Effects of compression on the classification of hyperspectral images
Euisun Choi, Sangwook Lee, Chulhee Lee · International Conference on Systems · 2008
In this paper, we investigate the effects of compression on the classification of hyperspectral images. It has been reported reconstructed images after compression often increase classification accuracy. In this paper, by using various experiments and analyses, we examine the causes and implications of this phenomenon. In particular, we used the three dimensional version of the set partitioning in hierarchical trees (SPIHT) algorithm as a compression method. We compared the classification accuracies in the original images and the reconstructed images at various bit rates. The experimental results and analyses indicated that higher classification accuracy did not necessarily mean improved performance since the classification results might differ from those of original data.