Large-breadth remote sensing image building extraction method used for hybrid scene

Abstract

The invention relates to a large-breadth remote sensing image building extraction method used for a hybrid scene. The large-breadth remote sensing image building extraction method comprises the following steps: 1: preprocessing a remote sensing image; 2: carrying out superpixel segmentation based on an entropy rate; 3: extracting three characteristics including shadows, homogeneity and strong edges; 4: generating a shade layer, a homogeneity layer and a strong edge layer; 5: extracting buildings independently on different layers; and 6: carrying out building superposition. The large-breadth remote sensing image building extraction method has the beneficial effects that more characteristics and semantic information of deeper levels are extracted, and the large-breadth remote sensing image building extraction method can adapt to building extraction in a larger geographic area, and is used for realizing relevant businesses including basic geographic database update and the like.

Description

technical field [0001] The invention relates to the field of remote sensing image processing, in particular to a large-format remote sensing image building extraction method for mixed scenes. Background technique [0002] Buildings are one of the main geographical elements of a city and an important content of various urban thematic maps. Researching the extraction of buildings is of great significance for comprehensively examining the urban geographic information environment. With the rapid development of high-resolution remote sensing image acquisition technology, the processing, analysis and application of remote sensing images have better data sources, and its digital products have wider and more in-depth applications. Various degrees of progress have been made in computer image processing technology, pattern recognition, artificial intelligence, etc., which provide the possibility to efficiently extract effective information from massive images. However, the extraction...

AI Technical Summary

Problems solved by technology

[0003] (1) The data sources are mainly two-dimensional remote sensing images, and most of the cases lack direct three-dimensional data;

[0004] (2) Different remote sensing images often have large differences due to factors such as spectral range, resolution, geometric image of the sensor, and imaging conditions;

[0005] (3) The appearance and texture details of different types of buildings are ever-changing, showing great differences in remote sensing images, and it is difficult to establish a unified building model library, which makes the automatic extraction of information quite difficult;

[0006] (4) The complexity of the scene where the building is located, such as low contrast, mutual occlusion of houses, shadows of the building itself, and shadows of other ground objects, etc., so it is necessary to automatically extract buildings with clear boundaries from the background more difficult

[0007] The extraction of buildings from large-format remote sensing images containing mixed scenes is a problem that must be solved in practical engineering applications. "Mixed scenes" refer to buildings that usually contain a variety of different features in large-format remote sensing images to be processed. And the distribution is irregular, it is difficult to use the regular partition method to divide the buildings with the same characteristics, resulting in the inability to select the corresponding building extraction method, or using a single building extraction method for extraction is prone to serious missed detection and false detection

Images