A Horizon Detection Method Based on Deep Learning and Random Forest

doi:10.16182/j.issn1004731x.joss.201807010

Abstract

Abstract: The detection effect of existing horizon line detection methods is greatly affected by the environment, and the computational complexity is high. Aiming at the problem of horizon line detection in complex road scene in real-life, a horizon line detection method based on deep learning and random forest is proposed. The deep learning model is used to extract the depth features, then the obtained depth features are used for random forest training. The results of horizon line detection are obtained by random forest regression-voting. The simulation results show that this method has good detection effect. The detection results are not only similar to the real value on the straight road, but also are basically coincident with the true value in the shadow and the curve area. It shows that the method is robust. It can be used to detect the horizon line in complex road scene.

Key words: horizon line, deep learning, random forest, complex road scene, simulation

CLC Number:

TP391

Ye Jihua, Shi Shuxia, Li Hanxi, Zuo Jiali, Wang Shimin. A Horizon Detection Method Based on Deep Learning and Random Forest[J]. Journal of System Simulation, 2018, 30(7): 2507-2514.

References

[1] T Ahmad, G Bebis, M Nicolescu, et al.An Edge-less Approach to Horizon Line Detection[C]// 14th International Conference on Machine Learning and Applications, Miami, United States, ICMLA 2015, 1095-1102.
[2] A Pouryazdanpanah, E E Regentova, V Muthukumar, et al.Real-time horizon line detection based on fusion of classﬁcation and clustering[J]. International Journal of Computer Applications (S0975-8887), 2015, 121(10): 5-11.
[3] R Tonge, S Maji, C V Jawahar.Parsing world's skylines using shape-constrained MRFs[C]//Columbus, Ohio:CVPR2014: 3174-3181.
[4] Donghao Zhang, Xuming He, Hanxi Li.Data-driven street scene layout estimation for distant object detection[C]//International Conference on Digital Image Computing: Techniques and Applications, Wollongong, New South Wales, Australia:DICTA2014: 1-7.
[5] 赵世峰, 张海, 范耀祖. 一种基于计算机视觉的飞行器姿态估计算法[J]. 北京航空航天大学学报, 2006, 32(8): 885-889.
Zhao Shifeng, Zhang Hai, Fan Yaozu.Attitude estmiation method for flight vehicles based on computer vision[J]. Journal of Beijing University ofAeronautics and Astronautics, 2006, 32(8): 885-889.
[6] Chen Y J, Abushakra A, Lee J.Vision-based Horizon Detection and Target Tracking for UAVs[C]//Las Vegas, NV, USA: ISVC2011: 310-319.
[7] Fefilatyev S, Smarodzinava V, Hall L O, et al.Horizon Detection using Machine Learning Technique[C]// Proceedings of the 5th International Conference on Machine Learning and Applications, Orlando, Florida, USA:ICMLA2006: 17-21.
[8] 程序, 郝群, 宋勇, 等. 基于直线模型的微型飞行器姿态角计算[J]. 北京理工大学学报, 2010, 30(7): 798-802.
Cheng Xu, Hao Qun, Song Yong, et al.Method for Calculating MAV Attitude Angles Based on Straight Line Model[J]. Transactions of Beijing Institute of Technology, 2010, 30(7): 798-802
[9] Zafarifar B, Weda H.Horizon detection based on sky-color and edge features[C]//Proceedings of SPIE, San Diego, California, United States: SPIE, 2008: 6822201-6822209.
[10] Nechyba S M, Ifju M C, Waszak P G.Vision-guided Flight Stability and Control for Micro Air Vehicles[C]// International Conference on Intelligent Robots and Systems, Lausanne, Switzerland:IROS2002: 2134-2140.
[11] Thurrowgood S, Soccol D, Moore R J D, et al. A Vision Based System for Attitude Estimation of UAVs[C]//The2009 IEEE/RSJ International Conference on Intelligent Robots and Systems, Louis, MO, USA: IROS2009: 5725-5727.
[12] 乔体洲, 戴树岭. 基于回归森林的面部姿态分析[J]. 计算机辅助设计与图形学学报, 2014, 26(7): 1151-1158.
Qiao Tizhou, Dai Shuling.Regression Forests for Head Pose Estimation Analysis[J]. Journal of Computer-Aided Design&Computer Graphics, 2014, 26(7): 1151-1158.
[13] 王丽婷, 丁晓青, 方驰. 基于随机森林的人脸关键点精确定位方法[J]. 清华大学学报, 2009, 49(4): 543-546.
Wang Liting, Ding Xiaoqing, Fang Chi.Accurate localization of facial feature points based on random forest classifier[J]. J Tsinghua Univ (Sci& Tech), 2009, 49(4): 543-546.
[14] Claudia Lindner, Paul A Bromiley, Mircea C lonita, et al. Robust and Accurate Shape Model Matching Using Random Forest Regression-Voting[C]//UNITED STATES, Transactions on Pattern Analysis and Machine Intelligence, 2015: 1862-1874.
[15] Wei Liu, Dragomir Anguelov, Dumitru Erhan, et al.SSD: Single Shot MultiBox Detector[C]//Amsterdam: ECCV2016.
[16] Girshick R.Fast R-CNN[C]//ICCV, IEEE International Conference on Computer Vision, Santiago, Chile: ICCV2015, 1440-1448.
[17] Ren S, He K, Girshick R, et al.Faster R-CNN: Towards real-time object detection with region proposal networks[C]//Montreal: International Conference on Neural Information Processing Systems, 2015, 39(6): 91-99.