Vegetation feature extraction and selection method based on long time sequence
A long-term sequence and feature extraction technology, applied in the field of image processing, can solve problems such as the inability to extract and distinguish ground vegetation, achieve the effect of improving classification accuracy and speed, and improving discrimination
Pending Publication Date: 2022-04-12
NANJING FORESTRY UNIV
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Faced with the diversity of vegetation and the density of vegetation coverage, traditional and single remote sensing data cannot extract and distinguish surface vegetation
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
[0044] The following examples are only used to more clearly illustrate the technical solution of the present invention, and the flow chart is as figure 1 :
[0045] Step 1. Preprocess the 12 Sentinel-2 time-series images in each month of the year and one Sentinel-1 image in November of the same year to ensure that the coordinate system and coverage of the 13 image data are unified, and to improve the accuracy of the images. Visual Interpretation Effect. Sentinel-1 needs to extract the covariance matrix C2 to facilitate subsequent polarization decomposition, while for each scene Sentinel-2 images need to be processed in batches including geometric correction, radiation correction and cropping to ensure that the range of image data of 13 scenes is consistent and the effect is clear;
[0046] Step 2, perform dual polarization decomposition on the preprocessed Sentinel-1 polarization image, and extract 5 decomposition features, 1 polarization vegetation index, 9 texture features ...
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 vegetation feature extraction and selection method based on a long time sequence, and the method comprises the steps: carrying out the preprocessing of 12-scene Sentinel-2 time sequence images of each month in one year and a scene Sentinel-1 image of November in the same year, carrying out the dual-polarization decomposition, and extracting 17 features, namely a polarization vegetation index, a texture feature and a backscattering coefficient; carrying out B2-B8, B11 and B12 wave band extraction and carrying out resampling; extracting an NDVI mean value and an IRECI mean value of each scene image data of the 12 scenes, drawing a broken line according to time, and extracting a VFC mean value; respectively carrying out time sequence broken line interpolation and smooth filtering, and extracting 14 related phenological parameters; obtaining a novel NDVI time sequence broken line and an IRECI time sequence broken line based on VFC, and extracting 14 novel phenological parameters from the novel NDVI time sequence broken line and the IRECI time sequence broken line; and according to an SVM sample training result, performing feature selection on the extracted 17 polarization features and 28 optical features by using an artificial bee colony feature selection algorithm, and screening out a classified optimal feature subset.
Description
technical field [0001] The invention belongs to the field of image processing, and mainly relates to feature extraction and selection of remote sensing image data. feature selection method. Background technique [0002] As surface organisms covering 31.7% of the earth's surface, vegetation is closely related to natural environmental elements such as climate, soil, topography, animal kingdom and water. Faced with the diversity of vegetation and the density of vegetation coverage, traditional and single remote sensing data cannot extract and distinguish surface vegetation. For this reason, according to the growth cycle of different vegetation, the present invention combines vegetation phenology information with vegetation coverage in feature extraction to propose a new type of characteristic parameter, fully considering the growth form and growth state of vegetation. The red edge of the Sentinel-2 image and the short-wave band have a good correlation with the vegetation leaf...
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): G06V10/46G06V10/764G06K9/62G06N3/00
Inventor 颜瑾陈媛媛郑加柱向云飞魏浩翰
Owner NANJING FORESTRY UNIV