication statusPublished - 2017
EventSPIE Optics + Photonics 2017 - San Diego, United States
Duration: 6 Aug 201710 Aug 2017
http://spie.org/conferences-and-exhibitions/past-conferences-and-exhibitions/optics-and-photonics-2017

Conference

ConferenceSPIE Optics + Photonics 2017
CountryUnited States
CitySan Diego
Period06/08/201710/08/2017
Internet address

Fingerprint Dive into the research topics of 'High bit depth infrared image compression via low bit depth codecs'. Together they form a unique fingerprint.

Cite this