With the rapid development of spectral imaging technology in recent years

hyperspectral remote sensing images can provide abundant data on surface features. However

the sizable data of hyperspectral images make their storage

transmittal

and application quite difficult. As a result

how to validly code hyperspectral images has become a hot issue. The distributed source coding based on coset codes has received much attention because of its good compression performance and low coding complexity. In this study

we present a scheme for lossless distributed source coding of hyperspectral images based on adaptive quadtree segmentation. Assuming that the first frame in every group of spectrum frames is the key frame

the other frames are Wyner-Ziv frames. First

we perform adaptive quadtree segmentation on the key frame and then optimum linear prediction on each block of each Wyner-Ziv frame. Afterwards

the index of the coset codes to be transferred and the least significant bits of every pixel in this block are ascertained using prediction residuals. In this study

adaptive quadtree segmentation scheme has been proved to strengthen the adaptability of the formed coset codes. The proposed scheme can achieve a good compromise between coding efficiency and calculating complexity. This scheme is better able to meet the lossless coding requirement for hyperspectral images under low-complexity environment.