Building base contour refined extraction method based on live-action three-dimensional model

Abstract

The invention relates to a building base contour refined extraction method based on a live-action three-dimensional model, and the method carries out the building base contour extraction based on the live-action three-dimensional model, and comprises the steps: firstly, extracting a vegetation pattern spot through a deep learning algorithm to refine the building pattern spot, effectively distinguishing the vegetation and the building which are close in height, and generating an initial building vector surface; then, generating a side image map based on the initial building vector surface and the live-action three-dimensional model, generating a first side image map through the side image map, and classifying the first side image map by using a deep learning algorithm, so that attachments on the side of the building can be effectively eliminated, and interference is avoided; and then, a window is extracted from the second side image to obtain floor information, so that protruding structures such as a building eave and illegal construction can be effectively removed, automatic correction of the eave is realized, and an accurate building base contour can be extracted.

Description

technical field [0001] The invention relates to the technical field of contour extraction of building bases, in particular to a method for finely extracting the contours of building bases based on a real-scene three-dimensional model. Background technique [0002] In the construction of smart cities, the building base profile plays an extremely important role, and it can be used in urban planning, disaster assessment, homestead survey and other fields. The existing building base contour extraction methods are mainly divided into the following two categories: [0003] (1) The method based on on-site surveying and mapping: operators use high-precision surveying and mapping instruments, such as total stations, to measure the outline of the building base; this method has high precision, but the workload is heavy, and it is not suitable for large-scale surveys that require quick and corresponding surveys. Task. [0004] (2) The (semi) automatic extraction method based on remote...

AI Technical Summary

Problems solved by technology

[0003] (1) The method based on on-site surveying and mapping: operators use high-precision surveying and mapping instruments, such as total stations, to measure the outline of the building base; this method has high precision, but the workload is heavy, and it is not suitable for large-scale surveys that require quick and corresponding surveys. Task

[0004] (2) The (semi) automatic extraction method based on remote sensing images: this method is limited by the defects of remote sensing images, such as scene occlusion, poor building projection, etc., and the extracted results are not the real outline of the building base, which affects the subsequent Quantitative application

For example, the Chinese patent with the publication number CN113011288A discloses a remote sensing building detection method based on the Mask R-CNN algorithm. This method obtains a detection network model suitable for irregular remote sensing buildings by modifying the RPN network and improving the direction of the mask mask. , which strengthens the ability to detect and segment irregular buildings, but cannot solve the problems of occlusion and projection difference between buildings

[0005] Another Chinese patent with the publication number CN113011288A discloses a method for automatically extracting buildings from tilted photographic data that combines height and spectral information. Through steps such as height threshold segmentation, vegetation information filtering, and post-processing, the results of building extraction are optimized to realize building information. Automatic extraction, but after filtering out vegetation information, it directly removes small-area noise objects and fills small internal holes. In the middle of filtering, there will be a mixture of bottom-level noise and high-level information. At the same time, it does not solve the problem between buildings Problems such as occlusion and projection difference between

Images