The Impact of Lossy Compression on Hyperspectral Data Adaptive Spectral Unmixing and PCA Classification
P. Keerthana1, A. Sivasankar2
1Keerthana.P, PG-Scholar, VLSI Design, Regional centre – Madurai Anna University, Madurai, India.
2Sivasankar.A, Assistant Professor, Department of Electronics and Communication Engineering, Regional Centre – Madurai Anna University, Madurai, India.
Manuscript received on June 05, 2013. | Revised Manuscript received on June 11, 2013. | Manuscript published on June 15, 2013. | PP: 35-37 | Volume-1 Issue-7, June 2013. | Retrieval Number: G0340061713
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: In the past, scientific data have been almost exclusively compressed by means of lossless methods, in order to preserve their full quality. However, more recently, there has been an increasing interest in the lossy compression which has not yet globally accepted by the remote sensing community, mainly because it is sensed that the lossy compressed images may affect the results of posterior processing stages. Hence here, the influence of lossy compression on two standard approaches for hyperspectral data exploitation known as adaptive spectral unmixing, and supervised classification using PCA are considered. The experimental result states that the adaptive spectral unmixing provides a user defined spatial scale which improves the process of extraction of end members and PCA improves the classification accuracy. It is also observed that, for certain compression techniques, a higher compression ratio may lead to more accurate classification results. This work further provides recommendations on best practices when applying lossy compression prior to hyperspectral data classification and/or unmixing.
Keywords: Hyperspectral data lossy compression, end member extraction, adaptive spectral unmixing, wavelet transform, support vector machine (SVM), Principal component analysis