高光谱图象因其信息丰富、地物辨识能力强在工农业和国防等领域都有着广泛的应用前景，然而其成象质量在获取传输过程中受诸多因素影响，导致其获得的图象质量退化或某些信息损失，进而限制其进一步利用。针对该类问题，本项目首先从物理层对高光谱成象链中的像素级污染、条带缺失和垃圾波段等退化的产生机理进行深入分析，论证高光谱图象的三维稀疏特性，以及利用受损后数据估计原始稀疏表达的可行性。在此基础上，建立适合高光谱图象数据结构和应用特征的三维稀疏模型，并以此为重点，系统研究应对多种退化的恢复新技术。与此同时，探索独立数据的非相似性恢复效果评价体系，直接判断恢复数据的光谱域、空间域形态质量及其应用价值。本研究技术体系的实施，将极大提高现有高光谱数据源的成象质量和应用潜力，更好的造福人类。同时，也有助于高光谱数据表征、恢复技术和三维稀疏理论的发展。

Hyperspectral Images have a broad application prospect in many fields such as industrial and agricultural production and national defense due to the rich information and feature recognition ability. However, the imaging quality is affected by kinds of factors in the progress of acquisition and transmission, which leads to image quality degradation,loss of certain information and the fact limits the further use of hyperspectral images. Aim at the above problems, the project first analyze the generation mechanism of pixel-level contamination, corrupted stripes and junk bands in hyperspectral imaging chain deeply from the physical layer; then demonstrate three dimensional sparse characteristics of hyperspectral images and the feasibility of estimating original sparse expression using damaged data. On this basis, we would like to establish a three dimensional sparse model suitable for hyperspectral image data structure and application characteristics and systematic study a recovery technique able to respond to a variety of degradation comprehensively; meanwhile, explore an independent data non-similarity recovery effects evaluation system which directly judge the quality and application value of the spectral and space domain of the recovered data. The implementation of the technical system in our study will greatly increase the degree of imaging quality and potential applications of hyperspectral data source, thus benefit the mankind. At the same time, the project is also helpful to the development of hyperspectral data representation, recovery technology and three dimensional sparse theories.