SVM active learning classification algorithm for large-scale training data

Abstract

The invention relates to the crossing field of remote sensing classification and image information processing technology, especially relates to a SVM active learning classification algorithm for large-scale training data. The SVM active learning classification algorithm is based on clustering and uncertainty evaluation method, and selects, from a large number of samples, boundary samples which are far from cluster centroid and closer to two kinds of interface, and implements iterative optimization of classifier by introducing the active learning method. The process of selecting the boundary samples is not blind, but scientific, and the difference between the uncertain information and the distribution information of the sample can be compared through an iterative learning system, and a compression set can be automatically controlled and adjusted according to the comparison result. An optimal training sample set can be obtained by means of inversion derivation, thereby completing automatic classification of remote sensing images and improving the quality of classification.

Description

technical field [0001] The invention relates to the intersecting field of remote sensing classification and image image information processing technology, in particular to an SVM active learning classification algorithm for large-scale training data. Background technique [0002] Remote sensing images objectively and truly record and reflect the intensity of electromagnetic radiation of surface objects, and are a form of expression for remote sensing detection of surface object information. Classification of ground features using remote sensing images has important applications in the fields of urban monitoring, agricultural monitoring, soil survey and forestry monitoring. The existing remote sensing image classification methods mainly focus on using the spectral information of remote sensing image pixels (or supplemented by texture and other spatial information), and using clustering criteria such as distance, angle, probability, or support vector machines and neural networ...

AI Technical Summary

Problems solved by technology

[0004] At present, the sample selection in the large training sample set usually adopts a simple stratified equidistant sampling method, but because it does not use any information of the data, this method is blind

The selection of a good training sample is a trial-and-error project, and the trial-and-error project is an iterative process that involves repeated four steps of sample selection, classification, evaluation, and update of the sample set until a satisfactory result is achieved. a very time consuming process

Images