联合深度卷积神经网络的遥感影像机场识别算法

针对亚米级高分辨率遥感影像中机场识别算法存在的定位精度和识别准确率低的问题，提出了一种基于深度卷积神经网络的高分辨率遥感影像机场识别算法。首先，使用双三次插值算法对原始影像进行下采样处理并转为灰度图，进行模糊增强以得到预处理图像。其次，利用Canny算子提取灰度图边缘信息并使用概率Hough变换提取其中的直线，通过判断平行线存在与否对直线区域进行初步筛选及合并。再次，对合并后的区域利用深度卷积神经网络进行判别以得到相应区域的识别概率值。最后，通过分析概率值得到机场目标。对某卫星两种高分辨率遥感影像数据进行实验，得到识别率100%、定位准确率87.53%的实验结果，证明了所提算法的有效性和通用性。

Abstract

In order to solve the problems of low locating precision and low recognition rate of the airport identification algorithm in sub-meter high-resolution remote sensing imagesa new identification algorithm based on Deep Convolutional Neural Network (DCNN) is proposed.Firstlythe bi-cubic interpolation algorithm is used to down-sample the original single-phase remote sensing images and convert them into grayscale imagesand the pre-processed images are obtained by fuzzy enhancement.Secondlythe edge information of the images is detected by using Canny edge detection operatorand the straight line segments are extracted by using probability Hough transform.The linear regions are preliminarily screened and merged by judging whether there are parallel lines.ThenDCNN is used for judging the merged regions to acquire the recognition probability of the corresponding regions.Finallythe airport area is obtained by analyzing the probability values of the candidate regions.Simulation experiments were made to the two kinds of remote sensing images with high resolutionthe recognition rate was 100% and the mean locating accuracy was 87.53%which proved the validity and versatility of the proposed algorithm.

参考文献

[1] ITTI LKOCH CNIEBUR E.A model of saliency-based visual attention for rapid scene analysis［J］.IEEE Tran-sactions on Pattern Analysis & Machine Intelligence1998 20(11):1254-1259.

[2] SCHLKOPF BPLATT JHOFMANN T.Graph-based visual saliency［J］.Advances in Neural Information Processing Systems200719:545-552.

[3] HOU XZHANG L.Saliency detection:a spectral residual approach［C］//IEEE Conference on Computer Vision and Pattern Recognition2007:1-8.

[4] ACHANTA RHEMAMI SESTRADA Fet al.Frequency-tuned salient region detection［C］//IEEE Conference on Computer Vision and Pattern Recognition2009:1597-1604.

[5] GUO C LMA QZHANG L M.Spatio-temporal saliency detection using phase spectrum of quaternion Fourier transform［C］//IEEE Conference on Computer Vision and Pattern Recognition2008:1-8.

[6] 王强,胡维平,陆志敏,等.HOUGH变换实时检测算法研究［J］.计算机工程与设计200122(3):76-80.

[7] WANG XLV QWANG Bet al.Airport detection in remote sensing images: a method based on saliency map.［J］.Cognitive Neurodynamics20137(2):143-54.

[8] 艾淑芳,闫钧华,李大雷,等.遥感图像中的机场跑道检测算法［J］.电光与控制,201724(2):43-46.

[9] 朱丹,王斌,张立明.基于直线邻近平行性和GBVS显著性的遥感图像机场目标检测［J］.红外与毫米波学报,2015,34(3):375-384.

[10] THEODORIDIS S.Chapter 18-neural networks and deep learning［M］//Machine Learning.Amsterdam：Elsevier Ltd2015:875-936.

[11] SZEGEDY CVANHOUCKE VIOFFE Set al.Rethinking the inception architecture for computer vision［C］//IEEE Conference on Computer Vision and Pattern Recognition（CVPR）2016.doi:10.1109/CVPR.2016.308.

[12] SZEGEDY CIOFFE SVANHOUCKE Vet al.Inception-v4inception-ResNet and the impact of residual connections on learning［EB/OL］.［2017-05-07］.https://arxiv.org/pdf/1409.1556.pdf.

[13] SIMONYAN KZISSERMAN A.Very deep convolutional networks for large-scale image recognition［EB/OL］.［2017-05-18］.https://arxiv.org/pdf/1602.07261.pdf.

[14] HE KZHANG X YREN S Qet al.Deep residual learning for image recognition［C］//IEEE Conference on Computer Vision and Pattern Recognition2015:770-778.

[15] KRIZHEVSKY ASUTSKEVER IHINTON G E.ImageNet classification with deep convolutional neural networks［C］//International Conference on Neural Information Processing Systems2012:1097-1105.

[16] CLEVERT D AUNTERTHINER THOCHREITER S.Fast and accurate deep network learning by Exponential Linear Units (ELUs)［C］//ICLR2016:1-14.

[17] HE K MZHANG X YREN S Qet al.Delving deep into rectifiers:surpassing human-level performance on imageNet classification［J］.IEEE International Conference on Computer Vision(ICCV)2015:1026-1034.

[18] XU BWANG N YCHEN T Qet al.Empirical evaluation of rectified activations in convolutional network［EB/OL］.［2017-06-16］.https://arxiv.org/pdf/1505.00853.pdf.

[19] HINTON G ESRIVASTAVA NKRIZHEVSKY Aet al.Improving neural networks by preventing co-adaptation of feature detectors［J］.Computer Science20123(4):212-223.

[20] 李航.统计学习方法［M］.北京：清华大学出版社,2012:13-15.

张作省, 杨程亮, 朱瑞飞, 高放, 于野, 钟兴. 联合深度卷积神经网络的遥感影像机场识别算法[J]. 电光与控制, 2018, 25(6): 83. ZHANG Zuo-xing, YANG Cheng-liang, ZHU Rui-fei, GAO Fang, YU Ye, ZHONG Xing. An Algorithm for Recognition of Airport in Remote Sensing Image Based on DCNN Model[J]. Electronics Optics & Control, 2018, 25(6): 83.

联合深度卷积神经网络的遥感影像机场识别算法

关于本站 Cookie 的使用提示

全站搜索

联合深度卷积神经网络的遥感影像机场识别算法

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索