Research on Dimensionality Reduction of Hyperspectral Image under Close Range

Hyperspectral imaging technology can make detailed division of data in both spatial and imaging bands. At the same time, due to the existence of noise factors, the data formed by imaging will have some redundancy. In order to explore the data redundancy and its impact on the subsequent analysis process, this paper analyzes the mechanism of data redundancy formation, including spatial correlation and inter-spectral correlation of hyperspectral images. At the same time, the process of data dimensionality reduction is discussed. In order to reduce the long-distance interference factors, the noise analysis of the forest image of close-range imaging was carried out during the experiment, and the model was used to reduce the dimension. The results show that the processed data dimension is greatly reduced, and the quality improvement is more obvious. Hyperspectral image dimensionality reduction can provide effective support for further computational analysis.