The fast remote detection of landslides impacting on railway tracks and roads has become a crucial transport safety issue. Automatic Change Detection analyses of images collected by cameras is one of the most promising solutions for this purpose. At present, the most common sensors for this application are based either on visible-light cameras, or on microwave radars. However, visible-light cameras are blind in the darkness and through fog, while, on the other hand, radars cannot provide a clear view of the obstacles, thus causing frequent false positive alarms.
In order to solve the above limitations, in this project we propose to implement newly developed micro-bolometer cameras for the long-wave infrared ranges into a prototype for railway/road surveillance system. The physics and technology aspects of microbolometers will be analyzed for optimal detection of greybody spectrum from different types of rocks and debris. A prototype of a telescope for small-object identification will be built designing and realizing special mid- and long-wave infrared optics. At the same time, a dedicated software based on change detection algorithms will be developed and calibrated for the automatic identification of landslide invasions on railways/roads.
In the frame of the project, the prototype will be tested firstly in the natural landslides laboratories of the Department of Earth Sciences of the "Sapienza" University of Rome (Poggio Baldi and Acuto Natural Labs). Then, selected locations along railways and roads will be identified with partners like the national Italian Railway network (RFI) and the National Italian Road Department (ANAS). The business strategy will be left for future evaluation (spin-off, patenting or licensing), but preliminary steps of intellectual property protection will be considered during the project development.
The use of advanced infrared vision systems for landslides monitoring along railways and roads is the main innovation introduced in the present project. We believe that the flat metal surface of the railway tracks and the asphalt of roads will have a good contrast in thermal imaging vision with respect with fresh rock and debris. Therefore, the idea should be tested experimentally in a suitable environment. In order to show the degree of innovation of our proposal, we compare the expected performances of the proposed solution with potential competing technologies such as enhanced vision cameras and radars.
- The lack of visibility through fog, smoke, dust and heavy rainfalls is an effect of light scattering due to small particles suspended in air (water droplets, ashes, sand) basically cancelling, over tens of meters of propagation, the direction of initial emission of the photons hence making image formation impossible. Cameras sensitive to single-photon levels in the visible range are currently commercialized, but it is not the lack of sensitivity that prevents vision through fog and rainfall. Even the most sophisticated enhanced vision cameras cannot compete with infrared vision through thick fog, dust and smoke [6]
- Microwave-based radar technology is in continuous development, but it is not yet at the stage of providing clear images to be used for machine or human vision, mainly due to the physical limits posed by diffraction. Briefly, Abbe's diffraction principle states that the shorter the wavelength, the smaller the feature size that can be detected. Synthetic Aperture Radars (SAR) using wavelengths of several centimeters can form two-dimensional images at distances of several hundreds of meters with details down to a few meters at most. Automotive radars using millimeter waves have high resolution in the distance measurement (ranging to avoid collision), but do not produce full two-dimensional images. Clear vision of centimeter-scale objects, such as a single railway track, at a hundred meter distance is not at reach of existing radar systems. To fill this technology gap, terahertz radar prototypes have been developed e.g. by the NASA, but their cost is estimated above 1 MEUR making their use in civil applications improbable.
- Finally, it can be noted that infrared telescopes present on the market are constituted of few pyroelectric detector pixels (e.g. by ASAS GmbH or by Eltec Inc.), and as such operate in anti-intrusion alarm-mode only, but have no capability of forming an image.
In a few words, optical cameras, recently used for identifying the invasion of tracks and roads by rock or debris, are showing several limitations during the night and also during complex weather conditions (e.g. fog, rainfalls), i.e. the most critical conditions for rockfalls and debris-flow triggering. On the other hand, many commercial infrared sensors are characterized by a low resolution that makes the sensibility to small blocks limited, especially at distances larger than few tenths of meters. For such a reason, we are planning the development of a monitoring solution that should be able to solve most of the limitations described above, thus providing an answer to the issue caused by the invasion of tracks due to rockfalls and debris-flows.
Hence, with the herein proposed Thermal InfraRed monitoring solution it will be possible to face slope monitoring needs that are still not solved by tools available in the market. Specifically, fast landslides like rockfalls and debris-flows are usually characterized by small volumes, widespread distribution over long railways segments and characterized by a sudden activation. These peculiar features make their temporal prediction very difficult, hence, tools for risk reduction along railways and roads are recently focusing, instead, on their early detection after the occurrence.
As a further induced benefit with high societal impact of this project, applications of a high-performance infrared vision system are foreseen in structural monitoring of buildings, health condition screening of passengers (remote body temperature), cultural heritage, non-destructive material testing.
[6] D. Wikner, Progress in Millimeter-Wave Imaging, Proc. SPIE Vol.7936 - 0D1.