High-resolution image oriented ground object feature screening method

Abstract

The invention discloses a high-resolution image oriented ground object feature screening method which comprises the steps of performing feature analysis and feature coarse screening on a target groundobject according to ground object actual distribution and a remote sensing image expression mode; according to different selected image characteristics, performing one-time feature extraction on thetarget ground object in the high-resolution original image, and acquiring a plurality of extraction results of the target ground object; performing sensitivity analysis on each extraction result by means of a constructed feature sensitivity model, and obtaining the effective extraction degrees of different image features to the target ground object; sequencing the different image feature sensitivities, and selecting a plurality of features with relatively high sensitivity to the target ground object for forming a sensitivity feature set of the target ground object, etc. Through consideration in multiple aspects, the high-resolution image oriented ground object feature screening method has remarkable effects of analyzing the sensitivity degree of the feature to the target ground object, improving objective authenticity of an extracting result, effectively settling a problem of high difficulty in ground object feature selection, and reducing interference of man factors to ground object information extraction.

Description

Technical field [0001] The invention relates to the technical field of remote sensing image processing, in particular to a method for screening feature features of high-resolution images. Background technique [0002] High-resolution remote sensing images contain rich detailed information of ground features, and obtain complex and diverse spectral information, as well as rich structural information such as space, texture, topology, and semantics. The phenomenon of high dimensionality and large number of feature data inevitably occurs, namely Feature "dimension disaster", how to optimize the selection of massive features is the key to the effective extraction of feature information. [0003] The existing feature selection methods are mainly studied from two aspects: search strategy and evaluation criteria. The former commonly used algorithms include exhaustive search, sequential search, genetic algorithm, ant colony algorithm, particle swarm algorithm, etc. The algorithm is mainly ...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies of the prior art, the object of the present invention is to provide a feature screening method for high-resolution images, which constructs feature features from two elements: the accuracy of feature feature information extraction and the complexity of mis-extraction. Sensitivity analysis model, which quantifies the effective extraction degree of different image features for target features, and solves the problem of difficult feature selection

Images