A joint superpixel graph cut optimization method for complex scene building change detection

Abstract

The invention discloses a combined super-pixel graph cut optimization method for complex scene building change detection, which comprises the following steps: step 1, acquiring digital surface modelsat two different periods, performing gridding processing, and performing combined segmentation on the digital surface models after fusion to obtain a combined super-pixel object; step 2, for the original image data of the two periods, realizing image classification by adopting a semantic segmentation algorithm based on a deep convolutional neural network, and detecting a building in the image; step 3, extracting a plurality of characteristics of the combined superpixel object by taking the combined superpixel object as a processing unit, constructing a graph cut optimized data item and a smooth item, and obtaining a global optimal solution by adopting a maximum flow minimum cut theory to obtain a change object of the building in two periods; and step 4, classifying the change types of thebuilding, including building, heightening, dismantling and lowering. According to the method, the building change detection precision and reliability can be remarkably improved.

Description

technical field [0001] The invention relates to the field of three-dimensional building change detection, in particular to a joint superpixel graph cut optimization method for complex scene building change detection. Background technique [0002] Building change detection, as one of the important contents of geographic national conditions monitoring, is of great significance to the identification of illegal buildings, urban dynamic monitoring and geographic information update. In addition to the lack of relevant legal links, the lack of automated monitoring methods for illegal buildings is also an important reason. There is an urgent need in the market for a highly automated, robust and reliable building change detection method to assist in promoting the remediation process of urban illegal buildings. [0003] In the early days, some scholars proposed a building change detection method based on high-resolution remote sensing images of different periods. However, due to the ...

AI Technical Summary

Problems solved by technology

However, the change detection results of such methods usually depend on the classification accuracy, and the single-period classification errors will be accumulated into the final change detection results.

Images