Hyperspectral data dimensionality reduction method based on tensor distance patch alignment
A data dimensionality reduction and hyperspectral technology, applied in the field of hyperspectral remote sensing image processing, can solve the problems of data internal structure damage, high dimensionality, dimensionality disaster, etc.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Example Embodiment
[0077] Example 1: The dimensionality reduction method is aimed at the tensor characteristics of hyperspectral data. First, the “window area” is used to convert the hyperspectral data into a tensor form through the central pixel and other pixels around it. , Maintain the spatial information of each pixel; second, introduce the tensor distance to construct a high-quality tensor neighbor graph containing discriminative information; third, obtain the global optimal spectrum according to the patch calibration framework extended to the tensor space- Spatial information; fourth, the solution of the quantum space is obtained by using the alternating least squares algorithm; finally, the category of each sample is determined according to the tensor nearest neighbor method;
[0078] Specific steps are as follows:
[0079] Step 1. Select the hyperspectral data to be analyzed, and convert the hyperspectral data into tensor form according to the window area;
[0080] Step 2. Calculate the tenso...
Example Embodiment
[0126] Example 2: Through the AVIRIS92AV3C hyperspectral data experiment, the TDPA proposed by the present invention is compared with the existing MDA and MPCA tensor dimensionality reduction algorithms. For the fairness of comparison, it is coordinated when seeking the nearest neighbor distance of the high-quality tensor neighbor graph The parameter β=1. And according to the tensor nearest neighbor method to distinguish the type of each test sample, each experiment is done 20 times, and the average value is taken. Prove the superiority of TDPA.
[0127] Combine figure 1 . The figure shows the key steps of using the tensor distance patch calibration method to reduce the dimensionality of hyperspectral data. It mainly includes seven steps: First: Calculate the tensor distance d between training samples TD ; Second: construct a high-quality tensor neighbor graph G according to the tensor distance; third: select χ according to the high-quality tensor neighbor graph i Patch sample ...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap