深度卷积神经网络(deep convolutional neural network,DCNN)在高分辨率遥感图像自动道路提取领域被广泛应用,但现有方法难以对预测结果中像素间的上下文关系建模.针对此问题,已有研究利用全连接条件随机场(fully con-nected conditional random field,FullCRF)结合上下文信息对语义分割结果进行二次优化,但无法有效改善道路结构不连续问题.为改善道路结构的完整性,提出一种结合DCNN的短距条件随机场模型(short range conditional random field,SRCRF),SRCRF利用DCNN强大的特征提取能力并控制FullCRF的推理范围缓解过度平滑现象,解决道路提取结果中的结构不连续、不完整问题.实验结果表明,在Zimbawe-Roads数据集与Cheng-Roads数据集中,SRCRF的F1分数相比DCNN分别上升约4.01％、3.73％,相比FullCRF分别上升约3.25％、2.28％.
Road Extraction from Remote Sensing Image by Integrating DCNN with Short Range Conditional Random Field
Objective:Deep convolutional neural network(DCNN)is widely used in automatic road extrac-tion from high-resolution remote sensing images(HRSIs).However,the existing methods make it difficult to model the context relationship of pixels in the predicted results.To solve this problem,some studies have used fully connected conditional random field(FullCRF)to perform secondary optimization of seman-tic segmentation results combined with context information,but the discontinuity problem of road structure cannot be effectively improved.Methods:In order to improve the integrity of road structure,this paper pro-poses a short range conditional random filed(SRCRF)model combined with DCNN.SRCRF mainly in-cludes a unary potential function based on road pre-segmentation,a binary potential function based on spec-tral spatial features,and a K-neighborhood mean field inference algorithm.First,the prior knowledge of road pre-segmentation results is obtained by using the powerful feature extraction capability of DCNN as the unary potential function of SRCRF.Then,the dependence of the binary potential function defined by the linear combination of Gaussian kernel functions on the surrounding nodes is modeled.The binary poten-tial function enables the classification results to have local consistency,that is,adjacent pixels with similar spectral features have the same label.Finally,the K-neighborhood mean field inference algorithm based on the mean-field approximation inference algorithm optimizes the inference range to make full use of the spa-tial context information and spectral feature context information of the road,and then calculates the optimal label corresponding to each pixel based on the space and spectral feature to optimize the road accurately.The convolution method is adopted to control the inference range of SRCRF within the radius of K in order to improve the proportion of feature vectors.Results:The experimental results show that SRCRF not only alleviates the transition smoothness of FullCRF,but also alleviates the structural discontinuity and incom-pleteness in the road acquisition results of HRSIs.In Zimbawe-Roads dataset and Cheng-Roads dataset,F1-scores of SRCRF are increased by about 4.01％and 3.73％respectively compared with DCNN,and about 3.25％and 2.28％respectively compared with FullCRF.Conclusions:The proposed SRCRF com-bines the advantages of DCNN,optimizes the fully connected structure of traditional conditional random fields into the K-neighborhood structure,reduces the inference scope,and improves the proportion of fea-ture vectors.Compared with FullCRF,SRCRF can make better use of image color features and spatial fea-tures to accurately optimize the road extraction results of deep learning output.The performance of SRCRF is improved compared with DCNN and FullCRF,and the time is shortened by one order of magnitude com-pared with FullCRF.In the future,we will further investigate the potential for learning Gaussian features and investigate more complex architecture of conditional random field to better capture global context infor-mation.Additionally,we are particularly interested in exploring the application potential of SRCRF in other fields,such as building extraction,vehicle extraction,lake extraction,etc.
remote sensing imageroad extractiondeep learningconditional random fieldsemantic seg-mentation
NETL
NSTL
万方数据
