Depth feature and graph cut method-based built-up area automatic extraction method

Abstract

The invention discloses a depth feature and graph cut method-based built-up area automatic extraction method, and belongs to the technical field of image data processing. The method comprises the steps of using a high-resolution full-color image and a multispectral image subjected to ortho-rectification and linear cutting stretching processing as data sources; dividing the full-color image into image blocks same in size; extracting depth features of the image blocks by using a deep convolutional neural network; by taking the image blocks as nodes and taking the depth features as node features,building a graph model; determining built-up areas by using a graph cut method; performing voting by taking the image blocks as primitives based on multiple spectral indexes of the multispectral image to remove a false alarm; eliminating the built-up areas with the excessively small areas and non-built-up areas; performing super-pixel segmentation on the image blocks of built-up area edges; performing the voting based on the multiple spectral indexes of the multispectral image to remove the false alarm, thereby obtaining refined built-up area edges; and finally extracting an edge vector graph. The method can quickly, effectively and accurately realize the extraction of the built-up areas.

Description

technical field [0001] The invention belongs to the technical field of image data processing, and more specifically relates to a method for automatically extracting built-up areas based on depth features and a graph cut method. Background technique [0002] The built-up area refers to the area in the urban administrative area that has actually been developed and constructed in a large area, and the municipal public facilities and public facilities are basically equipped. For the visual discrimination task of remote sensing images, the built-up area refers to the area where buildings are densely distributed and reaches a certain area. There is no unified standard for defining the area of ​​the built-up area. [0003] The extraction of built-up areas is an important task in the interpretation of remote sensing images, and the results of built-up area extraction have important uses in many fields. For example, for urban development planning, accurate statistics of the area of ...

AI Technical Summary

Problems solved by technology

[0006] A large number of features and classification methods have been applied in existing studies on the extraction of built-up areas from remote sensing images. These studies usually have the following three problems: First, the research uses a small amount of data, usually a small number of images in a certain area, covering The sample space of the model is narrow, the generalization ability of the model is poor, and it lacks practicability; secondly, the image is divided into built-up areas and non-built-up areas with pixels as primitives, which requires a huge amount of calculation, and for built-up areas without clear closed boundaries The effect of the region is poor; finally, the feature representation ability for built-up area extraction is insufficient, which is also the most important factor affecting the detection accuracy and scalability of the algorithm

[0007] Although the existing built-up area extraction algorithms have solved some problems to a certain extent, it is difficult to put them into practical use in remote sensing images of vast areas.

Images