Hyperspectral image classification method based on guiding filtering and linear spatial correlation information
A hyperspectral image and linear space technology, applied in the field of hyperspectral image classification based on guided filtering and linear spatial correlation information, can solve the problems of ignoring the auxiliary function of spatial correlation information and insufficient mining of spatial texture information, etc. Spatial correlation, the effect of improving classification accuracy
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0030] figure 1 It is a schematic flow chart of an embodiment of a hyperspectral image classification method based on guided filtering and linear spatial correlation information. Such as figure 1 As shown, a hyperspectral image classification method based on guided filtering and linear spatial correlation information, including:
[0031] S1, receiving a hyperspectral image dataset D;
[0032] S2, obtaining spatial texture information D according to the hyperspectral image data set s ;
[0033] In this embodiment, the spatial texture information D is obtained according to the hyperspectral image data set s The steps include: processing the hyperspectral dataset D through PCA dimensionality reduction to obtain a hyperspectral image dataset D with redistributed information PCA ; use guided filtering to D PCA The first 20 components are filtered to obtain the spatial texture information D s .
[0034] S3, obtaining linear spatial correlation information D according to the ...
Embodiment 2
[0054] The Indian agriculture and forestry hyperspectral dataset is used for testing. Among them, Indian agriculture and forestry comes from the spectrometer (AirborneVisible Infrared Imaging Spectrometer), which is a hyperspectral remote sensing image collected in Indiana in northwest Indiana in 1992. It has a spatial resolution of 20 meters and contains 144×144 pixels, 220 Bands, 20 bands are removed due to factors such as noise and water absorption, and the remaining 200 bands include 16 types of vegetation. Select all 16 categories, and randomly select 10% of each category to form a labeled training set, and the remaining 90% as For the test set, see Table 1 for the specific object types and the number of samples;
[0055] Table 1 Statistics of image classification data of Indian agriculture and forestry dataset
[0056]
[0057] The overall classification accuracy (Overall accuracy, OA), average classification accuracy (Average accuracy, AA) and Kappa statistical coef...
Embodiment 3
[0062] The Salinas Valley hyperspectral data set was used for testing. The Salinas Valley: from the Airborne Visible Infrared Imaging Spectrometer (Airborne Visible Infrared Imaging Spectrometer) was an image collected in the Salinas Valley, California, USA in 1992. With a spatial resolution of 3.7 meters, it contains 512×217 pixels and 224 bands. Due to factors such as noise and water absorption, 20 bands are removed, and the remaining 204 bands include 16 vegetation types. All 16 categories are selected Among them, 1% samples of each category are randomly selected to form a labeled training set, and the remaining 99% are used as a test set. See Table 2 for specific object categories and sample numbers;
[0063] Table 2 Salinas Valley dataset image classification data statistics
[0064]
[0065] The present embodiment adopts overall classification accuracy (Overall accuracy, OA), average classification accuracy (Average accuracy, AA) and Kappa statistical coefficient (Kap...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More