Traffic sign recognition method based on graphical model and convolutional neural network
-
摘要: 为了提高交通标志识别的鲁棒性, 提出了一种基于图模型与卷积神经网络(CNN)的交通标志识别方法, 建立了一个面向应用的基于区域的卷积神经网络(R-CNN)交通标志识别系统。构造了基于超轮廓图(UCM)超像素区域的图模型, 有效利用自底向上的多级信息, 提出了一种基于图模型的层次显著性检测方法, 以提取交通标志感兴趣区域, 并利用卷积神经网络对感兴趣候选区进行特征提取与分类。检测结果表明: 针对限速标志, 基于UCM超像素区域的图模型比基于简单线性迭代聚类(SLIC)超像素的图模型更有利于获取上层显著度图的大尺度结构信息; 基于先验位置约束与局部特征(颜色与边界)的层次显著性模型有效地融合了局部区域的细节信息与结构信息, 检测结果更加准确, 检测目标更加完整、均匀, 查准率为0.65, 查全率为0.8, F指数为0.73, 均高于其他同类基于超像素的显著性检测算法; 基于具体检测任务的CNN预训练策略扩展了德国交通标志识别库(GTSRB)的样本集, 充分利用了CNN的海量学习能力, 更好地学习目标内部的局部精细特征, 提高了学习与识别能力, 总识别率为98.85%, 高于SVM分类器的95.73%。Abstract: In order to improve the recognition robustness of traffic sign, a recognition method of traffic sign based on graphical model and convolutional neural network(CNN)was proposed, and aapplication-oriented recognition system of traffic sign based on the regions with convolutional neural network(R-CNN)was set up.A graphical model based on UCM superpixel region was constructed, the multi-scales information from bottom to up was used efficiently, and a hierarchical saliency detection method based on graphical model was proposed to extract the interest regions of traffic sign.The candidate interest regions were processed for the feature extraction and classification of traffic sign with CNN.Detecting result indicates that aiming at the speed-limited signs, the graphical model based on UCM superpixel can get more larger-scale construction information of upper saliency map contrasting to the graphical model based on simple linear iterative cluster(SLIC)superpixel.Because the hierarchical saliency detection model based on the prior location restriction and the local properties combines the detail information of localregions and the constructional information of whole image, the detected results are more precise, and the detected targets are more complete and homogeneous.The detecting precision of detection model is 0.65, the recall ratio is 0.8, Findex is 0.73, and the indexes are higher than the indexes of other saliency detection methods based on superpixel.The CNN pre-training strategy for the specific detection task expends the data base of German traffic sign recognition benchmark(GTSRB), and fully uses the learning skills of CNN to learn the local fine detail features of object, so the recognition precision of CNN rises, and the recognition rate of CNN is 98.85% beyond the rate of SVM with 95.73%.
-
[1] 隽志才, 曹鹏, 吴文静. 基于认知心理学的驾驶员交通标志视认性理论分析[J]. 中国安全科学学报, 2005, 15(8): 8-11. doi: 10.3969/j.issn.1003-3033.2005.08.002JUAN Zhi-cai, CAO Peng, WU Wen-jing. Study on driver traffic signs comprehension based on cognitive psychology[J]. China Safety Science Journal, 2005, 15(8): 8-11. (in Chinese). doi: 10.3969/j.issn.1003-3033.2005.08.002 [2] LIU Han, LIU Ding, LI Qi. Real-time recognition of road traffic sign in moving scene image using genetic algorithm[C]//IEEE. Proceedings of the 4th World Congress on Intelligent Control and Automation. New York: IEEE, 2002: 1027-1030. [3] VÁZQUEZ-REINA A, LAFUENTE-ARROYO S, SIEGMANN P, et al. Traffic sign shape classification based on correlation techniques[C]//WSEAS. Proceedings of the 5th WSEASInternational Conference on Signal Processing, Computational Geometry and Artificial Vision. Stevens Point: WSEAS, 2005: 149-154. [4] LAFUENTE-ARROYO S, SALCEDO-SANZ S, MALDONADO-BASCÓN S, et al. A decision support system for the automatic management of keep-clear signs based on support vector machines and geographic information systems[J]. Expert Systems with Applications, 2010, 37(1): 767-773. doi: 10.1016/j.eswa.2009.05.102 [5] OVERETT G, PETERSSON L. Large scale sign detection using HOG feature variants[C]//IEEE. 2011 IEEEIntelligent Vehicles Symposium(IV). New York: IEEE, 2011: 326-331. [6] WANG Gang-yi, REN Guang-hui, WU Zhi-lu, et al. A hierarchical method for traffic sign classification with support vector machines[C]//IEEE. The 2013International Joint Conference on Neural Networks. New York: IEEE, 2013: 1-6. [7] SALTI S, PETRELLI A, TOMBARI F, et al. A traffic sign detection pipeline based on interest region extraction[C]//IEEE. The 2013International Joint Conference on Neural Networks. New York: IEEE, 2013: 1-7. [8] XIE Yuan, LIU Li-feng, LI Cui-hua, et al. Unifying visual saliency with HOG feature learning for traffic sign detection[C]//IEEE. 2009IEEE Intelligent Vehicles Symposium. New York: IEEE, 2009: 24-29. [9] YAN Qiong, XU Li, SHI Jian-ping, et al. Hierarchical saliency detection[C]//IEEE. 2013IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2013: 1155-1162. [10] WEI Yi-chen, WEN Fang, ZHU Wang-jiang, et al. Geodesic saliency using background priors[C]//FITZGIBBON A, LAZEBNIK S, PERONA P, et al. 12th European Conference on Computer Vision. Berlin: Springer, 2012: 29-42. [11] PERAZZI F, KRÄHENBÜHL P, PRITCH Y, et al. Saliency filters: contrast based filtering for salient region detection[C]//IEEE. 2012 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2012: 733-740. [12] YANG Chuan, ZHANG Li-he, LU Hu-chuan, et al. Saliency detection via graph-based manifold ranking[C]//IEEE. 2013IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2013: 3166-3173. [13] YUAN Xue, GUO Jia-qi, HAO Xiao-li, et al. Traffic sign detection via graph-based ranking and segmentation algorithms[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2015, 45(12): 1509-1521. doi: 10.1109/TSMC.2015.2427771 [14] SERMANET P, CHINTALA S, LECUN Y. Convolutional neural networks applied to house numbers digit classification[C]//IEEE. 21st International Conference on Pattern Recognition. New York: IEEE, 2012: 3288-3291. [15] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//PEREIRA F, BURGES C J C, BOTTOU L, et al. Advances in Neural Information Processing Systems 25. South Lake Tahoe: NIPS Foundation, 2012: 1097-1105. [16] SERMANET P, LECUN Y. Traffic sign recognition with multi-scale convolutional networks[C]//IEEE. The 2011International Joint Conference on Neural Networks. New York: IEEE, 2011: 2809-2813. [17] WU Yi-hui, LIU Yu-long, LI Jian-min, et al. Traffic sign detection based on convolutional neural networks[C]//IEEE. The 2013 International Joint Conference on Neural Networks. New York: IEEE, 2013: 1-7. [18] JIA Yang-qing, SHELHAMER E, DONAHUE J, et al. Caffe: convolutional architecture for fast feature embedding[C]//ACM. Proceedings of the 22nd ACM International Conference on Multimedia. New York: ACM, 2014: 675-678. [19] SZEGEDY C, LIU Wei, JIA Yang-qing, et al. Going deeper with convolutions[C]//IEEE. 2015 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2015: 1-9. [20] ZEILER M D, FERGUS R. Visualizing and understanding convolutional networks[C]//FLEET D, PAJDLA T, SCHIELE B, et al. 13th European Conference on Computer Vision. Berlin: Springer, 2014: 818-833. [21] BESAG J. Spatial interaction and the statisticalanalysis of lattice systems[J]. Journal of the Royal Statistical Society. Series B: Methodological, 1974, 36(2): 192-236. [22] LAFFERTY J, M CALLUM A, PEREIRA F. Conditional random fields: probabilistic models for segmenting and labeling sequence data[C]//ACM. Proceedings of the 18th International Conference on Machine Learning. New York: ACM, 2001: 282-289. [23] YEDIDIA J S, FREEMAN W T, WEISS Y. Generalized belief propagation[C]//LEEN T K, DIETTERICH T G, TRESP V. Advances in Neural Information Processing Systems 13. Denver: NIPS Foundation, 2000: 689-695. [24] HOUBEN S, STALLKAMP J, SALMEN J, et al. Detection of traffic signs in real-world images: the German traffic sign detection benchmark[C]//IEEE. The 2013 International Joint Conference on Neural Networks. New York: IEEE, 2013: 1-8. [25] STSLLKAMP J, SCHLIPSING M, SALMEN J, et al. The German traffic sign recognition benchmark: a multi-class classification competition[C]//IEEE. The 2011International Joint Conference on Neural Networks. New York: IEEE, 2011: 1453-1460. [26] WEN Cheng-lu, LI J, LUO Huan, et al. Spatial-related traffic sign inspection for inventory purposes using mobile laser scanning data[J]. IEEE Transactions on Intelligent Transactions Systems, 2016, 17(1): 27-37. [27] ACHANTA R, SHAJI A, SMITH K, et al. SLIC superpixels[R]. Lausanne: École Polytechnique Féderale de Lausanne, 2010. [28] MARTIN D R, FOWLKES C C, MALIK J. Learning to detect natural image boundaries using local brightness, color, and texture cues[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2004, 26(5): 530-549. [29] CHENG Ming-ming, WARRELL J, LIN Wen-yan, et al. Efficient salient region detection with soft image abstraction[C]//IEEE. 2013 IEEE International Conference on Computer Vision. New York: IEEE, 2013: 1529-1536. [30] TONG Na, LU Hu-chuan, RUAN Xiang, et al. Salient object detection via bootstrap learning[C]//IEEE. 2015 IEEEConference on Computer Vision and Pattern Recognition. New York: IEEE, 2015: 1884-1892. [31] QIN Yao, LU Hu-chuan, XU Yi-qun, et al. Saliency detection via cellular automata[C]//IEEE. 2015IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2015: 110-119. -
下载:
点击查看大图
图(19)
计量
- 文章访问数: 2322
- HTML全文浏览量: 135
- PDF下载量: 2676
- 被引次数: 0
