基于视频区域动态特征的林火烟雾检测技术研究

刘长春; 刘鹏举; 季烨云

doi:10.12171/j.1000-1522.20200049

基于视频区域动态特征的林火烟雾检测技术研究

刘长春^{1, 2,},
刘鹏举^{1, 2, ,},
季烨云^{1, 2}

1.
中国林业科学研究院资源信息研究所，北京 100091
2.
国家林业和草原局林业遥感与信息技术重点实验室，北京 100091

基金项目: 中国林业科学研究院基本科研业务费专项（CAFYBB2017ZC001），“948”国家林业局引进项目（2014-4-01）

详细信息

作者简介:
刘长春。主要研究方向：地理信息系统技术与应用。Email：lcc175@126.com　地址：100091 北京市海淀区香山路东小府1号中国林业科学研究院资源信息研究所

责任作者:
刘鹏举，博士，副研究员。主要研究方向：林业GIS应用与开发。Email：liupeng@caf.ac.cn　地址：同上

中图分类号: S762.3+2
计量
- 文章访问数: 2310
- HTML全文浏览量: 875
- PDF下载量: 127
出版历程
- 收稿日期: 2020-02-22
- 修回日期: 2020-05-24
- 网络出版日期: 2020-07-26
- 发布日期: 2021-02-04

Research on forest fire smoke detection technology based on video region dynamic features

Liu Changchun^{1, 2,},
Liu Pengju^{1, 2, ,},
Ji Yeyun^{1, 2}

1.
Research Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing 100091, China
2.
Key Laboratory of Forestry Remote Sensing and Information System, National Forestry and Grassland Administration, Beijing 100091, China

摘要

摘要:
目的视频监控越来越多地应用到森林火灾烟雾的早期检测中。现有的视频林火烟雾检测方法大多是基于像素提取烟雾特征进行分析检测，烟雾发生早期或烟雾距离摄像头较远时，在视频图像上烟雾仅呈现较小区域，且烟雾的扩散具有无规则性，背景环境复杂多变，导致基于像素的特征不明显，因此使基于像素的烟雾自动化检测难度增大。本文根据可见光视频图像处理原理，提出一种基于局部区域图像动态特征的林火视频烟雾检测方法，以提高林火视频烟雾检测准确度和灵敏度。
方法以视频图像为研究对象，每秒取一帧生成图像序列，对图像序列进行多层次不同尺度分区；利用图像信噪比原理，计算分区后的连续图像序列的信噪比；根据背景图像信噪比得到自适应阈值，确定待检测图像序列发生亮度变化的图像块，即为疑似烟雾块；提取疑似烟雾块的LBP纹理特征，采用支持向量机区分出烟雾区域。
结果利用HSV颜色空间的亮度分量，可以有效提取烟雾区域。选择有林火烟雾的视频，对提出的烟雾变化检测方法进行验证，分析结果表明该方法能确定烟雾发生所在的图像块，且能排除部分非烟雾干扰因素。
结论本文提出了基于局部区域亮度特征和LBP纹理特征的视频林火烟雾检测技术，能准确定位烟雾发生区域，排除部分干扰因素，检测识别率平均达到92%以上，有助于实时林火烟雾自动检测，提高林火烟雾检测率，具有很强的实用性。
- 林火烟雾检测 /
- 视频图像 /
- 分区 /
- 信噪比 /
- LBP纹理特征
Abstract:
Objective Video surveillance is increasingly applied to the early detection of forest fire smoke. The existing video forest fire smoke detection methods are mostly based on pixel extraction of smoke characteristics for analysis and detection, but when the smoke is early or the smoke is far from the camera, the smoke only appears in a small area on the video image. Moreover, the diffusion of smoke is irregular, and the background environment is complex and changeable, resulting in insignificant pixel-based features, which makes it more difficult to automatically detect pixel-based smoke. Based on the principle of visible light video image processing, this paper proposes a forest fire video smoke detection method based on local area image dynamic characteristics to improve the accuracy and sensitivity of forest fire video smoke detection.
Method The video images were selected as the research object. One frame per second was taken to generate an image sequence, and the image sequences were divided into multiple levels and different scales; using the principle of image signal-to-noise ratio, we calculated the signal-to-noise ratio of continuous image sequences after blocking; the adaptive threshold was obtained according to the signal-to-noise ratio of the background image, and the image block whose brightness changes in the image sequence to be detected was determined to be the suspected smoke block; the LBP texture feature of the suspected smoke block was extracted, and the support vector machine was used to distinguish the smoke area.
Result Using the value component of the HSV color space, smoke areas can be effectively extracted. The videos with forest fire smoke were selected to verify the proposed smoke change detection method. The analysis results showed that the method can determine the image block where the smoke occurred and excluded some non-smoke interference factors.
Conclusion This paper proposes a video forest fire smoke detection technology based on brightness characteristics and LBP texture features of local area, which can accurately locate the smoke occurrence area and exclude some interference factors. The average detection recognition rate reaches more than 92%, which is helpful for real-time forest fire smoke automatic detection and improving the detection rate of forest fire smoke. It has a strong practicality.
- forest fire smoke detection /
- video image /
- block /
- signal-to-noise ratio /
- LBP texture feature

HTML全文

图 1 分块后编码

Figure 1. Block encoding

下载: 全尺寸图片幻灯片

图 2 算法流程图

Figure 2. Algorithm flowchart

下载: 全尺寸图片幻灯片

图 3 Video4烟雾发生区域连续图像

Figure 3. Continuous image of smoke occurrence area in video4

下载: 全尺寸图片幻灯片

图 4 烟雾HSV颜色通道SNR变化趋势

m表示参考帧取背景帧；n表示参考帧取前一帧。m indicates the reference frame taking the background frame; n indicates the reference frame taking the previous frame.

Figure 4. SNR changing trend of smoke HSV color channel

下载: 全尺寸图片幻灯片

图 5 SNR变化趋势

SNR_bg表示前一时间窗口内SNR的值。SNR_bg indicates the value of SNR in the previous time window.

Figure 5. Changing trend of SNR

下载: 全尺寸图片幻灯片

图 6 非烟雾SNR_m和SNR_n变化趋势

Figure 6. Changing trend of non-smoke SNR_m and SNR_n

下载: 全尺寸图片幻灯片

图 7 不同分块方法得到SNR变化趋势

Figure 7. Changing trend of SNR by diffirent block methods

下载: 全尺寸图片幻灯片

图 8 检测结果

Figure 8. Detection results

下载: 全尺寸图片幻灯片

表 1 视频相关信息

Table 1 Video related information

编号 No.	类别 Classification	视频名称 Video name	视频场景描述 Video scene description	帧速率/(帧·s⁻¹) Frame rate/ (frame·s⁻¹)	截取后总帧数 Total number of frames after clip
视频1 Video1	烟雾 Smoke	Pelco_Colakli	远距离，有风，烟雾扩散缓慢 Long distance, windy, smoke spread slowly	7	122
视频2 Video2	烟雾 Smoke	20070817 Aksehir_Duman_Test5_ en_Iyisi_near_487_frame	远距离，有风，烟雾扩散缓慢 Long distance, windy, smoke spread slowly	7	484
视频3 Video3	烟雾 Smoke	Smoke_Manavgat_Raw	远距离，烟雾较浓，扩散缓慢 Long distance, thick smoke, slow diffusion	25	242
视频4 Video4	烟雾 Smoke	20090409 ManavgatTEst	远距离，有风，有车辆驶过，烟雾扩散缓慢 Long distance, windy, vehicles passing by, smoke spread slowly	9	180
视频5 Video5	烟雾 Smoke	森林保护站拍摄视频1 Video1 taken at the forest protection station	近距离，有烟雾，卡车施工作业 Close distance, smoke, truck construction operation	25	238
视频6 Video6	非烟雾 Non-smoke	森林保护站拍摄视频2 Video2 taken at the forest protection station	近距离，有云 Close distance, cloudy	30	220
视频7 Video7	非烟雾 Non-smoke	森林保护站拍摄视频3 Video3 taken at the forest protection station	近距离，有行人经过 Close distance, pedestrians passing by	24	192
视频8 Video8	非烟雾 Non-smoke	森林保护站拍摄视频4 Video4 taken at the forest protection station	近距离，有车辆驶过 Close distance, vehicles passing by	25	168

下载: 导出CSV

表 2 2 × 2分块方式

Table 2 2 × 2 block pattern

编码 No.	块0 Block 0	块1 Block 1	块2 Block 2	块3 Block 3
待检测区域信噪比的方差SNR_var SNR_var of the area to be detected	5.477 5	0.041 3	0.054 2	0.053 1

下载: 导出CSV

表 4 8 × 8分块方式

Table 4 8 × 8 block pattern

编码 No.	块030 Block 030	块031 Block 031	块032 Block 032	块033 Block 033
待检测区域信噪比的方差SNR_var SNR_var of the area to be detected	21.453 7	16.541 4	25.686 1	40.761 4

下载: 导出CSV

表 3 4 × 4分块方式

Table 3 4 × 4 block pattern

编码 No.	块00 Block 00	块01 Block 01	块02 Block 02	块03 Block 03
待检测区域信噪比的方差SNR_var SNR_var of the area to be detected	0.083 5	1.738 1	0.677 1	14.888 5

下载: 导出CSV

表 5 测试结果

Table 5 Test results %

烟雾视频数据 Smoke video data	精确率 Precision (P)	F₁
视频1 Video1	90	95
视频2 Video2	93	96
视频3 Video3	90	95
视频4 Video4	95	98
视频5 Video5	93	96
注：F₁为精确率和召回率的调和平均数。Note: F₁ is harmonic mean of precision and recall.

下载: 导出CSV

参考文献(27)

[1]	史劲亭, 袁非牛, 夏雪. 视频烟雾检测研究进展[J]. 中国图象图形学报, 2018, 23(3):303−322. doi: 10.11834/jig.170439. Shi J T, Yuan F N, Xia X. Video smoke detection: a literature survey[J]. Journal of Image and Graphics, 2018, 23(3): 303−322. doi: 10.11834/jig.170439.
[2]	Çetin A E, Dimitropoulos K, Gouverneur B, et al. Video fire detection: review[J]. Digital Signal Processing, 2013, 23(6): 1827−1843. doi: 10.1016/j.dsp.2013.07.003.
[3]	张斌, 魏维, 何冰倩. 基于多特征融合的早期野火烟雾检测[J]. 成都信息工程大学学报, 2018, 33(4):408−412. Zhang B, Wei W, He B Q. An early wildfire smoke detection method based on multi-features fusion[J]. Journal of Chengdu University of Information Technology, 2018, 33(4): 408−412.
[4]	林宏, 刘志刚, 赵同林, 等. 基于视频的林火烟雾识别算法研究[J]. 安全与环境学报, 2013, 13(2):210−214. doi: 10.3969/j.issn.1009-6094.2013.02.045 Lin H, Liu Z G, Zhao T L, et al. Improved algorithm for smoking identification of the forest fire based on the video survey[J]. Journal of Safety and Environment, 2013, 13(2): 210−214. doi: 10.3969/j.issn.1009-6094.2013.02.045
[5]	周忠, 赵亚琴, 唐于维一, 等. 基于时空特征的林火视频烟雾区域提取[J]. 中国农机化学报, 2016, 37(2):196−199. Zhou Z, Zhao Y Q, Tang-Yu W Y, et al. Segmentation of forest fire video smoke region based on the temporal-spatial features[J]. Journal of Chinese Agricultural Mechanization, 2016, 37(2): 196−199.
[6]	Liu J L, Zhao H W. Research of smoke detection on visual saliency method[J]. Journal of Multimedia, 2014, 9(6): 781−788.
[7]	叶秋冬, 王越. 基于视频的烟雾检测预警系统在森林防火系统中的应用[J]. 计算机应用与软件, 2012, 29(1):270−272, 293. doi: 10.3969/j.issn.1000-386X.2012.01.076 Ye Q D, Wang Y. Application of video-based smoke detection and early warning system to forest fire prevention system[J]. Computer Applications and Software, 2012, 29(1): 270−272, 293. doi: 10.3969/j.issn.1000-386X.2012.01.076
[8]	刘恺, 刘湘, 常丽萍, 等. 基于YUV颜色空间和多特征融合的视频烟雾检测[J]. 传感技术学报, 2019, 32(2):237−243. Liu K, Liu X, Chang L P, et al. Video smoke detection based on YUV color space and multiple feature fusion[J]. Chinese Journal of Sensors and Actuators, 2019, 32(2): 237−243.
[9]	Tian H D, Li W Q, Ogunbona P O, et al. Detection and separation of smoke from single image frames[J]. IEEE Transactions on Image Processing, 2018, 27(3): 1164−1177. doi: 10.1109/TIP.2017.2771499
[10]	Ojala T, Pietikäinen M, Harwood D. A comparative study of texture measures with classification based on featured distributions[J]. Pattern Recognition, 1996, 29(1): 51−59. doi: 10.1016/0031-3203(95)00067-4
[11]	Yuan F N. Video-based smoke detection with histogram sequence of LBP and LBPV pyramids[J]. Fire Safety Journal, 2011, 46(3): 132−139. doi: 10.1016/j.firesaf.2011.01.001
[12]	杨秋霞, 罗传文. 基于LBP和稀疏表示的林火烟雾图像识别研究[J]. 安徽农业科学, 2014, 42(34):12342−12346. doi: 10.3969/j.issn.0517-6611.2014.34.116 Yang Q X, Luo C W. Forest fire smoke recognition based on local binary patterns and sparse representation[J]. Journal of Anhui Agricultural Sciences, 2014, 42(34): 12342−12346. doi: 10.3969/j.issn.0517-6611.2014.34.116
[13]	李红娣, 袁非牛. 采用金字塔纹理和边缘特征的图像烟雾检测[J]. 中国图象图形学报, 2015, 20(6):772−780. doi: 10.11834/jig.20150606 Li H D, Yuan F N. Image based smoke detection using pyramid texture and edge features[J]. Journal of Image and Graphics, 2015, 20(6): 772−780. doi: 10.11834/jig.20150606
[14]	兰久强, 刘金清, 刘引, 等. 基于颜色和纹理特征的林火烟雾识别[J]. 计算机系统应用, 2016, 25(3):101−106. Lan J Q, Liu J Q, Liu Y, et al. Forest fire smoke recognition based on color and texture features[J]. Computer Systems & Applications, 2016, 25(3): 101−106.
[15]	Ojala T, Pietikäinen M, Mäenpää T. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(7): 971−987. doi: 10.1109/TPAMI.2002.1017623
[16]	Yuan F N, Xia X, Shi J T. Holistic learning-based high-order feature descriptor for smoke recognition[J]. International Journal of Wavelets, Multiresolution and Information Processing, 2019, 17(2): 1940005. doi: 10.1142/S0219691319400058
[17]	Prema C E, Vinsley S S, Suresh S. Multi feature analysis of smoke in YUV color space for early forest fire detection[J]. Fire Technology, 2016, 52(5): 1319−1342. doi: 10.1007/s10694-016-0580-8
[18]	Dimitropoulos K, Barmpoutis P, Grammalidis N. Higher order linear dynamical systems for smoke detection in video surveillance applications[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(5): 1143−1154. doi: 10.1109/TCSVT.2016.2527340
[19]	Ko B C, Kwak J Y, Nam J Y. Wildfire smoke detection using temporospatial features and random forest classifiers[J]. Optical Engineering, 2012, 51(1): 017208. doi: 10.1117/1.OE.51.1.017208
[20]	Park J O, Ko B C, Nam J Y, et al. Wildfire smoke detection using spatiotemporal bag-of-features of smoke[C]//2013 IEEE Workshop on Applications of Computer Vision. Tampa: IEEE, 2013: 200−205.
[21]	张建明, 黄继风. 基于视频的实时烟雾检测算法[J]. 计算机应用与软件, 2016, 33(12):248−252. doi: 10.3969/j.issn.1000-386x.2016.12.059 Zhang J M, Huang J F. Real-time smoke detection algorithm based on video[J]. Computer Applications and Software, 2016, 33(12): 248−252. doi: 10.3969/j.issn.1000-386x.2016.12.059
[22]	罗一涵, 刘妍妍, 陈科. 探测信噪比计算方法及原理综述[J]. 电声技术, 2016, 40(6):37−43, 57. Luo Y H, Liu Y Y, Chen K. Computing methods and principles of detective SNR[J]. Audio Engineering, 2016, 40(6): 37−43, 57.
[23]	贺国旗, 陈向奎, 韩泉叶, 等. 一种自动提高图像信噪比的方法[J]. 计算机技术与发展, 2015, 25(12):60−63. He G Q, Chen X K, Han Q Y, et al. A method of automatically improving SNR of image[J]. Computer Technology and Development, 2015, 25(12): 60−63.
[24]	Bilkent Signal Procesing Group. Computer vision based fire detection software[EB/OL]. [2019−06−20]. http://signal.ee.bilkent.edu.tr/VisiFire/.
[25]	郑怀兵, 翟济云. 基于视频分析的森林火灾烟雾检测方法[J]. 南京理工大学学报(自然科学版), 2015, 39(6):686−691, 710. Zheng H B, Zhai J Y. Forest fire smoke detection based on video analysis[J]. Journal of Nanjing University of Science and Technology, 2015, 39(6): 686−691, 710.
[26]	何大超, 娄小平, 唐辉. 基于动态特性的实时视频烟雾检测[J]. 计算机应用与软件, 2014, 31(2):201−204. doi: 10.3969/j.issn.1000-386x.2014.02.054 He D C, Lou X P, Tang H. Dynamic features based real-time video smoke detection[J]. Computer Applications and Software, 2014, 31(2): 201−204. doi: 10.3969/j.issn.1000-386x.2014.02.054
[27]	奉国和. SVM分类核函数及参数选择比较[J]. 计算机工程与应用, 2011, 47(3):123−124. doi: 10.3778/j.issn.1002-8331.2011.03.037 Feng G H. Parameter optimizing for support vector machines classification[J]. Computer Engineering and Applications, 2011, 47(3): 123−124. doi: 10.3778/j.issn.1002-8331.2011.03.037

施引文献(29)

期刊类型引用(16)

1.	钟思琪，宁金魁，黄锦程，陈鼎泸，欧阳勋志，臧颢. 基于混合效应的杉木人工林冠幅模型. 森林与环境学报. 2024(02): 127-135 . 百度学术
2.	段平，王云川，晋秋梅，李佳. 基于无人机可见光影像的单木胸径估算方法. 测绘与空间地理信息. 2023(01): 14-17 . 百度学术
3.	魏智海，魏姿芃. 基于单木分割及点云特征提取的单木胸径估测. 陕西林业科技. 2023(02): 18-23 . 百度学术
4.	王杰芬，夏磊，林露花，胡璐璐，徐怀兴，王聚中，徐小军. 结合放射线法和无人机影像提取冠幅估算杉木碳储量研究. 浙江林业科技. 2023(05): 42-50 . 百度学术
5.	夏洪涛，郭晓斌，张珍，田相林，郭福涛，孙帅超. 基于不同立地质量评价指标的杉木大径材林分树高-胸径模型. 中南林业科技大学学报. 2023(10): 80-88 . 百度学术
6.	肖德卿，罗芊芊，范辉华，邱群，周志春. 栽植模式对木荷幼林生长和形质性状家系变异影响. 林业科学. 2022(05): 85-92 . 百度学术
7.	王志波，季蒙，李永乐，李银祥，马世明，张海东. 华北落叶松人工林差分地位指数模型构建. 林业资源管理. 2021(01): 156-163 . 百度学术
8.	杨洋，尤龙辉，叶功富，聂森，程分生，余锦林. 沙质海岸基干林木麻黄幼林模拟抚育预测. 福建农林大学学报(自然科学版). 2021(02): 206-215 . 百度学术
9.	田红灯，申文辉，谭一波，郑威，何琴飞，朱慧，甘国娟. 不同林龄杉木人工林冠幅与生长因子的关系. 中南林业科技大学学报. 2021(05): 93-101 . 百度学术
10.	朱晋梅，朱光玉，易烜，杨琬珑，牟村，王琢玙. 湖南省栎类次生林冠幅—胸径模型模拟研究. 湖南林业科技. 2021(03): 46-51 . 百度学术
11.	赵保国，朱江，艾训儒，姚兰，郭秋菊，洪建峰. 水杉原生种群胸径树高与树冠的通径分析. 东北林业大学学报. 2021(10): 16-20 . 百度学术
12.	于晓池，李凤，欧阳，张鹏，郭小龙，肖遥，赵秋玲，杨桂娟，王军辉，麻文俊. 基于表型的灰楸核心种质构建. 林业科学研究. 2021(06): 38-45 . 百度学术
13.	贾鹏刚，夏凯，董晨，冯海林，杨垠晖. 基于无人机影像的银杏单木胸径预估方法. 浙江农林大学学报. 2019(04): 757-763 . 百度学术
14.	张冬燕，王冬至，范冬冬，张健东，李大勇. 不同立地类型华北落叶松人工林冠幅与胸径关系研究. 林业资源管理. 2019(04): 69-73 . 百度学术
15.	伍小敏，徐春，杨汉波，陈炙，郭洪英，黄振，王泽亮. 四川桤木天然林和人工林的单木生长模型研究. 四川林业科技. 2018(04): 8-11+44 . 百度学术
16.	周凤艳. 沙地樟子松不同林龄树高、胸径等生长指标的关系研究. 吉林林业科技. 2017(01): 12-15 . 百度学术