Research And Improvement Of Road Extraction Algorithm In High Resolution Remote Sensing Images

As more and more high-resolution remote sensing satellites are launched,we can obtain more and more high-resolution remote sensing images.It is a major issue that how we extract the useful information from these massive data accurately and efficiently.In the past few decades,road extraction has got extensive attention and research.Roads not only have a wide range of application in areas such as urban planning and construction,tourism navigation,traffic management,change detection,and natural disaster analysis,but also as important components of geographic information systems,they are helpful for national security and campaign operation plan.In recent years,the research on road extraction from high-resolution remote sensing images has made great progress,while its performance fails to meet the application requirements.The main difficulties of road extraction from highresolution remote sensing images lie in the following aspects:(1)phenomena of “the same spectrum in different objects”,with the continuous improvement of resolution in remote sensing image,some parks and buildings have the similar spectral values.(2)phenomena of “the same object with different spectra”,roads in different area may be construct by different materials.(3)Some road areas are under occlusions of trees and vehicles as well as shadows of tall buildings,which makes the existing road extraction methods show little robustness against the occlusions.This dissertation is used to overcome these challenges,the main contributions can be summarized as follows:1.To overcome the deficiencies of directional segmentation,we propose an improved road network extraction from remote sensing images based on the shear transform,the directional segmentation,the road probability,shape features and a skeletonization algorithm.The proposed method includes the following steps.First,we combine shear transform with directional segmentation to get the initial road regions.Second,road map based on Mahalanobis distance and thresholding is fused with the initial road regions to improve accuracy.Third,road shape features filtering and hole filling are used to extract reliable road segments.Finally,the road centerlines are extracted by an automatic subvoxel precise skeletonization method based on fast marching.Road network is then generated by post-processing.Experimental results show that the proposed method can extract smooth and correct road centerlines.2.We propose an improved method for road centerlines extraction,which is based on shear transform,directional segmentation,shape features filtering,directional morphological filtering,tensor voting,multivariate adaptive regression splines(MARS)and enhanced broken lines connection.The proposed method consists of four steps.Firstly,directional segmentation based on spectral information and shear transform is used to segment the images for obtaining the initial road map.Shear transform is introduced to overcome the disadvantage of the loss of the road segment information.Secondly,we perform hole filling to remove the holes due to noise in some road regions.Reliable road segments are extracted by road shape features and directional morphological filtering.Directional morphological filtering can separate road from the neighboring non-road objects to ensure the independence of each road target candidate.Thirdly,tensor voting and MARS are exploited to extract smooth road centerlines,which overcome the shortcoming that the road centerlines extracted by the thinning algorithm have many spurs.Finally,we propose an enhanced broken lines connection algorithm to generate a complete road network,in which a new measure function is constructed and spectral similarity is introduced.We evaluate the performance on the highresolution aerial and Quick Bird satellite images.The results demonstrate that the proposed method is promising.3.In recent years,level set evolution has been used to extract the road,but it is difficult to automatically generate initial level curves for the level set evolution(LSE).In this paper,we propose an automatic approach for the generation of initial level curves and use it to extract the road.Firstly,CNN is used to classify the road or nonroad,then shape features are adopted to filter nonlinear features to get accurate road region.And then on this basis,we exploit tensor voting to detect the road junctions and utilize them as initial level curves;finally we fuse the results obtained by CNN and LSE.Experiments show that this algorithm can get accurate and complete road.However,the initial road map obtained by CNN can not align with real road boundaries.To address the above issue to some extent,we present a multiscale road centerlines method based on CNN and edge-preserving filtering.The proposed algorithm consists of the following four steps.Firstly,the aerial imagery is classified by a pixel-wise classifier based on convolutional neural network(CNN).Specifically,CNN is used to learn features from raw data automatically,especially the structural features.Then,edge-preserving filtering is conducted on the resulting classification map,with the original imagery serving as the guidance image.It is exploited to preserve the edges and the details of the road.After that,we do some post-processing based on shape features to extract more reliable roads.Finally,multiscale Gabor filters and multiple directional non-maximum suppression are integrated to get a complete and accurate road network.Experimental results show that the proposed method can achieve comparable or higher quantitative results,as well as more satisfactory visual performance.4.To solve the problem of limited labeled samples in road extraction task,a semi-supervised road centerline extraction is proposed,which incorporates high-level feature selection,markov random field and ridge transversal method.The proposed road extraction approach consists of three steps: multiple features extraction,semi-supervised road area extraction,and road centerlines extraction.In order to get more abstract and discriminative high-level features,instead of concatenation,multiple-feature adaptive sparse representation(MFASR)is applied to mid-level features generated by different prototype sets.To obtain an accurate road area result,we combine feature learning framework with markov random field(MRF).After getting accurate road area,we integrate Gabor filters and non-maxima suppression with ridge transversal method to extract centerlines.It is verified to achieve better performance than the state-of-the-art methods in terms of visual and quantitative aspects.