Remote sensing building detection method

Abstract

The invention discloses a remote sensing building detection method, which comprises the following steps of: in a preprocessing stage, firstly extracting a candidate building object set through image segmentation and a group of discrimination rules; secondly, screening differential profiles of the MAPs through a genetic algorithm, and providing a self-adaptive ACGA-DAPs to extract potential building pixels; on the basis, providing an unsupervised decision fusion framework by constructing a novel building index SSBI (Statistics-Space Building Index), and finally achieving automatic detection of the building. The method provided by the invention shows excellent performance in multiple groups of high-resolution remote sensing image experiments of different sensors in different regions, and the overall precision can reach 91.9% or above.

Description

technical field [0001] The invention relates to a remote sensing building detection method, which belongs to the technical field of high resolution remote sensing image data processing. Background technique [0002] With the rapid development of remote sensing earth observation technology, building detection based on high-resolution remote sensing images has become one of the research hotspots in the field of remote sensing. High-precision and highly reliable building detection results play a key role in many fields of human activities (such as land use dynamic monitoring, urban planning, and population density estimation, etc.). Compared with traditional mid- and low-resolution remote sensing images, high-resolution remote sensing images provide richer spatial structure information, which is conducive to the fine depiction of buildings in complex urban backgrounds. On the other hand, the low signal-to-noise ratio of high-resolution remote sensing images limits the accuracy...

AI Technical Summary

Problems solved by technology

However, in practical applications, the realization of high-precision, unsupervised high-resolution remote sensing image building detection based on MAPs must break through the following limitations: (1) Potential building pixels are differentiated by adjacent scale APs (Attribute Profiles) The extracted DAPs (Differential Attribute Profiles) are directly determined, and the MAPs theory does not give clear rules for setting scale parameters, so it is necessary to adaptively construct a reasonable set of scale parameters based on the original image

If the interval between scales is too large, it will be difficult to completely describe different types of buildings with different attributes; on the contrary, it is easy to retain too many other ground object pixels with similar attributes to buildings in the detection results

(2) As the basis for judging whether a certain pixel belongs to a building, DAPs extracted from different attributes may give opposite conclusions, and it is difficult to obtain reliable detection by directly taking the union of DAPs of all attributes and scales As a result, efficient decision rules need to be designed to deal with this conflict of evidence

(3) Buildings should be a type of geographical object with closed contours. How to convert the potential building pixels extracted based on MAPs into object-level detection results is another challenge that must be faced.

Images