首页 > 论文 > 激光与光电子学进展 > 57卷 > 20期(pp:201019--1)

结合卷积神经网络和张量投票的道路提取方法

Road Extraction Method Combining Convolutional Neural Network and Tensor Voting

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

深度学习通过训练样本进行特征识别,已经被广泛应用于道路提取领域。该方法不局限于特定类型的影像,但是受训练样本数量和计算机硬件的限制,所提取的道路会有断裂和噪声。针对上述问题,使用VGG卷积神经网络对道路进行初步提取后引入张量投票方法进行优化处理。首先通过影像变换、随机裁剪、过采样等方法对样本进行多模式扩充,进而训练VGG卷积神经网络模型;其次利用该网络从原始影像中初步分割道路面,接着对道路面的二值影像进行张量投票获取道路的显著性信息;最后在特征提取时针对显著性信息加入阈值获取道路面。实验结果表明,所提方法提取道路的召回率与正确率均达90%以上,与其他传统方法相比具有更高的精度,验证了所提方法的有效性。

Abstract

Deep learning is widely used in road extraction by training samples for feature recognition. This method is not limited to a specific type of image; however, the extracted road will be broken and noisy owing to the restriction of the number of training samples and computer hardware. Owing to the above problems, in this study, we use a VGG convolutional neural network to preliminarily extract roads and introduce a tensor voting method for optimization. First, multi-mode expansion of the samples is performed via image transformation, random cropping, and oversampling, and then, a VGG convolutional neural network model is trained. Second, the network is used to segment the road from the original image. Then, the tensor voting for the binary images of the road surface is used to obtain saliency information about the road. Finally, the threshold of significant information is set to obtain the road surface in the feature extraction process. The experimental results show that the recall rate and precision of the extracted road obtained by the proposed method are more than 90%, and the proposed method has a higher accuracy than other traditional methods, which verifies the effectiveness of the proposed method.

广告组1 - 空间光调制器+DMD
补充资料

中图分类号:P237

DOI:10.3788/LOP57.201019

所属栏目:图像处理

收稿日期:2019-12-10

修改稿日期:2020-03-10

网络出版日期:2020-10-01

作者单位    点击查看

李天琪:辽宁工程技术大学测绘与地理科学学院, 辽宁 阜新 123000自然资源部国土卫星遥感应用中心, 北京 100048
谭海:自然资源部国土卫星遥感应用中心, 北京 100048
戴激光:辽宁工程技术大学测绘与地理科学学院, 辽宁 阜新 123000
杜阳:辽宁工程技术大学测绘与地理科学学院, 辽宁 阜新 123000
王杨:辽宁工程技术大学测绘与地理科学学院, 辽宁 阜新 123000

联系人作者:谭海(tanhai001@139.com)

【1】Mena J B. State of the art on automatic road extraction for GIS update: a novel classification [J]. Pattern Recognition Letters. 2003, 24(16): 3037-3058.Mena J B. State of the art on automatic road extraction for GIS update: a novel classification [J]. Pattern Recognition Letters. 2003, 24(16): 3037-3058.

【2】Bonnefon R, Dhérété P, Desachy J. Geographic information system updating using remote sensing images [J]. Pattern Recognition Letters. 2002, 23(9): 1073-1083.

【3】Shi W Z, Zhu C Q, Wang Y. Road feature extraction from remotely sensed image review and prospects [J]. Acta Geodaetica et Cartographic Sinica. 2001, 30(3): 257-262.
史文中, 朱长青, 王昱. 从遥感影像提取道路特征的方法综述与展望 [J]. 测绘学报. 2001, 30(3): 257-262.

【4】Chen J S, Zhang A J. Review of lane detection methods based on road features China Science & Technology Panorama Magazine[J]. 0, 2017(8): 247-249, 251.
陈静思, 张爱军. 基于道路特征的车道线检测方法综述 中国科技纵横[J]. 0, 2017(8): 247-249, 251.

【5】Dai J G, Du Y, Fang X X, et al. Road extraction method for high resolution optical remote sensing images with multiple feature constraints [J]. Journal of Remote Sensing. 2018, 22(5): 777-791.
戴激光, 杜阳, 方鑫鑫, 等. 多特征约束的高分辨率光学遥感影像道路提取 [J]. 遥感学报. 2018, 22(5): 777-791.

【6】Lei X Q, Wang W X, Lai J. A method of road extraction from high-resolution remote sensing images based on shape features [J]. Acta Geodaetica et Cartographica Sinica. 2009, 38(5): 457-465.
雷小奇, 王卫星, 赖均. 一种基于形状特征进行高分辨率遥感影像道路提取方法 [J]. 测绘学报. 2009, 38(5): 457-465.

【7】Zhang Y H, Xia G H, Kan X, et al. Road extraction from multi-source high resolution remote sensing image based on fully convolutional network [J]. Journal of Computer Applications. 2018, 38(7): 2070-2075.
张永宏, 夏广浩, 阚希, 等. 基于全卷积神经网络的多源高分辨率遥感道路提取 [J]. 计算机应用. 2018, 38(7): 2070-2075.

【8】Liu J F. A concise and efficient method for accelerating convolution neural networks [J]. Science Technology and Engineering. 2014, 14(33): 240-244.
刘进锋. 一种简洁高效的加速卷积神经网络的方法 [J]. 科学技术与工程. 2014, 14(33): 240-244.

【9】Wei Q, Ai L M, Ye X N. An automatic road extraction method for high-resolution remote sensing images [J]. Computer Technology and Development. 2019, 29(6): 130-133.
魏清, 艾玲梅, 叶雪娜. 一种高分辨率遥感图像道路自动提取方法 [J]. 计算机技术与发展. 2019, 29(6): 130-133.

【10】Su P K. Automated extraction of road networks in satellite imagery based on convolutional neural networks [D]. Xiamen: Xiamen University. 2017.
粟佩康. 基于卷积神经网络的卫星影像道路网自动提取方法研究 [D]. 厦门: 厦门大学. 2017.

【11】Li P K, Zang Y, Wang C, et al. Road network extraction via deep learning and line integral convolution[C]//2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), July 10-15, 2016, Beijing, China. New York: , 2016, 1599-1602.

【12】Wang Y, Zhu L Q, Yu Z J, et al. Segmentation and recognition algorithm for high-speed railway scene [J]. Acta Optica Sinica. 2019, 39(6): 0610004.
王洋, 朱力强, 余祖俊, 等. 高速铁路场景的分割与识别算法 [J]. 光学学报. 2019, 39(6): 0610004.

【13】Hou Y M, Li Y P. Speech recognition of isolated words based on convolution neural networks [J]. Computer Engineering and Design. 2019, 40(6): 1751-1756.
侯一民, 李永平. 基于卷积神经网络的孤立词语音识别 [J]. 计算机工程与设计. 2019, 40(6): 1751-1756.

【14】Pei L, Liu Y, Gao L. Cloud detection of ZY-3 remote sensing images based on fully convolutional neural network and conditional random field [J]. Laser & Optoelectronics Progress. 2019, 56(10): 102802.
裴亮, 刘阳, 高琳. 结合全卷积神经网络与条件随机场的资源3号遥感影像云检测 [J]. 激光与光电子学进展. 2019, 56(10): 102802.

【15】Chang L, Deng X M, Zhou M Q, et al. Convolutional neural networks in understanding [J]. Acta Automatica Sinica. 2016, 42(9): 1300-1312.
常亮, 邓小明, 周明全, 等. 图像理解中的卷积神经网络 [J]. 自动化学报. 2016, 42(9): 1300-1312.

【16】Qin J. Tensor voting algorithm and its application [D]. Shanghai: East China Normal University. 2008.
秦菁. 张量投票算法及其应用 [D]. 上海: 华东师范大学. 2008.

【17】Wen P Z, Huang J F, Ning R H, et al. Boundary extraction of active contour based on tensor voting [J]. Computer Engineering. 2012, 38(6): 216-218.
温佩芝, 黄锦芳, 宁如花, 等. 基于张量投票的主动轮廓边缘提取 [J]. 计算机工程. 2012, 38(6): 216-218.

【18】Medioni G, Lee M S, Tang C K. The salient feature inference engine [M]. //Elsevier. A computational framework for segmentation and grouping. Amsterdam: Elsevier. 2000, 33-64.

【19】Ma W, Long Q Q, Qin Y, et al. Repairing high-definition ancient paintings based on decomposition of curves [J]. Journal of Computer-Aided Design & Computer Graphics. 2018, 30(9): 1652-1661.
马伟, 龙晴晴, 秦悦, 等. 基于画作线条结构分解的高清古画修复 [J]. 计算机辅助设计与图形学学报. 2018, 30(9): 1652-1661.

【20】Guo S Q, Zhu S S, Wang L Q, et al. De-noising methods for ultrasound image tensor voting [J]. Journal of Henan Polytechnic University (Natural Science). 2014, 33(1): 75-78.
郭树强, 朱世松, 王立群, 等. 基于张量投票法的超声波图像去噪 [J]. 河南理工大学学报(自然科学版). 2014, 33(1): 75-78.

【21】Li H X, Zhang B, Liu D, et al. Surface crack detection algorithm based on double threshold and tensor voting [J]. Laser & Optoelectronics Progress. 2018, 55(5): 051010.
李慧娴, 张斌, 刘丹, 等. 基于双重阈值和张量投票的表面裂纹检测算法 [J]. 激光与光电子学进展. 2018, 55(5): 051010.

【22】Mnih V. Machine learning for aerial image labeling [D]. Toronto: University of Toronto. 2013.

【23】Li H C, Chu H, Huo Y H. Multi-feature multiple kernels SVM-based urban road extraction Bulletin of Surveying and Mapping[J]. 0, 2018(2): 72-77.
李洪川, 楚恒, 霍英海. 城市道路的多特征多核SVM提取方法 测绘通报[J]. 0, 2018(2): 72-77.

【24】Zhang Z J, Yang F B. Road extraction algorithm for remote sensing images based on improved EM clustering [J]. Laser & Optoelectronics Progress. 2020, 57(6): 061005.
张宗军, 杨风暴. 基于改进最大期望聚类的遥感影像道路提取算法 [J]. 激光与光电子学进展. 2020, 57(6): 061005.

引用该论文

Li Tianqi,Tan Hai,Dai Jiguang,Du Yang,Wang Yang. Road Extraction Method Combining Convolutional Neural Network and Tensor Voting[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201019

李天琪,谭海,戴激光,杜阳,王杨. 结合卷积神经网络和张量投票的道路提取方法[J]. 激光与光电子学进展, 2020, 57(20): 201019

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF