Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Hyperspectral image classification method and system based on kernel-guided variable convolution and dual-window joint bilateral filter

A bilateral filter and classification method technology, applied in the field of hyperspectral image classification, can solve the problems of salt and pepper noise regional level, misclassification, inability to extract spatial spectrum information, etc., to improve classification accuracy, reduce inter-class similarity and intra-class similarity. Volatility, the effect of improving overall classification performance

Active Publication Date: 2022-03-22
HARBIN ENG UNIV
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The present invention is to solve the problems existing in the existing hyperspectral image classification methods, such as salt and pepper noise, region-level misclassification, and the inability to extract appropriate spatial spectrum information.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Hyperspectral image classification method and system based on kernel-guided variable convolution and dual-window joint bilateral filter
  • Hyperspectral image classification method and system based on kernel-guided variable convolution and dual-window joint bilateral filter
  • Hyperspectral image classification method and system based on kernel-guided variable convolution and dual-window joint bilateral filter

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0049] Aiming at the existing problems of hyperspectral image classification, the present invention proposes a hyperspectral image classification algorithm (KDCDWBF) framework based on kernel-guided variable convolution and dual-window joint filtering, figure 1 As shown, the present invention can combine the advantages of kernel-guided variable convolution and dual-window joint bilateral filter.

[0050] The hyperspectral image classification method based on kernel-guided variable convolution and dual-window joint bilateral filter described in this embodiment includes the following steps:

[0051] The hyperspectral image classification model based on kernel-guided variable convolution and dual-window joint filtering mainly includes two stages of feature extraction and reclassification. The feature extraction stage in the flowchart is the first stage of the algorithm, and the reclassification stage is the second stage of the algorithm.

[0052] 1) In the first stage, the size ...

specific Embodiment approach 2

[0092] The hyperspectral map classification system based on kernel-guided variable convolution and dual-window joint bilateral filter described in this embodiment is used to perform the hyperspectral classification based on kernel-guided variable convolution and dual-window joint bilateral filter Graph classification method.

Embodiment

[0094] Two hyperspectral data sets, PaviaU and SD, are used to illustrate the effect of the hyperspectral image classification algorithm based on kernel-guided variable convolution and dual-window joint filtering proposed by the present invention. The properties of these two datasets are listed in Table 1. The present invention randomly selects 1% and 5% of samples as training and verification data sets, and 94% of samples as test data sets. The experimental results use overall classification accuracy (OA), average classification accuracy (AA) and Kappa coefficient as measurement indicators, and the higher the value of the three indicators, the better the classification result.

[0095] Table 1 Detailed properties of two hyperspectral datasets PaviaU and SD

[0096]

[0097] Although δ s ,δ r Thresholds are parameters that can be adjusted and can affect the final classification results, but it is not necessary to perform parameter analysis on all parameters. Because eve...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

A hyperspectral image classification method and system based on kernel-guided variable convolution and double-window combined bilateral filter belong to the technical field of hyperspectral image classification. In order to solve the problems of existing hyperspectral image classification methods, salt and pepper noise and region-level misclassification, as well as the inability to extract appropriate spatial spectral information. The invention firstly utilizes multi-layer kernels to guide variable convolution layers to form a feature extraction network to extract accurate spatial spectrum features, and obtain an initial classification probability map; A double-window joint bilateral filter is used for classification; the above operation is performed on any pixel in the initial classification probability map, and the output result is the final classification probability map; the final classification probability map is subjected to the maximum probability value to obtain the final Classification result graph. Mainly used for the classification of hyperspectral images.

Description

technical field [0001] The invention relates to a hyperspectral image classification method and system, and belongs to the technical field of hyperspectral image classification. Background technique [0002] Hyperspectral images contain hundreds of different bands, which not only contain rich spectral information, but also spatial structure information of ground objects. Hyperspectral remote sensing images have the following characteristics: (1) hyperspectral images, the spectral range extends from visible light to near-infrared, mid-infrared and even far-infrared. In addition, hyperspectral images contain dozens or even hundreds of bands, the spectral sampling interval is small, and the spectral resolution reaches the nanometer level. The high spectral coverage and fine spectral resolution mean that the global features of the object spectrum will be more complete and the local subtle features will be more obvious. (2) The hyperspectral image has a large amount of data and...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06V10/764G06V10/74G06V10/774G06V10/30G06V20/10G06V10/82G06K9/62G06N3/04G06N3/08G06V10/58
CPCG06N3/08G06V20/194G06V10/30G06N3/045G06F18/22G06F18/2415G06F18/214
Inventor 冯收朱文祥赵春晖吴丹秦博奥成浩樊元泽丰瑞
Owner HARBIN ENG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com