Compression of multi-polarimetric SAR intensity images based on 3D-matrix transform

The algorithms for multi-polarimetric synthetic aperture radar (SAR) intensity image compression are investigated. First, the multi-polarimetric SAR intensity images (HH, HV and VV) are considered as a 3D-matrix unit, and then a 3D-matrix transform is adopted to remove the redundancies, which includes 1D discrete cosine transform (DCT) in the polarimetric channels and 2D discrete wavelet transform (DWT) in each polarimetric SAR image plane. After the 3D-matrix transform, two methods are proposed to encode the 3D mixed coefficients. One is a bit allocation encoding based on differential entropy and the other is a 3D set partitioning in hierarchical trees (SPIHT) encoding which is an improvement of the conventional SPIHT. The two methods can remove not only the redundancies of the image inside but also the redundancies among the polarimetric channels, because they do not process every channel image separately. Both the theory and experimental results show that the proposed methods are efficient.