基于深度卷積神經(jīng)網(wǎng)絡(luò)的場景自適應(yīng)道路分割算法

王海; 蔡英鳳; 賈允毅; 陳龍; 江浩斌

doi:10.11999/JEIT160329

基于深度卷積神經(jīng)網(wǎng)絡(luò)的場景自適應(yīng)道路分割算法

doi: 10.11999/JEIT160329

1.
江蘇大學(xué)汽車與交通工程學(xué)院鎮(zhèn)江 212013
2.
江蘇大學(xué)汽車工程研究院鎮(zhèn)江 212013
3.
克萊姆森大學(xué)汽車工程系美國南卡羅拉納州 29634

基金項(xiàng)目:

國家自然科學(xué)基金(U1564201, 61601203, 61573171, 61403172)，中國博士后基金(2014M561592, 2015T80511)，江蘇省重點(diǎn)研發(fā)計(jì)劃(BE2016149)，江蘇省自然科學(xué)基金(BK20140555)，江蘇省六大人才高峰項(xiàng)目(2015-JXQC-012, 2014-DZXX-040)

計(jì)量
- 文章訪問數(shù): 2113
- HTML全文瀏覽量: 270
- PDF下載量: 943
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2016-04-05
- 修回日期: 2016-08-22
- 刊出日期: 2017-02-19

Scene Adaptive Road Segmentation Algorithm Based on Deep Convolutional Neural Network

1.
School of Automotive and Traffic Engineering, Jiangsu University, Zhenjiang 212013, China
2.
Automotive Engineering Research Institute, Jiangsu University, Zhenjiang 212013, China
3.
Department of Automotive Engineering, Clemson University, South Carolina 29634, USA

Funds:

The National Natural Science Foundation of China (U1564201, 61601203, 61573171, 61403172), The China Postdoctoral Science Foundation (2014M561592, 2015T80511), The Key Research and Development Program of Jiangsu Province (BE2016149), The Natural Science Foundation of Jiangsu Province (BK20140555), The Six Talent Peaks Project of Jiangsu Province (2015-JXQC-012, 2014-DZXX-040)

摘要

摘要: 現(xiàn)有基于機(jī)器學(xué)習(xí)的道路分割方法存在當(dāng)訓(xùn)練樣本和目標(biāo)場景樣本分布不匹配時(shí)檢測效果下降顯著的缺陷。針對該問題，該文提出一種基于深度卷積網(wǎng)絡(luò)和自編碼器的場景自適應(yīng)道路分割算法。首先，采用較為經(jīng)典的基于慢特征分析(SFA)和GentleBoost的方法，實(shí)現(xiàn)了帶標(biāo)簽置信度樣本的在線選??；其次，利用深度卷積神經(jīng)網(wǎng)絡(luò)(DCNN)深度結(jié)構(gòu)的特征自動抽取能力，輔以特征自編碼器對源-目標(biāo)場景下特征相似度度量，提出了一種采用復(fù)合深度結(jié)構(gòu)的場景自適應(yīng)分類器模型并設(shè)計(jì)了訓(xùn)練方法。在KITTI測試庫的測試結(jié)果表明，所提算法較現(xiàn)有非場景自適應(yīng)道路分割算法具有較大的優(yōu)越性，在檢測率上平均提升約4.5%。
- 道路分割 /
- 場景自適應(yīng) /
- 深度卷積神經(jīng)網(wǎng)絡(luò) /
- 復(fù)合深度結(jié)構(gòu) /
- 自編碼器
Abstract: The existed machine learning based road segmentation algorithms maintain obvious shortage that the detection effect decreases dramatically when the distribution of training samples and the scene target samples does not match. Focusing on this issue, a scene adaptive road segmentation algorithm based on Deep Convolutional Neural Network (DCNN) and auto encoder is proposed. Firstly, classic Slow Feature Analysis (SFA) and Gentle Boost based method is used to generate online samples whose label contain confidence value. After that, using the automatic feature extraction ability of DCNN and performing source-target scene feature similarity calculation with deep auto-encoder, a composite deep structure based scene adaptive classifier and its training method are designed. The experiment on KITTI dataset demonstrates that the proposed method outperforms the existed machine learning based road segmentation algorithms which upgrades the detection rate on average of around 4.5%.
- Road segmentation /
- Scene adaptive /
- Deep Convolutional Neural Network (DCNN) /
- Composite deep structure /
- Auto-encoder

HTML全文

參考文獻(xiàn)(21)

余天洪, 王榮本, 顧柏園, 等. 基于機(jī)器視覺的智能車輛前方道路邊界及車道標(biāo)識識別方法綜述[J]. 公路交通科技, 2006, 38(8): 139-142.

YU Tianhong, WANG Rongben, GU Baiyuan, et al. Survey on the vision-based recognition methods of intelligent vehicle road boundaries and lane markings[J]. Journal of Highway and Transportation Research and Development, 2006, 38(8): 139-142.

ZHOU H, KONG H, WEI L, et al. Efficient road detection and tracking for unmanned aerial vehicle[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(1): 297-309. doi: 10.1109/TITS.2014.2331353.

SHIN B S, XU Z, and KLETTE R. Visual lane analysis and higher-order tasks: a concise review[J]. Machine Vision and Applications, 2014, 25(6): 1519-1547. doi: 10.1007/s00138- 014-0611-8.

HILLEL A B, LERNER R, LEVI D, et al. Recent progress in road and lane detection: a survey[J]. Machine Vision and Applications, 2014, 25(3): 727-745. doi: 10.1007/s00138-011- 0404-2.

PAZ L M, PINIES P, and NEWMAN P. A variational approach to online road and path segmentation with monocular vision[C]. 2015 IEEE International Conference on Robotics and Automation (ICRA), Seattle, USA, 2015: 1633-1639. doi: 10.1109/ICRA.2015.7139407.

PASSANI M, YEBES J J, and BERGASA L M. Fast pixelwise road inference based on Uniformly Reweighted Belief Propagation[C]. 2015 IEEE Intelligent Vehicles Symposium (IV), Seoul, Korea, 2015: 519-524. doi: 10.1109/ IVS.2015.7225737.

LADICKY L, STURGESS P, RUSSELL C, et al. Joint optimization for object class segmentation and dense stereo reconstruction[J]. International Journal of Computer Vision, 2012, 100(2): 122-133. doi: 10.1007/s11263-011-0489-0.

STURGESS P, ALAHARI K, LADICKY L, et al. Combining appearance and structure from motion features for road scene understanding[C]. BMVC 2012-23rd British Machine Vision Conference, Guildford, UK, 2012: 1-10.

KUHNL T, KUMMERT F, and FFITACH J. Monocular road segmentation using slow feature analysis[C]. 2011 IEEE Conference on Intelligent Vehicles Symposium (IV), Baden-　Baden, Germany, 2011: 800-806. doi: 10.1109/IVS.2011. 5940416.

肖良, 戴斌, 吳濤, 等. 基于字典學(xué)習(xí)與稀疏表示的非結(jié)構(gòu)化道路分割方法[J]. 吉林大學(xué)學(xué)報(bào)(工學(xué)版), 2013, 43(S1): 384-388.

XIAO Liang, DAI bin, WU Tao, et al. Unstructured road segmentation method based on dictionary learning and sparse representation[J]. Journal of Jilin University (Engineering and Technology Edition), 2013, 43(S1): 384-388.

李駿揚(yáng), 金立左, 費(fèi)樹岷, 等. 基于多尺度特征表示的城市道路檢測[J]. 電子與信息學(xué)報(bào), 2014, 36(11): 2578-2585. doi: 10.3724/SP.J.1146.2014.00271.

LI Junyang, JIN Lizuo, FEI Shumin, et al. Urban road detection based on multi-scale feature representation[J]. Journal of Electronics Information Technology, 2014, 36(11): 2578-2585. doi: 10.3724/SP.J.1146.2014.00271.

ALVAREZ J M, LECUN Y, GEVERS T, et al. Semantic road segmentation via multi-scale ensembles of learned features[C]. 2012 Workshops and Demonstrations Computer Vision of ECCV, Firenze, Italy, 2012: 586-595. doi: 10.1007/978-3-642- 33868-7_58.

WISKOTT L and SEJNOWAKI T J. Slow feature analysis: unsupervised learning of invariances[J]. Neural Computation, 2002, 14(4): 715-770. doi: 10.1162/089976602317318938.

KRIZHEVAKY A, SUTAKEVER I, and HINTON G E.

Imagenet classification with deep convolutional neural networks[C]. Advances in Neural Information Processing Systems. South Lake Tahoe, Nevada, USA, 2012: 1097-1105.

BROSTOW G J, FAUQUEUR J, and CIPOLLA R. Semantic object classes in video: A high-definition ground truth database[J]. Pattern Recognition Letters, 2009, 30(2): 88-97. doi: 10.1016/j.patrec.2008.04.005.

GEIGER A, LENZ P, and URTASUN R. Are we ready for autonomous driving? The kitti vision benchmark suite[C]. 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Providence, RI, USA, 2012: 3354-3361. doi: 10.1109/CVPR.2012.6248074.

LVAREZ J M and LOPEZ A M. Road detection based on illuminant invariance[J]. IEEE Transactions on Intelligent Transportation Systems, 2011, 12(1): 184-193. doi: 10.1109/ TITS.2010.2076349.

相關(guān)文章

施引文獻(xiàn)

資源附件(0)

訪問統(tǒng)計(jì)