Design And Implementation Of Building Extraction System For Remote Sensing Image Based On Unmanned Aerial Vehicle

Building is the main place for human production and life,how to effectively manage and supervise the construction land is very important.With the increasing development of science and technology,UAV remote sensing images have been normalized for daily land management.How to quickly,accurately and automatically extract construction land,the implementation of intelligent,automated management is an important content of this study.The influence of the three main features of construction land will be detected and extracted by the AdaBoost algorithm,color,texture and shape are extracted and concatenated together.The algorithm will be integrated in the system,the final design and implementation of UAV remote sensing image extraction system based on construction land.The main work of this paper includes the following four points:(1)INPHO software is used to preprocess the original UAV remote sensing images,including: distortion correction,aerial triangulation,generating orthophoto,mosaic color uniformity.In this paper,the color feature,texture feature and shape feature of building land recognition are studied,and the recognition accuracy of building land is extracted.(2)After analyzing the characteristics of various types of buildings,according to the characteristics of the building land shape,texture uniformity,color uniformity and other characteristics,the AdaBoost algorithm is used to cascade the three features.Experiments show that the AdaBoost algorithm is more accurate than the single feature extraction algorithm,which is more conducive to accurately identify the building land.The method is invariant to direction,position and scale,and can be applied to other ground classes extraction.(3)The basic framework of the system is designed,and the core function of the system is to extract the building land of the UAV,and other functions are assisted from the side.The system must have high efficiency,stability,high degree of automation,simple operation,good user experience and other basic characteristics.(4)Through the use of ArcGIS Engine GIS platform,based on.NET Framework visual development environment,using C# programming language,the realization of building land extraction system based on unmanned aerial vehicle remote sensing image.The system has the functions of image preprocessing,color transformation,multi feature extraction and building extraction.It realizes the integration and automation of the recognition and extraction of building land.From the experimental results,the average accuracy rate is 91% and the average detection rate is 92% based on the three feature cascade AdaBoost algorithm.The accuracy and detection rate of the algorithm are much higher than those of color feature,texture feature and shape feature.In this paper,the research results of the algorithm are built into a building extraction system,which makes science and technology into productive forces,saving much effort of artificial extraction of land objects.It also provides a reference for remote sensing images to achieve more accurate and automatic extraction of all objects.