A method of deep convolutional neural network photo classification for landuse scenes

doi:10.13474/j.cnki.11-2246.2020.0039

Abstract

Abstract: The multi-angle real-life proofed photos of the national land survey have multiple characteristics:high resolution, rich layers, and clear sections. The photos are accurate and detailed depictions of the occurrence status and scenes of landuse maps, which makes up for the lack of ground perspective of remote sensing images. Based on semantic segmentation, this paper proposes an innovative landuse scene classification method with a deep convolutional neural network (DCNN) for ground perspective photos. The method annotates photo scenes with multi-semantic technique and determines categories of photos of landuse with semantic composition. Successfully applied in the third automatic survey of land survey photos, this method reduces the workload of manual interpretation and improved the accuracy of automatic identification of landuse scenes.

Key words: land survey, real photos, deep convolutional neural networks, multiple semantic annotations, semantic composition, land use scenario classification

CLC Number:

P237

XU Shiwu, ZENG Jue, ZHANG Shihui, LI Changzheng, LI Tingyu. A method of deep convolutional neural network photo classification for landuse scenes[J]. Bulletin of Surveying and Mapping, 2020, 0(2): 24-28,42.

References

[1] SHI H, CHEN L, BI F, et al. Accurate urban area detection in remote sensing images[J]. IEEE Geoscience and Remote Sensing Letters, 2015, 12(9):1948-1952.
[2] OZDARICI-OK A, OK A, SCHINDLER K. Mapping of agricul-tural crops from single high-resolution multispectral images——data-driven smoothing vs. parcel-based smoothing[J]. Remote Sensing, 2015, 7(5):5611-5638.
[3] REDMON J, FARHADI A. YOLO9000:better, faster, stronger[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).[S.l.]:IEEE, 2017:6517-6525.
[4] 张富凯, 杨峰, 李策. 基于改进YOLOv3的快速车辆检测方法[J]. 计算机工程与应用, 2019, 55(2):18-26.
[5] 姚群力,胡显,雷宏.深度卷积神经网络在目标检测中的研究进展[J].计算机工程与应用,2018,54(17):1-9.
[6] 谢林江,季桂树,彭清,等.改进的卷积神经网络在行人检测中的应用[J].计算机科学与探索,2018,12(5):708-718.
[7] 许夙晖, 慕晓冬, 赵鹏,等.利用多尺度特征与深度网络对遥感影像进行场景分类[J]. 测绘学报, 2016,45(7):834-840.
[8] HE K, GKIOXARI G, DOLLÁR P, et al. Mask r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision.[S.l.]:IEEE, 2017:2961-2969.
[9] HU F, XIA G S, HU J, et al. Transferring deep convo-lutional neural networks for the scene classification of high-resolution remote sensing imagery[J]. Remote Sensing, 2015, 7(11):14680-14707.
[10] HU F, TONG X, XIA G S, et al. Delving into deep representations for remote sensing image retrieval[C]//Proceedings of IEEE the 13th International Conference on Signal Processing (ICSP).[S.l.]:IEEE, 2016:198-203.
[11] HUANG B, ZHAO B, SONG Y. Urban land-use mapping using a deep convolutional neural network with high spatial resolution multispectral remote sensing imagery[J]. Remote Sensing of Environment, 2018, 214:73-86.
[12] CHAKRABORTY S, BALASUBRAMANIAN V, SUN Q, et al. Active batch selection via convex relaxations with guaranteed solution bounds[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(10):1945-1958.
[13] 王振国, 陈宏宇, 徐文明. 利用DCNN融合特征对遥感图像进行场景分类[J].电子设计工程,2018,26(1):189-193.
[14] LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521:436-444.
[15] 韩丽燕. 基于散焦测距的几何尺寸测量技术研究[D]. 太原:中北大学, 2011.