基于Gabor滤波和稀疏表示的信号灯类型识别
|
摘要
汽车辅助驾驶中车辆前方交通信号灯的检测与识别具有重要意义。为解决图像中微小面积与边缘模糊的交通信号灯检测与识别准确率低的问题,文中提出一种信号灯检测与信号灯类型识别方法。采用亮度分割和几何形态滤波对交通信号灯进行定位,结合信号灯与其背板的相对位置和RGB与HSV颜色空间的判别结果判定信号灯颜色,最后对信号灯类型进行识别。对信号灯区域进行Gabor滤波,采用K均值奇异值分解算法进行字典学习,利用正交匹配追踪算法求解测试样本的稀疏系数,根据重构误差实现交通灯的类型判别。实验结果表明,信号灯检测准确率达到97.7%,圆形和4种箭头形信号灯的类型识别率达到98.75%。
Traffic light detection and recognition are important in advance driver assistance system. To solve the problem of low accuracy in detection and classification of traffic light with small area and edge blur in image,a traffic light detection and recognition method was proposed. Firstly,traffic light in the image was located using intensity segmentation and morphological filtering. Then,combined with the recognition results of RGB and HSV color space and the relative position of the lamp and lamp board,the color of the lamp was determined. At last,traffic light recognition was realized. The light region was filtered by Gabor wavelets,and K-SVD algorithm was used to learn the dictionary. OMP algorithm was used to extract the features,and traffic lights were classified according to the reconstruction error. Experimental results showed that the detection accuracy of traffic light was 97. 7%,and the recognition rate of circular and four types of arrow shaped lamps was 98. 75%.
Traffic light detection and recognition are important in advance driver assistance system. To solve the problem of low accuracy in detection and classification of traffic light with small area and edge blur in image,a traffic light detection and recognition method was proposed. Firstly,traffic light in the image was located using intensity segmentation and morphological filtering. Then,combined with the recognition results of RGB and HSV color space and the relative position of the lamp and lamp board,the color of the lamp was determined. At last,traffic light recognition was realized. The light region was filtered by Gabor wavelets,and K-SVD algorithm was used to learn the dictionary. OMP algorithm was used to extract the features,and traffic lights were classified according to the reconstruction error. Experimental results showed that the detection accuracy of traffic light was 97. 7%,and the recognition rate of circular and four types of arrow shaped lamps was 98. 75%.
引文
[1]谷明琴,蔡自兴,黄振威.城市环境中箭头型交通信号灯的实时识别算法[J].中南大学学报:自然科学版,2013,44(4):1404-1408.
[2]佀君淑,朱文兴,沙永贺.复杂背景下的交通信号灯综合识别方法[J].山东大学学报:工学版,2014,44(2):64-68.
[3]Shi Zhenwei,Zou Zhengxia,Zhang Changshui.Real-time traffic light detection with adaptive background suppression filter[J].IEEE Transactions on Intelligent Transportation Systems,2016,17(3):690-700.
[4]Sooksatra S,T Kondo.Red traffic light detection using fast radial symmetry transform[C].PA,USA:Proceedings of IEEE International Conference on Electrical Engineering/Electronics,Computer,Telecommunications and Information Technology,2014.
[5]De Charette R,Nashashibi F.Real time visual traffic lights recognition based on spot light detection and adaptive traffic lights templates[C].Paris:Intelligent Vehicles Symposium IEEE,2009.
[6]Shen Yehu,Umit Ozguner,Keith Redmill,et al.A robust video based traffic light detection algorithm for intelligent vehicles[C].Olrando:IEEE Intelligent Vehicles Symposium,2009.
[7]谷明琴,蔡自兴,李仪.应用圆形度和颜色直方图的交通信号灯识别[J].计算机工程与设计,2012,33(1):243-247.
[8]缪小冬,李舜酩,沈峘,等.复杂环境中交通标志的实时识别方法[J].江苏大学学报:自然科学版,2013,34(5):514-518.
[9]田谨,应捷,朱丹丹.一种新的箭头形交通信号灯识别方法[J].电子科技,2015,28(11):150-153.
[10]沈丽,韩彦芳.基于稀疏表示的超分辨率图像重建[J].电子科技,2015,28(9):144-147.
[11]曹蒙蒙,李新叶,范月坤.基于判别低秩矩阵恢复和稀疏表示的车标识别[J].电子科技,2015,28(4):57-60.
[12]Fan Zizhu,Ni Ming,Zhu Qi.Weighted sparse representation for face recognition[J].Neuro Computing,2015(151):304-309.
[13]Bhateja C,Ashok K,Sharma Shikhar,et al.Iris recognition based on sparse representation and K-nearest subspace with genetic algorithm[J].Pattern Recognition Letters,2016,73(1):13-18.
[14]Chiang Cheng-Chin,Ho Ming-Che,Liao Hong-sheng,et al.Detecting and recognizing traffic lights by genetic similar ellipse detection and spatial texture layouts[J].International Journal of Innovative Computing,Information and Control,2011,7(12):6919-6934.
[15]George Siogkas,Evangelos Skodras,Evangelos Dermatas.Traffic lights detection in adverse conditions using color,symmetry and spatiotemporal information[C].Russia:International Conference on Computer Vision Theory and Applications,2012.
[2]佀君淑,朱文兴,沙永贺.复杂背景下的交通信号灯综合识别方法[J].山东大学学报:工学版,2014,44(2):64-68.
[3]Shi Zhenwei,Zou Zhengxia,Zhang Changshui.Real-time traffic light detection with adaptive background suppression filter[J].IEEE Transactions on Intelligent Transportation Systems,2016,17(3):690-700.
[4]Sooksatra S,T Kondo.Red traffic light detection using fast radial symmetry transform[C].PA,USA:Proceedings of IEEE International Conference on Electrical Engineering/Electronics,Computer,Telecommunications and Information Technology,2014.
[5]De Charette R,Nashashibi F.Real time visual traffic lights recognition based on spot light detection and adaptive traffic lights templates[C].Paris:Intelligent Vehicles Symposium IEEE,2009.
[6]Shen Yehu,Umit Ozguner,Keith Redmill,et al.A robust video based traffic light detection algorithm for intelligent vehicles[C].Olrando:IEEE Intelligent Vehicles Symposium,2009.
[7]谷明琴,蔡自兴,李仪.应用圆形度和颜色直方图的交通信号灯识别[J].计算机工程与设计,2012,33(1):243-247.
[8]缪小冬,李舜酩,沈峘,等.复杂环境中交通标志的实时识别方法[J].江苏大学学报:自然科学版,2013,34(5):514-518.
[9]田谨,应捷,朱丹丹.一种新的箭头形交通信号灯识别方法[J].电子科技,2015,28(11):150-153.
[10]沈丽,韩彦芳.基于稀疏表示的超分辨率图像重建[J].电子科技,2015,28(9):144-147.
[11]曹蒙蒙,李新叶,范月坤.基于判别低秩矩阵恢复和稀疏表示的车标识别[J].电子科技,2015,28(4):57-60.
[12]Fan Zizhu,Ni Ming,Zhu Qi.Weighted sparse representation for face recognition[J].Neuro Computing,2015(151):304-309.
[13]Bhateja C,Ashok K,Sharma Shikhar,et al.Iris recognition based on sparse representation and K-nearest subspace with genetic algorithm[J].Pattern Recognition Letters,2016,73(1):13-18.
[14]Chiang Cheng-Chin,Ho Ming-Che,Liao Hong-sheng,et al.Detecting and recognizing traffic lights by genetic similar ellipse detection and spatial texture layouts[J].International Journal of Innovative Computing,Information and Control,2011,7(12):6919-6934.
[15]George Siogkas,Evangelos Skodras,Evangelos Dermatas.Traffic lights detection in adverse conditions using color,symmetry and spatiotemporal information[C].Russia:International Conference on Computer Vision Theory and Applications,2012.