딥러닝 기반 3차원 라이다의 반사율 세기 신호를 이용한 흑백 영상 생성 기법
Deep Learning Based Gray Image Generation from 3D LiDAR Reflection Intensity
1
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
In this paper, we propose a method of generating a 2D gray image from LiDAR 3D reflection intensity. The proposed method uses the Fully Convolutional Network (FCN) to generate the gray image from 2D reflection intensity which is projected from LiDAR 3D intensity. Both encoder and decoder of FCN are configured with several convolution blocks in the symmetric fashion. Each convolution block consists of a convolution layer with 3×3 filter, batch normalization layer and activation function. The performance of the proposed method architecture is empirically evaluated by varying depths of convolution blocks. The well-known KITTI data set for various scenarios is used for training and performance evaluation. The simulation results show that the proposed method produces the improvements of 8.56 dB in peak signal-to-noise ratio and 0.33 in structural similarity index measure compared with conventional interpolation methods such as inverse distance weighted and nearest neighbor. The proposed method can be possibly used as an assistance tool in the night-time driving system for autonomous vehicles.
Abstract
In this paper, we propose a method of generating a 2D gray image from LiDAR 3D reflection intensity. The proposed method uses the Fully Convolutional Network (FCN) to generate the gray image from 2D reflection intensity which is projected from LiDAR 3D intensity. Both encoder and decoder of FCN are configured with several convolution blocks in the symmetric fashion. Each convolution block consists of a convolution layer with 3×3 filter, batch normalization layer and activation function. The performance of the proposed method architecture is empirically evaluated by varying depths of convolution blocks. The well-known KITTI data set for various scenarios is used for training and performance evaluation. The simulation results show that the proposed method produces the improvements of 8.56 dB in peak signal-to-noise ratio and 0.33 in structural similarity index measure compared with conventional interpolation methods such as inverse distance weighted and nearest neighbor. The proposed method can be possibly used as an assistance tool in the night-time driving system for autonomous vehicles.
참고문헌 (18)
- 이무현, "2D 라이다 데이터베이스 기반 장애물 분류 기법", 대한임베디드공학회논문지, 10(3), pp.179~188 (2015), 대한임베디드공학회[DOI]
- 권성경, "저해상도 2D 라이다의 사람 특성 함수를 이용한 새로운 사람 감지 기법", 대한임베디드공학회논문지, 11(5), pp.267~276 (2016), 대한임베디드공학회[DOI]
- D. S. Hall, "Color lidar scanner", U.S. Patent 8,675,181, ~ (2014)
- T. D. Barfoot, "Into Darkness: Visual Navigation Based on a Lidar-intensity-image Pipeline", Proceedings of International Symposium Robotics Research, 114, pp.487~504 (2016)
- I. Ashraf, "An Investigation of Interpolation Techniques to Generate 2d Intensity Image from Lidar Data", Journal of IEEE Access, 5, pp.8250~8260 (2017)
- A. Asvadi, "Realtime Deep Convnet-based Vehicle Detection Using 3d-lidar Reflection Intensity Data", Proceedings of Iberian Robotics Conference, 694, pp.475~486 (2017)
- V. Badrinarayanan, "Segnet: A Deep Convolutional Encoder-decoder Architecture for Image Segmentation", Journal of IEEE Transaction Pattern Anal. Mach. Intel, 39(12), pp.2481~2495 (2017)
- R. Yasrab, "An Encoder-decoder Based Convolution Neural Network (cnn) for Future Advanced Driver Assistance System (adas)", Journal of Applied Sciences, 7(4), pp.312~ (2017)
- D. Varga, "Fully Automatic Image Colorization Based on Convolutional Neural Network", Proceedings of IEEE International Conference Pattern Recognit, pp.3691~3696 (2016)
- F. Baldassarre, "Deep Koalarization: Image Colorization Using Cnns and Inceptionresnet-v2", CoRR, pp.1~12 (2017)
- P. Isola, "Image-to-image Translation with Conditional Adversarial Networks", CoRR, pp.1~17 (2016)
- W. Chen, "Sketchygan: Towards Diverse and Realistic Sketch to Image Synthesis", Proceedings of IEEE Conference Comput. Vis. Pattern Recognit, pp.9416~9425 (2018)
- P. L. Surez, "Infrared Image Colorization Based on a Triplet Dcgan Architecture", Proceedings of IEEE Conference Comput. Vis. Pattern Recognit, pp.212~217 (2017)
- U. Qayynm, "Thermal Colorization Using Deep Neural Network", Proceedings of IEEE Interenational Conference Applied Science Tech, pp.325~329 (2018)
- A. Geiger, "Vision Meets Robotics: The Kitti Dataset", Journan of International J. Robotics Research, 32(11), pp.1231~1237 (2013)
- D. P. Kingma, "Adam: A Method for Stochastic Optimization", CoRR, pp.1~15 (2014)
- A. Hore, "Image Quality Metrics:Psnr vs. ssim", Proceedings of IEEE Conference Pattern Recognit, pp.2366~2369 (2010)
- S. Hochreiter, "The Vanishing Gradient Problem During Learning Recurrent Neural Nets and Problem Solutions", Journal of International J. Uncert., Fuzzi. and Knowledge-Based Syst., 6(2), pp.107~116 (1998)