Optical remote sensing image vegetation and water body information automatic extraction method

Abstract

The invention discloses an optical remote sensing image vegetation and water body information automatic extraction method comprising the following steps: obtaining an optical remote sensing data sample, and randomly extracting 10% of the optical remote sensing data sample as a sample subset; calculating normalized vegetation indexes and normalized water indexes of all the samples in the sample subset, obtaining a general characteristic spectrum to execute supervised classification based on a minimum spectral angle on all the samples in the sample subset, and recording vegetation types, water body types and the sizes of the minimum spectral angles corresponding to other types at the same time; taking the minimum first 50% of samples in the minimum spectral angles, performing k-means unsupervised classification based on the minimum Euclidean distance on the first 50% of samples, obtaining 10 characteristic spectrums for each type, totally 30 characteristic spectrums, and performing supervised classification based on the minimum Euclidean distance pixel by pixel on the global image to obtain an extraction result of vegetation and water. Any prior sample is not needed for supporting from beginning to end, manual intervention is avoided, and full-automatic extraction of vegetation and water body information is achieved.

Description

technical field [0001] The invention relates to the technical field of remote sensing image extraction, in particular to a method for automatically extracting vegetation and water body information from optical remote sensing images. Background technique [0002] Supervised classification method: also known as training field method and training classification method, it is based on the establishment of statistical recognition functions as the theoretical basis and classification technology based on typical sample training methods, that is, according to the samples provided by known training areas, by selecting characteristic parameters, find It is a method of pattern recognition to extract feature parameters as decision rules and establish discriminant functions to classify images to be classified. The training area is required to be typical and representative. If the discriminant criterion satisfies the classification accuracy requirement, the criterion is established; othe...

AI Technical Summary

Problems solved by technology

[0005] Among the above three methods, the supervised classification method and the random forest classification method both require training samples, but require a long training and learning process, and the extrapolation and generalization capabilities are insufficient. Although the unsupervised classification method does not require training samples, its classification The result of this method is only to distinguish different categories, but it cannot determine the attributes of the categories, and the classification results often cannot meet the actual needs.

Images