Research On Road Extraction Method From Remote Sensing Image Based On Adaptive Morphology

Extracting roads from remote sensing images is a very important and challenging task,which is of great significance for map composition,traffic management and urban planning.Traditional road extraction is usually judged by manual visual method,which not only wastes a lot of manpower,material resources and financial resources,but also has low detection efficiency and accuracy.With the rapid development of high-resolution satellites,a large amount of usable data has been generated.Therefore,how to use these massive data to quickly and completely extract road information and effectively use it has become a hot and difficult problem in this field.The content of remote sensing image is generally complex,which may lead to false extraction and missing extraction when extracting roads from remote sensing image.Using mathematical morphology can detect the spatial structure features of remote sensing image and better maintain the geometric structure of ground objects.Therefore,this paper proposes a method of extracting roads in remote sensing images based on adaptive mathematical morphology algorithms.The main work of the thesis is as follows:(1)A brief introduction to the research status and development significance of road extraction methods from remote sensing images and adaptive mathematical morphology at home and abroad is given,the basic theories and methods of mathematical morphology are analyzed,and a large number of simulation experiments are carried out.(2)When using classical mathematical morphology to extract road objects from remote sensing images,due to the use of structuring elements with fixed shape and size,some details of road morphology and spatial features are lost,so it is difficult to obtain high segmentation accuracy.Therefore,a road extraction method based on adaptive morphological reconstruction for high-resolution remote sensing images is proposed.Firstly,multi-scale structuring elements are employed to reconstruct the image,and perform point-by-point maximum value operation on the reconstructed image to obtain the adaptive reconstruction result.Then,the adaptive reconstruction opening and closing operations are combined to obtain a level operator.The level operator is used to measure the morphological features of the image,so as to preliminary separate the road area and non road area.The accurate road area is obtained by shape filtering and holes filling,and the complete the center lines of the road network are extracted by fast parallel refinement algorithm.The experimental results show that this method can make full use of the road shape and spatial features of remote sensing images and accurately extract the smooth road centerline.(3)The content of remote sensing images is complex and diverse,and noises such as vehicles,pedestrians,and trees on both sides of the road will affect the accuracy of road extraction.When using the classical morphological method for denoising,the larger structuring elements can filter out most of the noise,but it is easy to cause the edge of the image to become blurred and lose part of the edge information.The smaller structuring elements can retain most of the information of the image,but can not completely filter out the noise in the image,which affects the subsequent road extraction results.Therefore,a method of morphological remote sensing image road extraction based on MST-SASE(Salience Adaptive Structuring Eelment Base on Minimum Spanning Tree)is proposed.In this method,the gradient image of the input image is first calculated and the edge image is obtained by non-maximum suppression of the gradient image,and performs chamfer distance transformation on the edge image to obtain a salience map(SM,Salience Map).Then,the radius of structuring elements is determined by calculating the maximum and minimum values of SM.Next,the minimum spanning tree is calculated on the SM,the calculated radius is used to construct a structure element whose shape and size adaptively change with the local features of the input image.At the same time,the basic morphological operatoin elements,and compared with the classical morphological operators,so as to filter the noise in remote sensing image.Finally,the gray window slice is used to segment the road area,and the road center line is extracted accurately through a series of operations such as shape filtering.Experiments show that this method has high completeness,correctness and extraction quality.