Method for lossy compression of hyperspectral image based on classified DCT (discrete cosine transform)

• Main Authors: HAN RAN, WANG KEYAN, GE CHIRU, LI YUNSONG, HU ZIFAN, GUO JIE, ZHANG JING
• Format: Patent
• Language: English
• Published: 20.05.2015
• Subjects: ELECTRIC COMMUNICATION TECHNIQUE; ELECTRICITY; PICTORIAL COMMUNICATION, e.g. TELEVISION

The invention discloses a method for lossy compression of a hyperspectral image based on classified DCT (discrete cosine transform), which mainly solves the problems of incomplete compression decorrelation and poor compression effect because the spectrum vector correlation is not considered in the spectral transformation of the prior art. The method comprises the following steps of (1) performing the space two-dimensional wavelet transform on the hyperspectral image; (2) utilizing a classifying algorithm based on spectrum vector characteristics to classify the spectrum vectors consisting of space wavelet transform coefficients, obtaining a classification graph, and subtracting the average vector from each type of spectrum vector according to the classification graph, so as to obtain a residual vector; (3) utilizing the spectral one-dimensional DCT to transform the residual vector, so as to obtain a three-dimensional transform coefficient; (4) encoding the three-dimensional transform coefficient, so as to obtain a compression code stream with accurate and controllable code rates. The method has the advantages that the statistic property between the spectrums of the hyperspectral image is sufficiently utilized, the decorrelation is more thorough, the better compression property is obtained at the same code rate, and the method can be used for hyperspectral data treatment and transmission.