Road Boundary Extraction In Vehicle-borne Laser Point Clouds Based On Diverse LBP Features

With the rapid development of smart city and intelligent transportation.accurate and detailed road spatial information is becoming more and more important.And what's more,road boundary information is of great significance for high-precision map production.road 3D model reconstruction.traffic network management and so on.As an advanced surveying and mapping system,vehicle-borne mobile mapping system has the advantages of high precision.high efficiency.active.non-contact measurement and so on.And it has become one of the important means of road information acquisition and update.There are many problems in road boundary.such as various types,complex scenes,laser scanning occlusion and so on.And the problems result in great difficulty for extracting the road boundary automatically from vehicle-borne laser point clouds.Therefore.based on the in-depth analysis of point clouds spatial and geometric characteristics at road boundary.the paper introduces the principle of LBP operator and proposes an automatic extraction method for road boundary based on diverse LBP features.The main contents of this paper are as follows.(1)It combs the development status,development trend the vehicle-borne mobile mapping system and characteristics of vehicle-borne laser scanning point clouds in detail.Then it summarizes the existing road boundary extraction methods.These contents lay a theoretical foundation for the proposed method in this paper.(2)It analyzes the characteristics of the vehicle-mounted laser point clouds at road boundary position in detail and deeply studies the principle and calculation method of improved LBP operator.Aiming at the diversity of road boundary's spatial features,three improved operators,elevation LBP,dispersion LBP and spatial shape LBP.are designed by using the one-dimensional attributes,two-dimensional distribution and three-dimensional shape features of boundary point clouds.And it performs statistical analysis on the feature image pixel values of the three LBP improvement operators.The statistical results are consistent with the spatial distribution and geometric characteristics of different objects such as road boundary and road surface.(3)In view of two different types of road boundary,curb and greenbelt boundary,a road boundary automatic extraction method based on diverse LBP features is proposed.The method utilizes three improved operators of elevation LBP,dispersion LBP and spatial shape LBP,and focuses on constructing a diverse LBP feature semantic recognition model for road boundary,which solves the problem of modeling and classification of road boundary point clouds features.Finally,it performs road-direction denoising and linear vectorization on the road boundary points to obtain road boundary vector data.(4)Experiments on many different types of urban road and rural road point clouds,and compares the result integrity rate and accuracy rate with other methods.The results show that the method can accurately extract the curb boundary in the urban road environment with high integrity rate and accuracy.which has strong applicability and robustness.For greenbelt boundary.the accuracy of the method is high,and the integrity rate needs to be improved.