Hyperspectral image de-noising method based on Fisher dictionary learning and low-rank representation
A hyperspectral image and low-rank representation technology, applied in the field of hyperspectral image denoising, can solve the problems of single spectral information or spatial information denoising, lower data reliability, and insufficient denoising effect, achieving excellent denoising performance, High use value, effect of improving classification accuracy
Active Publication Date: 2017-10-20
NANJING NORMAL UNIVERSITY
View PDF4 Cites 7 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, in the process of collecting and transmitting hyperspectral images, they are often polluted by various types of noise, which greatly reduces the reliability of the data. Currently, hyperspectral images based on spectral signals and two-dimensional image denoising technology The denoising algorithm has achieved good results
However, due to the rich spectral information and spatial information of hyperspectral images, it is far from enough to denoise only by using spectral information or spatial 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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0112] This embodiment includes the following parts:
[0113] Step 1. Transform the data space:
[0114] In order to facilitate the comprehensive processing of data, it is necessary to transform the three-dimensional hyperspectral image data into a two-dimensional matrix of space-spectral joint.
[0115] For any hyperspectral image X∈R m×n×b , where m and n are the number of rows and columns of its spatial structure, respectively, and b is the number of bands. Record the value of each pixel of the hyperspectral image on all bands as a vector d h ∈R b (h=1,2,...,mn), then all pixels d h Put together to form a two-dimensional matrix D=[d 1 , d 2 ,...,d mn ]∈R b×mn .
[0116] Step 2. Learn the dictionary:
[0117] Obtain a new dictionary through the Fisher discriminant criterion, and satisfy the linear representation of the sub-dictionary corresponding to the class. The ability to represent samples of this class is strong and the ability to represent other classes is we...
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
Login to View More Abstract
The invention discloses a hyperspectral image de-noising method based on Fisher dictionary learning and low-rank representation. The method comprises the steps of transforming data space, learning dictionaries, replacing the dictionaries, improving LRR, inputting and outputting data and obtaining a noise-free image through reverse transformation. Through the method, various noises in a hyperspectral image can be effectively removed, and the data quality and application value of the hyperspectral image are improved. Besides, Fisher dictionary learning is adopted to judge whether parameters in a dictionary pair model in a dictionary replacement model have robustness, and therefore the method has high use value.
Description
technical field [0001] The invention relates to the technical field of hyperspectral image processing, in particular to a hyperspectral image denoising method based on Fisher dictionary learning and low-rank representation. Background technique [0002] Modern remote sensing technology originated in the 1960s. It is a discipline and technology that observes or obtains certain characteristic information by using the electromagnetic waves reflected or radiated by the receiving object or target without contacting the object or target being studied. Remote sensing technology has the advantages of not being restricted by factors such as geography, man-made and weather, and can continuously provide dynamic and large-scale observations of a variety of surface information. And many other fields have a wide range of applications. However, in the process of collecting and transmitting hyperspectral images, they are often polluted by various types of noise, which greatly reduces the ...
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
Login to View More IPC IPC(8): G06T5/00
CPCG06T2207/20172G06T5/70
Inventor 杨明俞珍秒吕静高阳
Owner NANJING NORMAL UNIVERSITY