A hyperspectral image classification method and system
By extracting regions at multiple scales and modeling features, and combining fine-grained, wide-area feature vector integration, the problems of fragmented patches and blurred boundaries in hyperspectral image classification are solved, achieving high-precision pixel-level classification.
CN122391844APending Publication Date: 2026-07-14COALFIELD GEOLOGICAL CENTER OF THE GEOLOGICAL BUREAU OF XINJIANG UYGUR AUTONOMOUS REGION +1
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- COALFIELD GEOLOGICAL CENTER OF THE GEOLOGICAL BUREAU OF XINJIANG UYGUR AUTONOMOUS REGION
- Filing Date
- 2026-03-17
- Publication Date
- 2026-07-14
Smart Images
Figure CN122391844A_ABST
Abstract
The application provides a hyperspectral image classification method and system, and relates to the technical field of remote sensing image processing.The method comprises the following steps: performing multi-scale region extraction on an original three-dimensional hyperspectral image to obtain a central region and a wide-area surrounding region of the original three-dimensional hyperspectral image, and performing dimension reduction on the central region and the wide-area surrounding region to obtain a low-dimensional feature vector of the central region and a low-dimensional feature vector of the wide-area surrounding region; performing feature modeling on the low-dimensional feature vector of the central region to obtain a fine-grained feature vector, and performing modeling optimization on the low-dimensional feature vector of the central region to obtain a high-order feature vector; performing feature enhancement and context feature expression on the low-dimensional feature vector of the wide-area surrounding region to obtain a wide-area feature vector; and integrating the fine-grained feature vector, the high-order feature vector and the wide-area feature vector to obtain a pixel-level classification map.The application improves the classification accuracy of the hyperspectral image.
Need to check novelty before this filing date? Find Prior Art