Hyperspectral image classification method and device
An image classification and hyperspectral technology, which is applied in the direction of instruments, biological neural network models, character and pattern recognition, etc., can solve the problem of low classification accuracy, achieve the effect of improving classification accuracy, simple implementation process, and high classification accuracy
Pending Publication Date: 2021-12-03
PLA STRATEGIC SUPPORT FORCE INFORMATION ENG UNIV PLA SSF IEU
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0006] The purpose of the present invention is to provide a hyperspectral image classification method and device to solve the problem of low classification accuracy of existing hyperspectral image classification methods
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
[0023] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments.
[0024] Method example:
[0025] In this embodiment, the trained spatial feature extraction unit is obtained by sequentially connecting the first 5 convolutional layers in the trained VGG19 network model. trained on the ImageNet dataset.
[0026] Among them, the structure of the VGG19 network model is as follows figure 1 As shown, CONV3-64 in the first dashed box indicates that 64 convolution kernels with a size of 3×3 are used in the convolutional layer, Block*2 indicates that the previous convolutional layer is repeated twice, and MaxPool indicates the maximum pooling layer , the other dotted boxes are similar to this, FC4096 represents a fully connected layer with 4096 hidden nodes, FC1000 represents a fully connected layer with 1000 hidden nod...
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 provides a hyperspectral image classification method and device, and belongs to the technical field of remote sensing image processing and application. The method comprises the following steps: preprocessing a to-be-classified hyperspectral image, so that the space size and the spectral segment number of the to-be-classified hyperspectral image meet the input requirements of a spatial feature extraction unit; enabling the spatial feature extraction unit to comprise n convolution layers connected in sequence, and n is greater than or equal to 2; inputting the preprocessed hyperspectral image into a trained spatial feature extraction unit, unifying the spatial size of each convolutional layer output image into the spatial size of the to-be-classified hyperspectral image, and splicing the convolutional layer output images after the spatial size unification to obtain a spatial feature map of the to-be-classified hyperspectral image; splicing the spatial feature map with the to-be-classified hyperspectral image to obtain a spatial-spectral feature map of the to-be-classified hyperspectral image; and inputting the space-spectrum feature map into a trained classification model to obtain a classification result of the to-be-classified hyperspectral image. According to the invention, the classification precision of the hyperspectral image can be improved.
Description
technical field [0001] The invention relates to a hyperspectral image classification method and device, and belongs to the technical field of remote sensing image processing and application. Background technique [0002] Hyperspectral image classification has been widely used in environmental monitoring, precision agriculture, mineral exploration and other fields. Among them, feature extraction is an important step in hyperspectral image classification tasks, and the quality of feature extraction results directly determines the quality of classification results. [0003] Traditional hyperspectral image classification methods rely on manual design of feature extraction rules to achieve ideal classification results. Commonly used feature extraction methods include: extended morphological attribute profile, Gabor feature, local binary mode, etc. Although these artificially designed feature extraction rules combined with classifiers such as support vector machines can achieve ...
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 Applications(China)
IPC IPC(8): G06K9/62G06N3/04
CPCG06N3/045G06F18/24323G06F18/214
Inventor 刘冰余岸竹王瑞瑞郭文月秦进春
Owner PLA STRATEGIC SUPPORT FORCE INFORMATION ENG UNIV PLA SSF IEU