Assessing the impact of image compression on SAR automatic target detection and cuing

Author

Gorman, John D. ; Werness, Susan A S ; Wei, Susan C.

Author_Institution

Environmental Res. Inst. of Michigan, Ann Arbor, MI, USA

fYear

1996

fDate

13-16 May 1996

Firstpage

118

Lastpage

123

Abstract

We present the results of a recent study on the impact of data compression on automatic target cuing (ATC) in magnitude-detected SAR imagery. Four “off-the-shelf” compression algorithms were tested: JPEG, an optimally-pruned tree-stuctured vector quantizer (PTSVQ), a wavelet coder with scalar quantization, and a hybrid wavelet/PTSVQ coder. At compression ratios of 2:1 and 1:1, all four compression algorithms yielded a similar image quality and we observed no statistically significant impacts on ATC performance. We also evaluated the impact of a noisy transmission channel on the overall system performance and found no measurable ATC performance loss at bit error rates of 10^-5 when error mitigation techniques were used

Keywords

image coding; noise; radar detection; radar imaging; radar target recognition; search radar; synthetic aperture radar; transform coding; trees (mathematics); vector quantisation; wavelet transforms; ATC performance; JPEG; SAR automatic target detection; automatic target cuing; bit error rates; compression algorithms; compression ratios; data compression; error mitigation techniques; hybrid wavelet/PTSVQ coder; image compression; magnitude-detected SAR imagery; noisy transmission channel; optimally-pruned tree-stuctured vector quantizer; scalar quantization; system performance; wavelet coder; Compression algorithms; Data compression; Image coding; Image quality; Loss measurement; Performance loss; Quantization; System performance; Testing; Transform coding;

fLanguage

English

Publisher

ieee

Conference_Titel

Radar Conference, 1996., Proceedings of the 1996 IEEE National

Conference_Location

Ann Arbor, MI

Print_ISBN

0-7803-3145-1

Type

conf

DOI

10.1109/NRC.1996.510667

Filename

510667