Study On The Extraction Of Buildings Based On The Airborne LiDAR Data And Image Processing Technology

Airborne laser radar system is a new generation of remote sensing system which includes laser ranging device,Dynamic differential GPS,Inertial Navigation System.It can gain the high-accuracy terrain data,and because the working period of it is short,the high accuracy data and the good environmental adaptability,the laser system has been used in basic surveying,forest monitoring,the mapping of the power and the railway system and the city planing.However,the backward processing technology limits the development of the airborne LiDAR in the 3D city modeling and the extraction of features.With the development of the digital city,it has been the trend to study using the high accuracy LiDAR data to extract the buildings.And the study of the multi source data in LiDAR filter is the key point and the difficulty.Based on the LiDAR data and analysing the feature of the LiDAR data and the noise of the LiDAR image,combining the technology of the image processing,this paper mainly researches the application of the morphology filtering algorithm in extracting buildings with the Li DAR image.This paper has finished the following researches according to the main tech in the processing of the LiDAR image.1.This article advances a new annular-complementary morphological filtering algorithm which is on the basis of the feature of the noises in the LiDAR image and the advantage of morphological algorithm.The new algorithm is consisted of two different structural elements,and the annular one is used to detect the noise with morphological open operation,and the other one is used to fill the hiatus with morphological close operation.2.The paper advances the algorithm that based on the statistical area theory.The annular-complementary morphological filtering algorithm can filter the small noise except for the noise spots on the buildings.So this paper has studied the feature of the noise spots,advanced the new algorithm.The new algorithm computes the distribution of the noise that locate in the scope of the template,then compare the results to the threshold to filter the noises.3.Single template area element algorithm can effectively filter out the noise spots,due to the simplex discriminant function,we have to add iterations to get the result.However,with the increasing of the iterations,the edge of the buildings become serious.According to the above problems,this paper improves the Single template area element algorithm,proposing a new double templates area element algorithm.The outer template is used to detect the size of the noise,and the inner one to compute the pixel information.The algorithm filters out the noise and fills up the holes of the image that is basis of the area theory.At last,it uses the multi-size morphological edge detection algorithm to extract the edge of the buildings.According to the results of the experiment,the annular-complementary morphological filtering algorithm can finish the work effectively,especially the spice-salt noises.But it is useless to the bigger noise spots.The single template and double templates algorithms can filter out the noise which is caused by the discreteness and blindness of the LiDAR data.In the aspect of protecting the edge information,area element algorithm also works effectively.