计算机科学 ›› 2021, Vol. 48 ›› Issue (9): 140-145.doi: 10.11896/jsjkx.200800002

• 计算机图形学&多媒体 • 上一篇    下一篇

基于边缘特征融合的高分影像建筑物目标检测

赫晓慧1, 邱芳冰2, 程淅杰2, 田智慧1, 周广胜3   

  1. 1 郑州大学地球科学与技术学院 郑州450052
    2 郑州大学信息工程学院 郑州450001
    3 中国气象科学研究院郑州大学生态气象联合实验室 郑州450052
  • 收稿日期:2020-08-01 修回日期:2020-09-10 出版日期:2021-09-15 发布日期:2021-09-10
  • 通讯作者: 赫晓慧(hexh@zzu.edu.cn)
  • 基金资助:
    第二次青藏高原综合科学考察研究项目(2019QZKK0106)

High-resolution Image Building Target Detection Based on Edge Feature Fusion

HE Xiao-hui1, QIU Fang-bing2, CHENG Xi-jie2, TIAN Zhi-hui1, ZHOU Guang-sheng3   

  1. 1 School of Earth Science and Technology,Zhengzhou University,Zhengzhou 450052,China
    2 School of Information Engineering,Zhengzhou University,Zhengzhou 450001,China
    3 Joint Laboratory of Eco-Meteorology,Chinese Academy of Meteorological Sciences,Zhengzhou University,Zhengzhou 450052,China
  • Received:2020-08-01 Revised:2020-09-10 Online:2021-09-15 Published:2021-09-10
  • About author:HE Xiao-hui,born in 1978,professor.Her main research interests include artificial intelligence,computer vision,remote sensing image processing,and data mining.
  • Supported by:
    Second Tibetan Plateau Scientific Expedition and Research(STEP) Program(2019QZKK0106)

摘要: 高分辨率遥感图像建筑物目标检测在国土规划、地理监测、智慧城市等领域有着广泛的应用价值,但是由于遥感图像背景复杂,建筑物目标的部分细节特征与背景区分度较低,在进行检测任务时,容易出现建筑物轮廓失真、缺失等问题。针对这一问题,设计了自适应加权边缘特征融合网络(VAF-Net)。该方法针对遥感图像建筑物检测任务,对经典编解码器网络U-Net进行拓展,通过融合RGB特征图和边缘特征图,弥补了基础网络学习中的细节特征缺失;同时,借助网络的学习自动更新融合权重,实现自适应加权融合,充分利用不同特征的互补信息。该方法在Massachusetts Buildings数据集上进行了实验,其准确率、召回率和F1-score分别达到了82.1%,82.5%和82.3%,综合指标F1-score相比于基础网络提升了约6%。VAF-Net有效提高了编解码器网络对于高分影像建筑物目标检测任务的表现性能,具有良好的实用价值。

关键词: 目标检测, 特征融合, 神经网络, 边缘特征, U-Net

Abstract: High-resolution remote sensing image building target detection has a wide range of application value in territorial planning,geographic monitoring,smart city and other fields.However,due to the complex background of remote sensing images,some detailed features of building targets are less distinguishable from the background.During the task,it is prone to problems such as distortion and missing of the building outline.Aiming at this problem,an adaptive weighted edge feature fusion network (VAF-Net) is designed.This method is aimed at remote sensing image building detection tasks,expands the classic codec network U-Net network,and makes up for the lack of detailed features in basic network learning through the fusion of RGB feature maps and edge feature maps.At the same time,relying on network learning to automatically update the fusion weight,adaptive weighted fusion can be achieved,and the complementary information of different features can be full made use of.This method is tested on the Massachusetts Buildingsdata set,and its accuracy,recall and F1-score reach 82.1%,82.5% and 82.3%,respectively.The comprehensive index F1-score increases by about 6% compared to the basic network.VAF-Net effectively improves the perfor-mance of the codec network for high-resolution image building target detection tasks,and has good practical value.

Key words: Target detection, Feature fusion, Neural network, Edge feature, U-Net

中图分类号: 

  • TP391.4
[1]YANG Z,MU X D,WANG S Y.Scene classficition of remote sensing images based on multiscale features fusion[J].Optics and Precision Engineering,2018,26(12):3099-3107.
[2]LONG J,SHELHAMER E,DARRELL T.Fully Convolutional Networks for Semantic Segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,39(4):640-651.
[3]XU Z J,YANG X B,HE L M,et al.Multiscale remote sensing semantic segmentation Network[J].Computer Engineering and Applications,2020,56(21):210-217.
[4]RONNEBERGER O,FISCHER P,BROX T.U-Net:Convolu-tional Networks for Biomedical Image Segmentation[J].arXiv:1505.04597,2015.
[5]LIU H,LUO J C,HUANG B,et al.DE-Net:Deep EncodingNetwork for Building Extraction from High-Resolution Remote Sensing Imagery[J].Remote Sensing,2019,11(20):2380.
[6]XU Z H,LIU Y,QUAN J C,et al.Buildings egmentation of remote sensing images based on VGG16 pre-encoding[J].Science Technology and Engineering,2019,19(17):250-255.
[7]REN X L,WANG Y P,YANG J Y.Building Detection form Remote Sensing Images Based on Improved U-net[J].Laser and Optoelectronics Progress,2019,657(22):195-202.
[8]LI Z,ZHOU F.FSSD:Feature Fusion Single Shot Multibox Detector[J].arXiv:1712.00960,2018.
[9]SHENG Y T,ZHAO Z,WANG T T.Building Areas Extraction in GF-3 Images Based on the Integration of Span Image and Texture Features[J].Beijing Surveying and Mapping,2020,34(1):73-78.
[10]FENG F J,LI J P,DING Y Z.Target Detection from High Re-solution Remote Sensing Images Based on Combination of Multi-scale Visual Features[J].Journal of Applied Sciences,2018,36(3):471-484.
[11]LIU S,HUANG D,WANG Y.Learning Spatial Fusion for Single-Shot Object Detection[J].arXiv:1911.09516,2019.
[12]FENG F,WANG S T.Building Extraction Based on Multi-input-multi-output and Multi-feature Fusion[J].Journal of Zhengzhou Institute of Surveying and Mapping,2020,37(6):575-580.
[13]WANG Z H,LIU H Q.Building Recognition Based on Transfer Learningand Adaptive Feature Fusion[J].Computer Technology and Development,2019(12):40-43.
[14]LI X Y.Object Detection in Remote Sensing Images Based on Deep Learning[D].Hefei:University of Science and Technology of China,2019.
[15]FENG J W,ZHANG L M,DENG X Y.Image segmentationbased on multi-source fusion FCN[J].Application Research of Computers,2018,35(9):2877-2880.
[16]ZHU G Y.Research on Building Extraction from Remote Sen-sing Images Based on Deep Lreaning[D].Hangzhou:Zhejiang University,2019.
[17]JIN F,WNAG L F,LIU Z,et al.Double U-Net Remote Sensing Image Road Extraction Method[J].Journal of Geomatics Science and Technology,2019,36(4):377-381,387.
[18]LIU R Y,SUN Q C,WANG C Y.Research on Edge Detection Algorithm in Digital Image[M].Science Press,2015.
[19]PANG Y W,XIU Y X.Lane Semantic Segmentation NeuralNetwork Based on Edge Feature Merging and Skip Connections[J].Journal of Tianjin University(Science and Technology),2019,52(8):779-787.
[20]ZHANG H,ZHAO J H,ZHANG X G.High-resolution Image Building Extraction Using U-net Neural Network[J].Remote Sensing Information,2020,35(3):143-150.
[21]XU Z H,LIU Y,QUAN J C,et al.Buildings Segmentation of Remote Sensing Images Based on U-Net Pre-encoding[J].Science Technology and Engineering,2019,19(17):250-255.
[1] 黄颖琦, 陈红梅. 基于代价敏感卷积神经网络的非平衡问题混合方法[J]. 计算机科学, 2021, 48(9): 77-85.
[2] 徐涛, 田崇阳, 刘才华. 基于深度学习的人群异常行为检测综述[J]. 计算机科学, 2021, 48(9): 125-134.
[3] 张新峰, 宋博. 一种基于改进三元组损失和特征融合的行人重识别方法[J]. 计算机科学, 2021, 48(9): 146-152.
[4] 袁磊, 刘紫燕, 朱明成, 马珊珊, 陈霖周廷. 融合改进密集连接和分布排序损失的遥感图像检测[J]. 计算机科学, 2021, 48(9): 168-173.
[5] 张晓宇, 王彬, 安卫超, 阎婷, 相洁. 基于融合损失函数的3D U-Net++脑胶质瘤分割网络[J]. 计算机科学, 2021, 48(9): 187-193.
[6] 赵金龙, 赵中英. 基于异质信息网络表示学习与注意力神经网络的推荐算法[J]. 计算机科学, 2021, 48(8): 72-79.
[7] 白勇, 张占龙, 熊隽迪. 基于FP-Growth算法和GRNN的电力知识文本挖掘[J]. 计算机科学, 2021, 48(8): 86-90.
[8] 王乐, 杨晓敏. 基于感知损失的遥感图像全色锐化反馈网络[J]. 计算机科学, 2021, 48(8): 91-98.
[9] 龚浩田, 张萌. 基于关键点检测的无锚框轻量级目标检测算法[J]. 计算机科学, 2021, 48(8): 106-110.
[10] 叶中玉, 吴梦麟. 融合时序监督和注意力机制的脉络膜新生血管分割[J]. 计算机科学, 2021, 48(8): 118-124.
[11] 王施云, 杨帆. 基于U-Net特征融合优化策略的遥感影像语义分割方法[J]. 计算机科学, 2021, 48(8): 162-168.
[12] 屈立成, 吕娇, 屈艺华, 王海飞. 基于模糊神经网络的运动目标智能分配定位算法[J]. 计算机科学, 2021, 48(8): 246-252.
[13] 王炽, 常俊. 基于3D卷积神经网络的CSI跨场景手势识别方法[J]. 计算机科学, 2021, 48(8): 322-327.
[14] 王超, 魏祥麟, 田青, 焦翔, 魏楠, 段强. 基于特征梯度的调制识别深度网络对抗攻击方法[J]. 计算机科学, 2021, 48(7): 25-32.
[15] 程松盛, 潘金山. 基于深度学习特征匹配的视频超分辨率方法[J]. 计算机科学, 2021, 48(7): 184-189.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 朱淑芹,王文宏,李俊青. 针对基于感知器模型的混沌图像加密算法的选择明文攻击[J]. 计算机科学, 2018, 45(4): 178 -181 .
[2] 张盼盼, 彭长根, 郝晨艳. 一种基于隐私偏好的隐私保护模型及其量化方法[J]. 计算机科学, 2018, 45(6): 130 -134 .
[3] 郭莹莹, 张丽平, 李松. 障碍环境中线段组最近邻查询方法研究[J]. 计算机科学, 2018, 45(6): 172 -175 .
[4] 沈夏炯, 张俊涛, 韩道军. 基于梯度提升回归树的短时交通流预测模型[J]. 计算机科学, 2018, 45(6): 222 -227 .
[5] 吕巨建, 赵慧民, 陈荣军, 李键红. 基于自适应稀疏邻域重构的无监督主动学习算法[J]. 计算机科学, 2018, 45(6): 251 -258 .
[6] 赵小艳,刘宏哲,袁家政,杨少鹏. 图像重排序技术的研究进展[J]. 计算机科学, 2018, 45(5): 15 -23 .
[7] 薛善良,杨佩茹,周奚. 基于模糊神经网络的WSN无线数据收发单元故障诊断[J]. 计算机科学, 2018, 45(5): 38 -43 .
[8] 庄陵,尹耀虎. 认知异构网络中基于不完全频谱感知的资源分配算法[J]. 计算机科学, 2018, 45(5): 49 -53 .
[9] 王波涛,梁伟,赵凯利,钟汉辉,张玉圻. 基于HBase的支持频繁更新与多用户并发的R树[J]. 计算机科学, 2018, 45(7): 42 -52 .
[10] 黄铉. 特征降维技术的研究与进展[J]. 计算机科学, 2018, 45(6A): 16 -21 .