A 3D non-linear orientation prediction wavelet transform for interference hyperspectral images compression

According to the characteristic of large aperture static interference imaging spectrometer (LASIS), a non-linear orientation prediction three-dimensional (3D) wavelet transform method is proposed in this paper on the basis of the 3D orientation prediction wavelet transform method proposed by Li, Ma and Wu in 2008 [1]. The method proposed in this paper still combines directional prediction into 3D lifting wavelet transformation, but compared with the 3D orientation prediction wavelet transform method, it made a breakthrough in the limitation that the orientation predicted must be a straight linear direction. The experimental results showed that this method improved the performance of wavelet obviously, especially on the LASIS image with quite severe and unstable directional characteristic as seen in the study of Ma and Ma (2009) [2]. Meanwhile, the characteristic of spectrum image recovered by the proposed method also possessed better performance.