1 / 20

Rome, 15 May 2014 - 09.00 - 16.00 hrs

Progress of the activities at DIAEE: towards a strategy for comparing the monitoring systems available within the consortium. Pablo Marzialetti , DIAEE, pablomarzialetti@psm.uniroma1.it Giancarlo Santilli, DIAEE, santilli@psm.uniroma1.it. Rome, 15 May 2014 - 09.00 - 16.00 hrs.

Télécharger la présentation

Rome, 15 May 2014 - 09.00 - 16.00 hrs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Progress of the activities at DIAEE: towards a strategy for comparing the monitoring systems available within the consortium Pablo Marzialetti, DIAEE, pablomarzialetti@psm.uniroma1.it Giancarlo Santilli, DIAEE, santilli@psm.uniroma1.it Rome, 15 May 2014 - 09.00 - 16.00 hrs 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  2. Introduction • One of the tasks of the ODS3F project concerns the comparison between the different available systems. • The comparison, to be meaningful and correct, should take into account the different environmental conditions (topography, vegetation type, weather, etc.) in which the systems operate. • Expected results: The results of the activity regard: • the assessment of the advantage and disadvantage associated with each available system; • the evaluation of the accuracy, adequacy and completeness of the information provided by the system. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  3. Monitoring Systems Feasibility • Limitations • Optical • Visibility • Fog occurrences • Contrasts analysis • False Alarms • Minimum smoke column heights • Algorithm robustness • Infrared • Topographic barriers • Flame energy • Algorithm robustness • Products • Optical • Contrast maps, in order to train identification and classification methods • Slope Position classification and Topographic Position Index • Fog Stability Index • Visibility Index • Fog/Low stratus cloud map • Smoke Viewshed analysis • Infrared • Topographic barriers • DSM vs DTM • Binary Viewshed analysis • Our objective: identify main territorial characteristics in order to compare different monitoring systems in next summer tests. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  4. Monitoring Systems Feasibility There are several environmental factors that potentially affects the visibility and that they changes depending on the territorial geography. FOG is often described as a stratus cloud resting near the ground. Its formation is complex and its occurrence is widely variable in space and time, forming under a wide range of meteorological circumstances. Valley fogforms where cold dense air settles into the lower parts of a valley condensing and forming fog. It is often the result of a temperature inversion with warmer air passing above the valley. Valley fog is confined by local topography and can last for several days in calm conditions during the winter. Fog Stability Index (FSI) The FSI(HOLTSLAG 2010, WANTUCH 2001) is an empirical method, developed by the US Air Force. Is calculated according to the following formula : FSI = 2 * (TS - T850 ) + 2 * ( TS - DP ) + W850 Bad weather conditions 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  5. Monitoring Systems Feasibility Fog Stability Index (FSI) FSI = 2 * (TS - T850 ) + 2 * ( TS - DP ) + W850 stability humidity wind speed FSI<31 indicates a high probability of fog formation, 31<FSI<55 implies moderate risk of fog, and FSI>55 suggests low fog risk. Fog formation is favored for high humidity (TS-DP small), the atmosphere is stable (weak mixing, TS-T850 is small) and low wind speed (no mixing, W850 is small) (HOLTSLAG 2010, FREEMAN 1998). In order to find a connection with visibility in (WANTUCH 2001) was made a comprehensive statistical analysis of direct measurements and derived physical quantities, finding that the best correlation was: Visibility = -1.33 + 0.45 * FSI 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  6. Monitoring Systems Feasibility 5 years Historical Serie of Fog Stability Index & Visibility Index (Jan-2009 Feb-2014) Spatial Resolution: 0.125º degrees Temporal Resolution: daily at 06:00,12:00 & 18:00 Source Meteo data: ECMWF European Centre for Medium-Range Weather Forecasts FSI 06:00 VIS 06:00 VIS 12:00 VIS 18:00 FSI 12:00 FSI 18:00 Multitemporal Pixel statistics (grouped by month) distribution of daily ROIs. means 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  7. Monitoring Systems Feasibility 5 years Historical Serie of Fog Stability Index & Visibility Index (Jan-2009 Feb-2014) Source Meteo data: ECMWF European Centre for Medium-Range Weather Forecasts 70 FSI January 12:00 PM  20 FSI January 06:00 AM • Seasonality of FSI & VIS index during the 5 years analysis. • Maximum Constant Difference in FSI index from 06:00 to 12:00 products. • Higher FSI values for Italy and Greece & Lower FSI values for Spain ROI at midday . • During winter rather similar values between ROIs. at 06:00 AM. & at 06:00 PM, while during winter, at 06:00 AM, Spain ROI evidence greater FSI risk than the other ones.. January July FSI January 06:00 PM FSI July 06:00 AM FSI July 12:00 PM France & ROI Greece & ROI Italy & ROI Spain & ROI 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  8. Monitoring Systems Feasibility Terrain Illumination & Aspect, could affect feature extraction processing, improving or reducing potential contrasts. Visibility of objects depends among other things on the perception of luminance contrasts between the objects and their surroundings. The greater the contrast of an object with its background, the greater its visibility. Illumination, Sunset & Sunrise 9 different illumination layers were processed for each ROI. 1 represents a synthesis of the day. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  9. Terrain characterization Slope Position Classification & FSI Slope Position Classification A Topographic Position Index (WEISS 2001) was introduced in order to classify at different scales the ROIs landscapes into Slope Positions. Geoprocessing results put into evidence the preeminence of valley-class, where valley fogs could be present. Topographic Position Index 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  10. Terrain characterization Viewshed Analysis Binary Viewshed Analysis Viewshed identifies the cells in an input raster that can be seen from one or more observation points or lines. Each cell in the output raster receives a value that indicates how many observer points can be seen from each location. If you have only one observer point, each cell that can see that observer point is given a value of one. All cells that cannot see the observer point are given a value of zero. The observer points feature class can contain points or lines. The nodes and vertices of lines will be used as observation points (ESRI ArcGIS Help). Clear topographic differences, which evidence visible surfaces seen by cameras (main condition for thermal cameras location) SORIA Binary Viewshed ARTA Binary Viewshed 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  11. Terrain characterization Viewshed Analysis Smoke Viewshed Analysis When we talk about optical systems, we can go further the binary viewshed analysis detailed before. Smoke columns could go higher, becoming visible depending on Camera and DEM quotas. For that case, a Line of Sight (DEM-pixel to Tower) process was developed, in order to create a new viewshed, capable to quantify the minimum smoke column heights necessary to be seen by the camera. Camera PROVENCE DEM & Camera PROVENCE smoke column viewshed At left we can see deepest areas in dark-black, classified as not visible by Binary Viewshed analysis. While at right, we can see traditional Binary Viewshed analysis results in white, and in the opposite and in dark-red, smoke column minimum quota that need to be reached to be seen by the camera. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  12. Terrain characterization Viewshed Analysis The Smoke Viewshed Analysis, put into evidence local topography and its high impact, not only for thermal monitoring systems, but for optical ones. From the first examples below, significant information can be extracted, only viewing minimun quota magnitudes and distribution. MONTE CAVO Smoke Viewshed PROVENCE Smoke Viewshed • Note: In further analysis, should be introduced precise heights of cameras to not underestimate critical areas. • (covered area and minimum quotas change significantly at different cameras heights). 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  13. Monitoring Systems Feasibility • Products developed: • Topographic Index and Classification (roughness index)  visibility • Fog Stability and Visibility Index (historical data)  visibility • Relative Humidity (historical data)  visibility • Aspect and Illumination  contrast • Binary Viewshed analysis  visibility • Minimum smoke heights analysis  visibility • Distances from Camera (buffering zones at 500, 1000, 2500, 5000 and 10000 meters, and distance to seashores)  visibility • … under development • Evapotranspiration •  with potential influence on the visibility • Fog/Low Stratus Cloud satellite product •  visibility and contrast impacts • Dispersion Index •  visibility 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  14. Monitoring Systems Feasibility Evapotranspiration (Penman-Monteith method) Known as the sum of evaporation and plant transpiration from the Earth's land and ocean surface to the atmosphere, will have potentially influence on the visibility. FAO Penman-Monteithmethod, with historical meteo data, (and potentially Local near real time data, because the Penman methods may require local calibration of the wind function to achieve satisfactory results). Through evapotranspiration, trees in cloud forests collect the liquid water in fog or low clouds onto their surface, which drips down to the ground. Fog/Low Stratus Cloud satellite product (Cermak 2008) This near real time product (temporal resolution of 15 minutes), this product could add information to improve knowledge of impacts on visibility and contrast. Dispersion Index (Lavdas 1995) Measure the atmosphere’s ability to ventilate smoke from areas of prescribed burning activity. It will be used to improve the visibility index based on FSI. • Next step • Terrain characterization 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  15. Monitoring Systems Feasibility • Terrain characterization • We have exposed territoral characteristics that could enhance or impact on Fire Monitoring Systems performance. In the next step, we will make a terrain characterization approach, dividing each product within its main classes. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  16. Monitoring Systems Feasibility • Terrain characterization Events & characteristics Training process • characteristics weighted • according to events. • potential influences • of the territory. Terrain characteristics Note: Event = feature detected by the system, not necessarily a fire/smoke event Commissionerrors(event Detected that is a False Alarm) >>Need to be reduced Omission errors (event Not Detected that is a not a False Alarm) >> Need to be reduced 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  17. Monitoring Systems Feasibility • Products availability • Some products exposed can be consulted by the EOSIAL WMS, at: • http://eosial.psm.uniroma1.it:8081/geoserver • Historical series of Fog Stability Index: • http://eosial.psm.unirom1.it:8081/geoserver/FSI_ODS3F/wms? • General Project(under development): • (DEM, Aspect, Illumination, CLC, Cameras, Buffers, ROIs.limits, etc.) • http://eosial.psm.unirom1.it:8081/geoserver/ODS3F/wms? • Smoke Viewshed Analysis: • http://eosial.psm.unirom1.it:8081/geoserver/SMOKEVIEWSHED_ODS3F/wms? • FOG/Low Stratus Clouds (under development): • http://eosial.psm.unirom1.it:8081/geoserver/FOG_LOWSTRATUS/wms? 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  18. Monitoring Systems Feasibility • Conclusions & requirements • Because of several difficulties exposed by partners to introduce an external image dataset, the algorithms could not be tested in exact equal conditions. • Nevertheless, the products described will help us to evidence those terrain characteristics that could impact in detection systems. • High Spatial Resolution DEMs. for every area of interest (to improve the accuracy of products). • Precise heights of cameras to not underestimate viewshed analysis. • Dataset of events detected and their characteristics, events not detected and false alarms. • Further works will look to improve FSI/Visibility correlation depending on local measurements. • Potential addition of WMS links in monitoring systems to help decision makers to discriminate false alarms from real fire events. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  19. some references ToreyinB.U., "Fire Detection Algorithms Using Multimodal Signal and Image Analysis”, PhD thesis, Bilkent University, Department of Electrical and Electronics Engineering, Ankara, Turkey, 2009. Chan-Yun Y., Wei-Wen T., Jr-Syu Y, “Reducing False Alarm of Video-Based Smoke Detection by Support Vector Machine”, ISI 2008 Workshops, pp. 102-113, 2008. Damir K., Darko S., Toni J, “Histogram-Based Smoke Segmentation In Forest Fire Detection System”, INFORMATION TECHNOLOGY AND CONTROL ,2009 Dongil H., Byoungmoo L., “Flame and smoke detection method for early real-time detection of a tunnel fire”, Fire Safety Journal , pp. 951-961,2009 Ho C., “Machine vision-based real-time early flame and smoke detection”, MEASUREMENT SCIENCE AND TECHNOLOGY, 2009 Jayavardhana G., Slaven M., & Marimuthu P., “Smoke Detection in Video Using Wavelets and Support Vector Machines”, Fire Safety Journal, pp. 1110-1115, 2009 Jerome V., Philippe G., “An image processing technique for automatically detecting forest fire”, International Journal of Thermal Sciences, pp. 1113–1120, 2002. Nobuyuki F., Kenji T., “Extraction of a Smoke Region Using Fractal Cording”, International Symposium on Communications and Information Technologies, Sapporo, 2004. Philippe G., Jerome V., “Real-time identification of smoke images by clustering motions on a fractal curve with a temporal embedding method”, Society of Photo-Optical Instrumentation Engineers, 2001. Qinjuan L., Ning H., “Effective Dynamic Object Detecting for Video-based Forest Fire Smog Recognition”, Image and Signal Processing CISP '09. 2nd International Congress on, pp. 1-5, 2009. Steven V., Peter L., Rik Van D. W., “State of the art in vision-based fire and smoke detection”, 14th International Conference on Automatic Fire Detection, 2009. Toreyin B. U., Cetin A. E., “Wildfire detection using LMS based active learning”, in Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2009. Toreyin B. U., Dedeoglu Y., Cetin A. E., “Contour based smoke detection in video using wavelets”, 14th European Signal Processing Conference EUSIPCO. Florence, 2006. Toreyin B. U., Dedeoglu Y., Cetin, A. E., “Wavelet based real-time smoke detection in video”, European Signal Processing Conference (EUSIPCO), 2005 Yu Cui H. D., “An Early Fire Detection Method Based on Smoke Texture Analysis and Discrimination”, 2008 Congress on Image and Signal Processing, pp. 95-99, 2008. Yuan F., “A fast accumulative motion orientation model based on integral image for video smoke detection”, Pattern Recognition Letters, pp. 925-932, 2008. Zhou Y., Yi X., Xiaokang Y., “Fire Surveillance Method Based on Quaternionic Wavelet Features”, MMM 2010, pp. 477-488, 2010. Ziyou X., Rodrigo C., Hongcheng W., Alan M. F., Muhidin A. L., Peng a. P.-Y., “Video-based Smoke Detection: Possibilities, Techniques, and Challenges”, Suppression and Detection Research and Applications – A Technical Working Conference (SUPDET 2007), 2007. Turgay Ç., Hüseyin Ö., Hasan D., “Fire and smoke detection without sensors: Image processing based approach”, 15th European Signal Processing Conference EUSIPCO 2007, Eurasip, pp. 1794 – 1798, 2007. Gomez-Rodriguez F., Arrue B. C., Ollero A., “Smoke monitoring and measurement using image processing: application to forest fires”, Proc. SPIE pp. 404, 2003. Thou Ho C., YenHui Y., ShiFeng H., “The Smoke Detection for Early Fire Alarming System Based on Video Processing”, IEEE 2006 International Conference on Intelligent Information Hiding and Multimedia Signal Processing (IIHMSP), pp. 427-430, 2006. Yu C., Zhang Y., Fang J., Wang J., "Texture Analysis of Smoke for Real-Time Fire Detection," Second International Workshop on Computer Science and Engineering, vol. 2, pp.511-515, 2009 Piccinini P., Calderara S., Cucchiara R., "Reliable smoke detection in the domains of image energy and color," Image Processing, 2008. ICIP 2008. 15th IEEE International Conference on , vol., no., pp.1376-1379, 2008 Cucchiara R., Grana C., Piccardi M., Prati A., “Detecting moving objects, ghosts and shadows in video streams,” IEEE Trans. on PAMI. Vol. 25, no. 10, pp. 1337-1342, 2003. Maruta H., Kato Y., Nakamura A., Kurokawa F., “Smoke detection in open areas using its texture features and time series properties”, Industrial Electronics, 2009. ISIE 2009. IEEE International Symposium on, vol., no., pp.1904-1908, 2009. Jing Y., Feng C., Weidong Z., "Visual-Based Smoke Detection Using Support Vector Machine," Natural Computation, 2008. ICNC '08. Fourth International Conference on , vol.4, no., pp.301-305, 2008. DongKeun K., Yuan-Fang W., "Smoke Detection in Video”, WRI World Congress on Computer Science and Information Engineering , vol. 5, pp.759-763, 2009. Zhengguang X., Jialin X., "Automatic Fire Smoke Detection Based on Image Visual Features," International Conference on Computational Intelligence and Security Workshops (CISW 2007), pp.316-319, 2007. Zheng W., Xingang W., Wenchuan A., Jianfeng C., "Target-Tracking Based Early Fire Smoke Detection in Video", Fifth International Conference on Image and Graphics, pp.172-176, 2009 KaewTraKulPong, P., Bowden R., “An improved adaptive background mixture model for real-time tracking with shadow detection”, Video-based Surveillance Systems: Computer Vision and Distributed Processing, pp. 135–144, 2002. Chao-Ching H., Tzu-Hsin K., "Real-time video-based fire smoke detection system," Advanced Intelligent Mechatronics, 2009. AIM 2009. IEEE/ASME International Conference on, vol., no., pp.1845-1850, 2009. Conner W.D., Hodkinson J.R., "Optical properties and visual effects of smoke-stack plumes". A cooperative study: Edison Electric Institute and U.S. Public Health Service. Environmental Protection Agency. Office of Air Programs. Research Triangle Park, N.C. Revised May 1972. Fisher P.F., “Extending the Applicability of Viewsheds in Landscape Planning». Photogrammetric Engineering & Remote Sensing, Vol. 62, No. 11, November 1996, pp. 1297-1302. Lavdas and Achtemeier, “A Fog and smoke risk index for estimating roadway visibility hazard”. National Weather Digest, Volume 20, Number 1. October 1995. Cermak J. And Bendix J., “A novel approach to fog/low stratus detection using Meteosat 8 data”. Atmospheric Research 87 (2008) 279–292. 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

  20. Thank you for your attention 4rd ODS3F Progress Meeting, 15 - 16 May 2014, Rome

More Related