Unlock instant, AI-driven research and patent intelligence for your innovation.
A Semi-Supervised Classification Method Combining Hybrid Cell Decomposition and Active Learning
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 technology of mixed pixel decomposition and active learning, applied in the field of hyperspectral remote sensing, can solve problems such as insufficient precision, achieve the effects of improving precision, reducing required time, and reducing workload
Active Publication Date: 2018-07-17
INST OF REMOTE SENSING & DIGITAL EARTH CHINESE ACADEMY OF SCI
View PDF2 Cites 0 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
[0005] The technical problem to be solved by the present invention is to solve the problem of insufficient accuracy of the existing hyperspectral semi-supervised classification method in the case of fewer training samples
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
[0045] Such as figure 1 As shown, this embodiment provides a semi-supervised classification method combining mixed pixel decomposition and active learning, and the specific steps are as follows:
[0046] S11. Construct a labeled sample set and an unlabeled sample set, and set active learning parameters. The sub-steps in step S11 are specifically described as follows:
[0047] S111. Construct a labeled sample set and an unlabeled sample set.
[0048] The image samples of the image to be classified can be divided into labeled samples and unlabeled samples. Among them, each image sample corresponds to a pixel, the marked sample represents the pixel that has been marked, and the unlabeled sample represents the pixel that has not been marked.
[0049] From the images to be classified, select samples of known categories as marked samples to form a marked sample set S T . The marked samples are represented by formula (1):
[0050] (x 1 ,y 1 ),(x 2 ,y 2 )...,(x N ,y N ) (1...
Embodiment 2
[0082] In this embodiment, the method provided in Embodiment 1 is used to classify aerial hyperspectral images, and the specific description is as follows.
[0083] The aerial hyperspectral data images obtained by the scanning imaging spectrometer PHI (Pushbroom Hyperspectral Imager) are as follows: figure 2 As shown in , the number of bands is 80, and the spatial resolution is 1.7m.
[0084] First, from figure 2 In the image shown, samples of known classes are selected as labeled samples. Wherein, the number of samples N=5, and the number of categories n=8.
[0085] Second, set active learning parameters. Wherein, the number of samples added by active learning M=160; the number of active learning iterations t=4; the weight parameter w=0.5.
[0086] According to step S2 to step S3 in embodiment one, construct active learning sample set;
[0087] According to step S4 in Embodiment 1, the sample set S will be actively learned A The samples are merged into the marked samp...
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 the technical field of hyperspectral remote sensing, and discloses a semi-supervised classification method combining mixed pixel decomposition and active learning. The method includes: in the image samples to be classified, selecting samples of known categories as marked samples to form a marked sample set; performing mixed pixel decomposition on unmarked samples in the image to obtain sample abundance information; according to the marked samples and The abundance information of unlabeled samples is used to construct an active learning sample set; the samples of the active learning sample set are merged into the marked sample set, and the merged marked sample set is used to classify the image to obtain the classification result. By combining mixed pixel decomposition and active learning to classify images, the accuracy of classification can be improved in the case of fewer samples, the workload of sample labeling can be effectively reduced, and the time required for classifier training can be reduced.
Description
technical field [0001] The invention relates to the technical field of hyperspectral remote sensing, in particular to a semi-supervised classification method combining mixed pixel decomposition and active learning. Background technique [0002] Since hyperspectral remote sensing data has hundreds of spectral bands, the demand for training samples in supervised classification has increased significantly, and obtaining enough training samples often requires more time and effort, especially for completely unknown research areas. , Ground surveys need to consume a lot of manpower and material resources. Therefore, in the case of only a small number of labeled samples, how to use certain labeled samples to mine the potential labels of unlabeled samples and add them to the classifier to assist in classification, thereby improving the performance of the classifier, has become a hyperspectral One of the important problems of data classification. [0003] It is against this backgro...
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
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62
CPCG06F18/2413
Inventor 张霞张立福刘佳王树东孙艳丽
Owner INST OF REMOTE SENSING & DIGITAL EARTH CHINESE ACADEMY OF SCI
Login to View More 
Login to View More