Hyperspectral Image Classification Method Based on SRCM and Convolutional Neural Network
A convolutional neural network and hyperspectral image technology, applied in the field of image processing, can solve the problems of missing band information, small amount of characteristic information of hyperspectral images, and inability to comprehensively utilize characteristic information, so as to improve the ability of feature expression and the degree of discrimination , the effect of improving the classification accuracy
Active Publication Date: 2021-09-03
XIDIAN UNIV
View PDF8 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The disadvantage of this method is that the principal component analysis dimensionality reduction breaks the connection between the spectral bands of the hyperspectral image to be classified, loses the original band information of the hyperspectral image to be classified, and reduces the high The completeness of the spectral features of the spectral image affects the expressive ability of the spectral features of the hyperspectral image to be classified
The disadvantage of this method is that the deep convolutional neural network cannot comprehensively utilize the feature information of different scales extracted by different layers, resulting in a small amount of feature information of the hyperspectral image to be classified, which affects the hyperspectral image to be classified. Classification Accuracy of Spectral Images
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 Construction
[0048] The present invention will be further described below in conjunction with the accompanying drawings.
[0049] refer to figure 1 , to further describe the specific steps of the present invention.
[0050] Step 1, construct a convolutional neural network.
[0051] refer to figure 2 , to further describe the structure of the constructed convolutional neural network.
[0052] Construct a 20-layer convolutional neural network, and its structure is as follows: input layer→1st convolutional layer→1st pooling layer→2nd convolutional layer→2nd pooling layer→3rd volume Product layer → 3rd pooling layer → 4th convolutional layer → 4th pooling layer, 1st pooling layer → 5th convolutional layer → 1st fully connected layer, 2nd pooling Layer → 6th convolutional layer → 2nd fully connected layer, 3rd pooling layer → 7th convolutional layer → 3rd fully connected layer, 4th pooling layer → 8th convolutional layer →4th fully connected layer, 1st fully connected layer→feature cascad...
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
A hyperspectral image classification method based on spectral reflectance curve matrix SRCM and convolutional neural network, which mainly solves the problem of low classification accuracy of hyperspectral images in the prior art. The specific steps of the present invention are as follows: (1) construct a convolutional neural network; (2) add noise to the hyperspectral image to be classified; (3) normalize the image data by band; (4) generate a set of spatial spectral feature matrices; 5) Generate a set of stacked spatial spectral reflectance curve matrices; (6) generate a training data set and a test data set; (7) train a convolutional neural network; (8) classify the test data set. The present invention utilizes a convolutional neural network that fuses features of different layers to perform feature learning on the stacked spatial spectral reflectance curve matrix of hyperspectral images, thereby performing classification, and has the advantage of high precision for hyperspectral image classification problems.
Description
technical field [0001] The invention belongs to the technical field of image processing, and further relates to a hyperspectral image classification method based on a spectral reflectance curve matrix SRCM (Spectral Reflectance Curve Matrix) and a convolutional neural network in the technical field of target recognition. The invention can be used in the fields of agricultural remote sensing, map drawing, environmental monitoring, cancer detection, vegetation survey and the like for ground object recognition. Background technique [0002] Hyperspectral images can describe the two-dimensional radiation information and spectral information of the spatial distribution of ground objects at the same time, forming a unique spectrogram with triple information of space, radiation and spectrum, which has been widely used in the field of remote sensing applications. Classification is an important content in hyperspectral image processing technology, and its ultimate goal is to assign a...
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 Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/241G06F18/253
Inventor 王桂婷李诗卉公茂果钟桦吴飞杨晓婕陈贝贝马锐解玮
Owner XIDIAN UNIV