摘要
Abstract
Real estate entity surveying involves objects with complex shapes and ambiguous boundaries,making it difficult for a single unmanned aerial vehicle(UAV)to capture complete imagery.Additionally,feature point pair matching during image processing can be disrupted by actions such as image rotation,leading to stitching failure and affecting surveying accu-racy.To address this,this paper proposed a real estate entity surveying method based on intelligent processing of multiple UAV remote sensing images.High-resolution images of real estate entities were obtained through multi-UAV collaborative operations.By combining the improved fast corner detection(FAST)with rotation-invariant binary robust independent elementary features(BRIEF)descriptors,and using the efficient image feature algorithm ORB,corners can be quickly detected in images and used as key points.The generated descriptors had rotational invariance,ensuring correct matching of feature points.Affine transformation was then applied to complete the image stitching,ensuring full coverage and accurate stitching.A dense matching algorithm was used to generate depth maps,and Poisson surface reconstruction was employed to build three-dimensional(3D)models of the real estate entity.Geometric measurements were then performed on the 3D model,calculating the geometric parameters of points,lines,surfaces,and real estate data such as land area and elevation,achieving precise real estate surveying.Test results show that when using this method for real estate entity surveying,the interquartile range(IQR)of elevation data is lower,achieving ideal surveying accuracy.关键词
无人机遥感/不动产实体/测绘技术/影像处理/测绘精度Key words
unmanned aerial vehicle(UAV)remote sensing/real estate entity/surveying technology/image processing/surveying accuracy分类
天文与地球科学
