基于可见-近红外光谱的路面状况识别的研究

熊显名; 张乾坤; 秦祖军

基于可见-近红外光谱的路面状况识别的研究

桂林电子科技大学，光电信息处理重点实验室，广西桂林 541004

基金项目:

国家自然科学基金 61665001

详细信息

作者简介:
熊显名（1964-），男，学士，研究员，硕士生导师，研究方向为光电信息处理、光纤传感器。E-mail：5311128@qq.com

中图分类号: TN219
计量
- 文章访问数: 991
- HTML全文浏览量: 180
- PDF下载量: 89
- 被引次数: 0
出版历程
- 收稿日期: 2019-07-19
- 修回日期: 2019-10-08
- 刊出日期: 2021-02-20

Research on Highway State Detection Based on Visible-Near-Infrared Spectrum

Key Laboratory of Photoelectric Information Processing, Guilin University of Electronic Science and Technology, Guilin 541004, China

摘要

摘要: 光谱技术在公路状态识别（是否结冰、积水或积雪）方面有着积极的应用前景，但太阳光作为光源识别公路状态的研究较少。分别采用阳光和卤钨灯作为白天和夜间的实验光源，通过微型光谱仪数据分别得到冰、水、雪和公路本底的可见-近红外波段的光谱曲线。白天时，结冰和积水状态在不同光照情况下会出现“异物类谱”现象，根据阳光光照特性，本文提出将“环境变量”作为特征值的解决方法，并基于光谱曲线及归一化后的“环境变量”特征值，将光谱数据组合成新的数据波形，基于Dropout与Adam优化器的神经网络模型对数据进行训练和识别，最终识别率为99.375%。夜间，由于各类样本光谱区域差异明显，采用“组合-阈值”法识别。实验证明通过两种光源结合的识别方法，能够有效识别路面状态。
- 公路状态识别 /
- 可见-近红外光谱 /
- 神经网络 /
- 环境变量 /
- 组合-阈值法
Abstract: Spectral technology is a promising prospect for highway state detection(whether frozen, water accumulated, or snow accumulated). However, there is little research on using sunlight as a light source to identify highway states. Sunlight and halogen tungsten lamps were used as experimental light sources in the day and night. Spectral curves of the visible-near-infrared bands of ice, water, snow, and highway backgrounds were obtained using a micro-spectrometer. During the day, the state of icing and stagnant water resulted in a phenomenon known as "Different substances with similar spectra" under different illumination conditions. Then, based on the characteristics of sunlight illumination, the solution of "environmental variables" as eigen values was proposed. The curve of the spectrum and the normalized "environmental variables" were combined into a new data waveform, and a neural network model based on Dropout and an Adam optimizer was established for training and recognition. The final recognition rate was 99.375%. At night, due to the evident differences in the spectra of various samples, the spectral curves of each sample were identified using the "combination-threshold" method. Experiments proved that the method of combining two light sources can effectively identify the road surface state.
- highway state detection /
- visible-near infrared spectrum /
- neural network /
- the environment variables /
- combination-threshold

HTML全文

图 1 BP神经网络结构示意图

Figure 1. Structure of BP neural network

下载: 全尺寸图片幻灯片

图 2 结冰和积水A/D输出值的“异物类谱”图

Figure 2. The A/D output values of ice and water have similar spectral curves

下载: 全尺寸图片幻灯片

图 3 背景物质光谱曲线图

Figure 3. Background material spectrum curves

下载: 全尺寸图片幻灯片

图 4 数据采集结构示意图

Figure 4. Data acquisition schematic

下载: 全尺寸图片幻灯片

图 5 单个微型光谱仪数据采集结构示意图

Figure 5. Data acquisition schematic of a single micro-spectrometer

下载: 全尺寸图片幻灯片

图 6 白天时，公路本底表面各状态整合计算光谱图

Figure 6. Spectra of various states of road background surface during the daytime

下载: 全尺寸图片幻灯片

图 7 夜间公路本底表面各状态整合计算光谱图

Figure 7. Absorbance spectra of road background surface at night

下载: 全尺寸图片幻灯片

图 8 白天和夜晚时公路本底表面干燥状态整合计算光谱图

Figure 8. Dry absorbance spectra of road background surface during day and night

下载: 全尺寸图片幻灯片

图 9 结冰和积水“异物类谱”图

Figure 9. Ice and water may have similar absorbance signatures

下载: 全尺寸图片幻灯片

图 10 结冰和积水的环境特征值图

Figure 10. Background material spectrum curves

下载: 全尺寸图片幻灯片

图 11 组合数值比较图

Figure 11. Combined numerical comparison

下载: 全尺寸图片幻灯片

图 12 损失值变化曲线

Figure 12. Loss value variation curve

下载: 全尺寸图片幻灯片

图 13 准确率变化曲线

Figure 13. Accuracy variation curves

下载: 全尺寸图片幻灯片

表 1 样本分析统计表

Table 1. Sample analysis statistics

	Ice	Snow	Water	Dry
Average value	7.043	-5.456	8.565	-0.109
Standard deviation	0.159	0.311	0.140	0.08
Maximum deviation	0.243	0.417	0.447	0.105

下载: 导出CSV

参考文献(11)

[1]	欧彦, 浦翔, 周旭驰, 等. 路面结冰检测技术研究进展[J]. 公路, 2013(4): 191-195. OU Yan, PU Xiang, ZHOU Xunchi, et al. Review on icing detection techniques of pavement[J]. High Way, 2013(4): 191-195.
[2]	童魁. 路面冰水检测系统技术研究[D]. 南京: 东南大学, 2011. TONG Kui. Research On System Technology of Road Ice And Water Detection[D]. Nanjing: Southeast University, 2011.
[3]	张镇, 葛俊锋, 叶林, 等. 基于神经网络的主动式红外结冰探测[J]. 华中科技大学学报: 自然科学版, 2010, 38(6): 1-3. https://www.cnki.com.cn/Article/CJFDTOTAL-HZLG201006002.htm ZHANG Zhen, GE Junfeng, YE Lin, et al. Active infrared icing detection using neural networks[J]. Journal of Huazhong University of Science and Technology: Natural Science Edition, 2010, 38(6): 1-3. https://www.cnki.com.cn/Article/CJFDTOTAL-HZLG201006002.htm
[4]	Colace L, Santoni F, Assanto G. A near-infrared optoelectronic approach to detection of road conditions[J]. Optics and Lasers in Engineering, 2013, 51(5): 633-636. doi: 10.1016/j.optlaseng.2013.01.003
[5]	梁曹佳, 叶林, 葛俊锋. 非接触式路面状态检测技术研究进展[J]. 传感器与微系统, 2019, 38(2): 1-4. https://www.cnki.com.cn/Article/CJFDTOTAL-CGQJ201902001.htm LIANG Caojia, YE Lin, GE Junfeng. Research progress of non-contact road surface condition detection technology[J]. Transducer and Microsystem Technologies, 2019, 38(2): 1-4. https://www.cnki.com.cn/Article/CJFDTOTAL-CGQJ201902001.htm
[6]	Johan Casselgren, Mikael Sjodahl, James LeBlanc. Angular spectral response from covered asphalt[J]. Applied Optics, 2007, 46(20): 4277-4288. doi: 10.1364/AO.46.004277
[7]	王琮琪. 非接触式路面状况检测系统的研究[D]. 杭州: 浙江大学, 2014. WANG Zongqi. No-contact Measuring System of the Condition of Road[D]. Hangzhou: Zhejiang University, 2014.
[8]	高斌, 赵鹏飞, 卢昱欣, 等. 基于BP神经网络的血液荧光光谱识别分类研究[J]. 光谱学与光谱分析, 2018, 38(4): 3136-3143. https://www.cnki.com.cn/Article/CJFDTOTAL-GUAN201810029.htm GAO Bin, ZHAO Pengfei, LU Yuxin, et al. Study on recognition and classification of blood fluorescence spectrum with BP neural network[J]. Spectroscopy and Spectral Analysis, 2018, 38(4): 3136-3143. https://www.cnki.com.cn/Article/CJFDTOTAL-GUAN201810029.htm
[9]	刘厚林, 吴贤芳, 王勇, 等. 基于BP神经网络的离心泵关死点功率预测[J]. 农业工程学报, 2012(11): 45-49. https://www.cnki.com.cn/Article/CJFDTOTAL-NYGU201211009.htm LIU Houlin, WU Xianfang, WANG Yong, et al. Power prediction for centrifugal pumps at shut off condition based on BP neural network[J]. Transactions of the Chinese Society of Agricultural Engineering, 2012(11): 45-49. https://www.cnki.com.cn/Article/CJFDTOTAL-NYGU201211009.htm
[10]	王威, 李青, 孙叶青, 等. 基于卷积神经网络的红外热成像罐车内壁裂纹识别[J]. 红外技术, 2018, 40(12): 1198-1205. http://hwjs.nvir.cn/article/id/hwjs201812014 WANG Wei, LI Qing, SUN Yeqing, et al. Inner crack identification on car tanks using thermal imaging based on convolutional neural network[J]. Infrared Technology, 2018, 40(12): 1198-1205. http://hwjs.nvir.cn/article/id/hwjs201812014
[11]	杨观赐, 杨静, 李少波, 等. 基于Dropout与ADAM优化器的改进CNN算法[J]. 华中科技大学学报: 自然科学版, 2018, 46(7): 122-127. https://www.cnki.com.cn/Article/CJFDTOTAL-HZLG201807023.htm YANG Guanci, YANG Jing, LI Shaobo, et al. Modified CNN algorithm based on dropout and ADAM optimizer[J]. Journal of Huazhong University of Science and Technology: Natural Science Edition, 2018, 46(7): 122-127. https://www.cnki.com.cn/Article/CJFDTOTAL-HZLG201807023.htm