Point Cloud Classification and Accuracy Analysis Based on Feature Fusion

doi:10.11947/j.JGGS.2021.0304

Abstract

Abstract:

A method for land-cover classification was proposed based on the fusion of features generated from waveform data and point cloud respectively. It aims to partially overcome the ineffectiveness of many traditional classifiers caused by the fact that point cloud is lacking spectral information. The whole flowchart of the method is as follows: Firstly, Gaussian decomposition was applied to fit an echo full-waveform. The parameters associated with the Gaussian function were optimized by LM (Levenberg-Marquard) algorithm. Six and thirteen features were generated to describe the waveform characteristics and the local geometry of point cloud, respectively. Secondly, a random forest was selected as the classifier to which the generated features were input. Relief-F was used to rank the weights of all the features generated. Finally, features were input to the classifier one by one according to the weights calculated from feature ranking, where classification accuracies were evaluated. The experimental results show that the effectiveness of the fusion of features generated from waveform and point cloud for LiDAR data classification, with 95.4% overall accuracy, 0.90 kappa coefficient, which outperform the results obtained by a single class of features, no matter whether they were generated from point cloud or waveform data.

Key words: airborne LiDAR; full-waveform data; feature fusion; land-cover classification

Xiaochen WANG,Hongchao MA,Liang ZHANG,Zhan CAI,Haichi MA. Point Cloud Classification and Accuracy Analysis Based on Feature Fusion[J]. Journal of Geodesy and Geoinformation Science, 2021, 4(3): 38-48.

Figures/Tables 10

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Tab.1

Fig.7

Fig.8

Tab.2

References 39

[1]	ELIZAROV V V, GRISHKANICH A S, KASCHEEV S V, et al. Lidar scanning module for remote environmental monitoring[C]//Proceedings of 2016 International Conference Laser Optics. St. Petersburg, Russia: IEEE, 2016.
[2]	DONG Weihua, LAN Jianhang, LIANG Shunlin, et al. Selection of LiDAR geometric features with adaptive neighborhood size for urban land cover classification[J]. International Journal of Applied Earth Observation and Geoinformation, 2017,60:99-110. doi: 10.1016/j.jag.2017.04.003
[3]	SHAN Jie, TOTH C K. Topographic laser ranging and scanning: Principles and processing[M]. Boca Raton, FL: CRC Press, 2009: 3-7.
[4]	LUO Shezhou, WANG Cheng, XI Xiaohuan, et al. Fusion of airborne discrete-return LiDAR and hyperspectral data for land cover classification[J]. Remote Sensing, 2016,8(1):3. doi: 10.3390/rs8010003
[5]	SANKEY T, DONAGER J, MCVAY J, et al. UAV lidar and hyperspectral fusion for forest monitoring in the southwestern USA[J]. Remote Sensing of Environment, 2017,195:30-43. doi: 10.1016/j.rse.2017.04.007
[6]	AWAD M M. Toward robust segmentation results based on fusion methods for very high resolution optical image and LiDAR data[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017,10(5):2067-2076. doi: 10.1109/JSTARS.4609443
[7]	POLIYAPRAM V, WANG Weimin, NAKAMURA R. A point-wise LiDAR and image multimodal fusion network (PMNet) for aerial point cloud 3D semantic segmentation[J]. Remote Sensing, 2019,11(24):2961. doi: 10.3390/rs11242961
[8]	WANG Xue, LI Peijun. Extraction of urban building damage using spectral, height and corner information from VHR satellite images and airborne LiDAR data[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020,159:322-336. doi: 10.1016/j.isprsjprs.2019.11.028
[9]	NORMAN M, SHAFRI H Z M, MANSOR S, et al. Fusion of multispectral imagery and LiDAR data for roofing materials and roofing surface conditions assessment[J]. International Journal of Remote Sensing, 2020,41(18):7090-7111. doi: 10.1080/01431161.2020.1754493
[10]	CAO Qiong, MA Ailong, ZHONG Yanfei, et al. Urban classification by multi-feature fusion of hyperspectral image and LiDAR data[J]. Journal of Remote Sensing, 2019,23(5):892-903.
[11]	WANG Minye, LIU Rufei, LU Xiushan, et al. The use of mobile lidar data and Gaofen-2 image to classify roadside trees[J]. Measurement Science and Technology, 2020,31(12):125005. doi: 10.1088/1361-6501/aba322
[12]	NGUYEN T H, DANIEL S, GUÉRIOT D, et al. Super-resolution-based snake model-an unsupervised method for large-scale building extraction using airborne LiDAR data and optical image[J]. Remote Sensing, 2020,12(11):1702. doi: 10.3390/rs12111702
[13]	OZKAN S, AKAR G B. Hyperspectral data to relative lidar depth: An inverse problem for remote sensing[C]//Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. Long Beach, CA: IEEE, 2019.
[14]	BRUGGISSER M, RONCAT A, SCHAEPMAN M E, et al. Retrieval of higher order statistical moments from full-waveform LiDAR data for tree species classification[J]. Remote Sensing of Environment, 2017,196:28-41. doi: 10.1016/j.rse.2017.04.025
[15]	ZORZI S, MASET E, FUSIELLO A, et al. Full-waveform airborne LiDAR data classification using convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019,57(10):8255-8261. doi: 10.1109/TGRS.36
[16]	MARSELIS S M, ABERNETHY K, ALONSO A, et al. Evaluating the potential of full‐waveform lidar for mapping pan-tropical tree species richness[J]. Global Ecology and Biogeography, 2020,29(10):1799-1816. doi: 10.1111/geb.v29.10
[17]	YANG Xuebo, WANG Cheng, XI Xiaohuan, et al. Extraction of multiple building heights using ICESat/GLAS full-waveform data assisted by optical imagery[J]. IEEE Geoscience and Remote Sensing Letters, 2019,16(12):1914-1918. doi: 10.1109/LGRS.2019.2911967
[18]	LAI Xudong, YUAN Yifei, LI Yongxu, et al. Full-waveform LiDAR point clouds classification based on wavelet support vector machine and ensemble learning[J]. Sensors, 2019,19(14):3191. doi: 10.3390/s19143191
[19]	SHINOHARA T, XIU Haoyi, MATSUOKA M. FWNet: Semantic segmentation for full-waveform LiDAR data using deep learning[J]. Sensors, 2020,20(12):3568. doi: 10.3390/s20123568
[20]	MALLET C, SOERGEL U, BRETAR F. Analysis of full-waveform lidar data for classification of urban areas[J]. Photogrammetrie Fernerkundung Geoinformation, 2008,5(5):337-349.
[21]	MALLET C, BRETAR F. Full-waveform topographic lidar: State-of-the-art[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2009,64(1):1-16. doi: 10.1016/j.isprsjprs.2008.09.007
[22]	RONCAT A, WAGNER W, MELZER T, et al. Echo detection and localization in full-waveform airborne laser scanner data using the averaged square difference function estimator[J]. The Photogrammetric Journal of Finland, 2008,21(1):62-75.
[23]	WAGNER W, ULLRICH A, DUCIC V, et al. Gaussian decomposition and calibration of a novel small-footprint full-waveform digitising airborne laser scanner[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2006,60(2):100-112. doi: 10.1016/j.isprsjprs.2005.12.001
[24]	JUTZI B, STILLA U. Range determination with waveform recording laser systems using a Wiener Filter[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2006,61(2):95-107. doi: 10.1016/j.isprsjprs.2006.09.001
[25]	WU Jiaying, VAN AARDT J A N, ASNER G P. A comparison of signal deconvolution algorithms based on small-footprint LiDAR waveform simulation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2011,49(6):2402-2414. doi: 10.1109/TGRS.2010.2103080
[26]	KOMA Z, KOENIG K, HÖFLE B. Urban tree classification using full-waveform airborne laser scanning[C]//Proceedings of the ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Prague, Czech Republic: ISPRS, 2016: 185-192.
[27]	MA Hongchao, ZHOU Weiwei, ZHANG Liang. DEM refinement by low vegetation removal based on the combination of full waveform data and progressive TIN densification[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2018,146:260-271. doi: 10.1016/j.isprsjprs.2018.09.009
[28]	ALEXANDER C, TANSEY K, KADUK J, et al. Backscatter coefficient as an attribute for the classification of full-waveform airborne laser scanning data in urban areas[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2010,65(5):423-432. doi: 10.1016/j.isprsjprs.2010.05.002
[29]	WEINMANN M, JUTZI B, MALLET C. Semantic 3D scene interpretation: A framework combining optimal neighborhood size selection with relevant features[C]//Proceedings of the ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Zurich, Switzerland: ISPRS, 2014: 181-188.
[30]	GUO Bo, HUANG Xianfeng, ZHANG Fan, et al. Classification of airborne laser scanning data using JointBoost[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015,100:71-83. doi: 10.1016/j.isprsjprs.2014.04.015
[31]	DONG Weihua, LAN Jianhang, LIANG Shunlin, et al. Selection of LiDAR geometric features with adaptive neighborhood size for urban land cover classification[J]. International Journal of Applied Earth Observation and Geoinformation, 2017,60:99-110. doi: 10.1016/j.jag.2017.04.003
[32]	GRESSIN A, MALLET C, DEMANTKÉ J, et al. Towards 3D lidar point cloud registration improvement using optimal neighborhood knowledge[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2013,79:240-251. doi: 10.1016/j.isprsjprs.2013.02.019
[33]	BREIMAN L. Random forests[J]. Machine Learning, 2001,45(1):5-32. doi: 10.1023/A:1010933404324
[34]	MALLET C, BRETAR F, ROUX M, et al. Relevance assessment of full-waveform lidar data for urban area classification[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2011,66(S6):S71-S84. doi: 10.1016/j.isprsjprs.2011.09.008
[35]	CAI Zhan, MA Hongchao, ZHANG Liang. Feature selection for airborne LiDAR data filtering: A mutual information method with Parzon window optimization[J]. GIScience & Remote Sensing, 2020,57(3):323-337.
[36]	LI Xiaolan. The study and application of feature selection algorithms based on relief[D]. Dalian: Dalian University of Technology, 2013.
[37]	KONONENKO I. Estimating attributes: Analysis and extensions of RELIEF[C]//Proceedings of the European Conference on Machine Learning. Italy: Springer, 1994.
[38]	KIRA K, RENDELL L A. A practical approach to feature selection[C]//Proceedings of the Ninth International Workshop on Machine Learning. Aberdeen, Scotland: Morgan Kaufmann Publishers Inc., 1992.
[39]	MA Hongchao, CAI Zhan, ZHANG Liang. Comparison of the filtering models for airborne LiDAR data by three classifiers with exploration on model transfer[J]. Journal of Applied Remote Sensing, 2018,12(1):016021.

Predicted label
		Ground	Low-vegetation	High-vegetation	Building	Omission/(%)
True label	Ground	132844	1664	214	1745	2.65
	Low-vegetation	2951	29488	198	409	10.77
	High-vegetation	8	251	9932	33	2.86
	Building	1017	338	83	11770	10.89
	Commission/(%)	2.91	7.10	4.75	15.67	—
Overall accuracy: 95.4% Kappa coefficient: 0.90

	Point cloud		Waveform			Point cloud& waveform			Relief-F (first 14 features)			TerraScan
	Ⅰ	Ⅱ		Ⅰ	Ⅱ		Ⅰ	Ⅱ		Ⅰ	Ⅱ		Ⅰ	Ⅱ
Ground	4.03	3.22	7.45		13.23	2.65		2.91	3.12		2.56	26.5		1.63
Low-vegetation	11.8	8.56	33.5		30.86	10.7		7.10	8.73		7.81	5.59		55.32
High-vegetation	2.59	4.56	63.6		32.65	2.86		4.75	3.23		4.64	8.21		16.45
Building	11.7	23.75	48.1		31.92	10.9		15.67	11.9		17.88	26.5		1.74
Overall accuracy /(%)	94.20		82.33			95.40			95.31			78.07