Effects of Hyperspectral Image Compression Methods on Classification

Abstract

Because of having large data size in hyperspectral imaging, compression becomes an important necessity in terms of the transmission and storage of the data. There are several compression approaches proposed in the literature, and it is not enough to use only quality metrics in comparison. Depending on the application area, different compression methods can provide advantages to each other. In this paper, performances of PCA+JPEG2000, DWT+JPEG2000, 3DTARP, 3DTARP and JPEG2000 compression methods, which are frequently used in literature, are evaluated based on their classification performances.