Remote sensing image semantic segmentation method combining deep learning and random forest
What is Al technical title?
Al technical title is built by PatSnap Al team. It summarizes the technical point description of the patent document.
A semantic segmentation and random forest technology, applied in neural learning methods, computer parts, character and pattern recognition, etc., can solve the problems of low segmentation accuracy and limited versatility of methods, so as to improve classification accuracy, improve efficiency and universality. Sex, the effect of less bands
Pending Publication Date: 2020-07-31
CAPITAL NORMAL UNIVERSITY
View PDF6 Cites 16 Cited by
Summary
Abstract
Description
Claims
Application Information
AI Technical Summary
This helps you quickly interpret patents by identifying the three key elements:
Problems solved by technology
Method used
Benefits of technology
Problems solved by technology
Although this invention can efficiently and quickly realize the pixel-level classification of various ground objects in high-resolution remote sensing images, and simplifies the complex process of traditional classification methods, the segmentation accuracy is low and the generality of the method is limited.
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 more
Image
Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
Click on the blue label to locate the original text in one second.
Reading with bidirectional positioning of images and text.
Smart Image
Examples
Experimental program
Comparison scheme
Effect test
Embodiment 1
[0065] as attached figure 1 As shown, a remote sensing image semantic segmentation method combining deep learning and random forest includes the following steps:
[0066] 101. Create a training data set for the research area, using samples and sample labels as the training data set.
[0067] Acquisition of samples and sample labels: First obtain the GF-2 image, synthesize the red, green and blue bands of the image, and crop the composite image into a sample. The cropping specification is 512*512, as attached figure 2 As shown, mark on the labelme software, as attached image 3 As shown, the saved data is a json file, and the json file is converted into a dataset file. At this time, the label data is 8-bit depth label data, and finally it is given "true color" and converted into a 24-bit depth sample label file. as attached Figure 4 shown.
[0068] The training data set in the research area uses 512*512*24 bit depth samples and sample labels as the specifications of the t...
Embodiment 2
[0109] Based on the above-mentioned embodiment 1, as attached Figure 5 As shown, a GF-2 optical image data is used as the data source for the data of the study area. The samples and sample labels in the above-mentioned embodiment 1 are used as the training data set.
[0110] Use the Python language to write a fully convolutional neural network model based on the TensorFlow framework, use the above samples and sample labels to train the model, use the trained model to test the data in the research area, extract its deep features, and visualize it.
[0111] Use the Band Math function of ENVI5.3 software to extract the normalized vegetation index and normalized water index of GF-2 image data according to the formula (3) and formula (4); the Band Math function is the pixel value of each pixel Perform numerical operations.
[0112] Use the Co-occurrence Measures function of ENVI5.3 software to extract the variance, entropy, dissimilarity and angular second-order moment of four t...
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 relates to a remote sensing image semantic segmentation method combining deep learning and a random forest, and the method comprises the following steps: making a training data set of aresearch region, and employing samples and sample labels as the training data set; establishing a full convolutional neural network model, and training the model by using the sample and the sample label; extracting deep features of the research area by using a full convolutional neural network model; meanwhile, extracting shallow features of the GF-2 image of the research area; performing multi-feature combination on the deep features and the shallow features; performing semantic segmentation by adopting a random forest. According to the method, through the manufacturing of the data set, the used image wavebands are few, easy to obtain, high in universality and high in segmentation precision; the deep learning and random forests are combined, shallow features and deep features are innovatively fused in the method, the shallow features and the deep features are combined and complemented with each other, and the defects of a single method are overcome; the method has a good effect on remote sensing image semantic segmentation, the classification precision is effectively improved, and the method has a good effect on extraction of water bodies, vegetation and impermeable layers.
Description
technical field [0001] The invention relates to the technical field of remote sensing image classification, in particular to a remote sensing image semantic segmentation method combined with deep learning and random forest. Background technique [0002] Ground object information has always been extremely important information in remote sensing images. At present, the resolution of remote sensing images has been greatly improved. High-resolution remote sensing images have rich and fine ground feature information, and the details of ground features are clearer, which provides a good research basis for the extraction of ground feature information in remote sensing images. But at the same time, the structure of ground objects is more complicated, and the interference information is more difficult to deal with. [0003] Semantic segmentation of remote sensing images is the pixel-level classification of images, and is an important research direction in the application of remote s...
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
Application Date:The date an application was filed.
Publication Date:The date a patent or application was officially published.
First Publication Date:The earliest publication date of a patent with the same application number.
Issue Date:Publication date of the patent grant document.
PCT Entry Date:The Entry date of PCT National Phase.
Estimated Expiry Date:The statutory expiry date of a patent right according to the Patent Law, and it is the longest term of protection that the patent right can achieve without the termination of the patent right due to other reasons(Term extension factor has been taken into account ).
Invalid Date:Actual expiry date is based on effective date or publication date of legal transaction data of invalid patent.
Login to view more 
Login to view more