WO2007137585A1

WO2007137585A1 - System for tracing refugee shelters by a satellite

Info

Publication number: WO2007137585A1
Application number: PCT/DK2007/000255
Authority: WO
Inventors: Mikkel Vestergaard Frandsen; Torben Vestergaard Frandsen; Christian Dalsgaard; Michael Schultz Rasmussen; Niels Henrik Eisum
Original assignee: Vestergaard Frandsen Sa
Priority date: 2006-05-31
Filing date: 2007-05-31
Publication date: 2007-12-06
Also published as: AP2008004699A0

Abstract

A system for tracing a refugee shelter having an upper surface of which one part is white and the other part is green in order to be more pronounced for satellite recognition. A satellite is configured to image the surface of the earth with multi colour camera for investigation with an image analysis system configured to analyse images with respect to objects having a green part and a white part, each part having a size equal to or larger than the resolution of the camera.

Description

System for tracing refugee shelters by a satellite

FIELD OF THE INVENTION

The present invention relates to refugee shelters, in particular the colours of covers or tents used by refugees. The invention also includes use of such a shelter for refugee tracing through satellite surveillance and includes a system for tracing refugee shelters with a satellite configured to image the surface of the earth.

BACKGROUND OF THE INVENTION

War and catastrophes, especially in the third world, often leave refugees without much protection in primitive camps waiting for help from the outside and waiting for the situation to improve and the war to end. Regularly, such refugee camps near war zones experience attacks from militant groups causing refugees to seek protection in other locations. These may be forests, mountains, hills or different types of waste lands. Typically, the distance for fleet is 10-20 km, but in certain cases, the refugees move as far as several hundreds of km, making humanitarian search for the hiding people very difficult.

The refugees seeking protection usually do not possess much apart from the tent or shelter they have received from help organisations when located at the refugee camp. The tent or cover is vital and may determine the survival rate, but other belongings are scarce such that there is a vital need for the help organisations to get in contact with the refugees as fast as possible. However, help organisations are typically not able to trace the refugees in vaste remote, deserted or forested areas, and the people are left in a totally unsafe situation. Thus, there exists a tremendous desire to improve the situation, especially the traceability, for this kind of people.

The material for tarpaulins and tents used for shelter of refugees typically has a blue or a white colour. The blue coloured plastic sheets as used by the United Nations Organisations are less visible on satellite images under certain circumstances, because the blue signal from the ground is reduced by scattering constituents in the atmosphere. White sheets are hard to recognize against bare soil, sand and dusty background (all bright surfaces) and thin clouds may equally add to obscuring their presence. Thus, there exists a need for improvement.

Bright red shelters would improve the situation in some respect. However, the refugees would be highly visible from large distances also by the militant groups, and the risk for recognition would result in a tendency not to take the otherwise vital shelter along, when the refugees are fleeing.

DESCRIPTION / SUMMARY OF THE INVENTION

It is therefore the object of the invention to provide means for improvement of the traceability for refugees. Especially, it is the purpose of the invention to provide a col- our pattern for shelter tarpaulins or tents that on the one hand improves the traceability for refugees by satellites but on the other hand do not leave the refugees highly visible to their pursuers.

This object is achieved with a shelter for refugees, such as a tarpaulin or tent, where the shelter has an upper surface of which one part is white and the other part is green, wherein the size of the white part and the size of the green part is at least 1.2 m by 1.2 m, or preferably 2 m by 2 m, in order to match the current resolution in optical instruments carried by surveillance satellites. For satellite born instruments for scanning or imaging the Earth different terminologies are used, such as Reflection Radiometer, CCD scanner, camera and others, hi the following, the general term camera is intended to cover the different terms.

The optical resolution of satellite camera, as found on the QuickBird satellite or the Ikonos satellite, is 2.4 m and 4 m, respectively, hi the panchromatic mode, the same instruments provide spatial resolutions of 0.6 m and 1 m, respectively. The multispec- tral information can be fused with the panchromatic data and hereby providing multis- pectral information at the level of 0.6 m and 1 m image units or pixels. Taking into consideration that a pixel can be a mixture of a tarpaulin and the soil background, the size of the coloured parts of the shelter must be in the order of at least two pixels, e.g. between 1.2 m and 2 m.

Very High Resolution satellite images (VHR) such as QuickBird and Dconos can be used for simple visual interpretation as done on traditional aerial photography. Improvement of the traceability can be achieved by a combination of the signals from the two types of cameras on each satellite. When combining the white and the green colour, the following advantage is achieved. The white colour has a generally high reflection in all bands, including in the panchromatic image. Using the white and green combination has the advantage of providing a unique pattern that can be recognised from multispectral satellite data, because the white part increases the overall reflectance and the green part enhances green reflectance that makes it distinct from the soil background. The green tarpaulin can easily be separated from the green vegetation by using a vegetation index. A vegetation index expresses the difference between re- fleeted near-infrared light and red (or green) and hereby identify vegetation that doing photosynthesis. At the same time, the green and the white surfaces when viewed from a distance or from the air, do not attract as much attention as, for instance, a red surface would do.

In addition, a shelter having an average size of around 4 m, the size is just large enough to approximately match the resolution in the multispectral satellite images.

For example, a shelter according to the invention may be a tent or sheeting, such as a polymer tarpaulin. Such tarpaulins are already in use for refugees, typically either white or blue coloured. According to the invention, the same type of material can be used, just with a different colour pattern.

For example, according to the invention, one half of the upper surface of the shelter may be green and the other half of the shelter white. Especially, the object of the in- vention is achieved with a shelter for refugees, where the shelter is at least 4 m times 4 m and preferably 4 m times 5 meter or 4 times 6 meter, such that the white and the green area each are at least 2 m by 4 m. Maximal visibility has been achieved with a light green colour having a peak of reflectivity between 520 ran and 600 nm, preferably at around 550 ran. The reflectivity should be higher than 20 % in order to differentiate the shelter against vegetation. However, better results are achieved, if the reflectivity is more than 30%, or even bet- ter more than 40 %

In a special embodiment, the colour has a FWHM of between 80 and 110 nm, preferably around 90 nm in the reflectivity curve, for example between 85nm and 95nm. With this bandwidth, the green spectral band of the satellite camera is covered, result- ing in a high measured intensity, that can be identified when doing the image analysis. The light green colour combined with the white will together provide two important advantages, first the overall reflection will be high and hereby allowing detection, either simple visual interpretation of images or through digital image processing. Furthermore, the green colour will permit the tarpaulins to be identified using multispec- tral satellite data. This is so because natural green vegetation will at the same time of being green (reflecting green radiation) also reflect near infrared radiation. The green tarpaulin will reflect green radiation and only limited near infrared radiation, hi addition, the geometric contrast between the white and the light green on the tarpaulin makes it a well defined geometrical feature to observe and the green colour reduces the risk that a pure white tarpaulin will be confounded with bare soil, sand and dust background or being obscured by thin clouds.

Furthermore, the light green colour having a peak of reflectivity at around 550 nm with a FWHM of around 90 nm is substantially different from the military camouflage green colour, such that confusion is avoided. The striping of green and white further reduces the risk of association with military targets. Even further, this green colour has proven not to attract tsetse flies, which is another remarkable and highly important beneficial property, hi contrast to this, it should be recognised that the blue colour used in current tarpaulins for shelter purposes do attract tsetse flies.

In a practical embodiment, the upper surface of the shelter with the green and the white part is made of a polymer tarpaulin. Such tarpaulins are already used for shelters with a durable protection against sunlight and rain. Experimentally, shelter made of a tarpaulin with a material thickness of 200g/m² has proven to be feasible. One of the materials used in the preliminary experiments was a tarpaulin with a center weave of thin plastic bands having a width of around 2 mm and a thickness of 0.2 mm that are interwoven to a web with a width of 2 m. The center weave was chosen to be black in order to prevent sunlight to be transmitted through the final tarpaulin. This center weave was then laminated on both sides with a low density poly ethylene (LPDE) plastic. Coloring agents were added to the lamination plastic before the lamination. The webs were welded together, cut into correct lengths and widths and folded at the edges with the insertion of a reinforcing cord into the folding in order to enhance stability and durability. Each corner was provided with a metal ring for fixation during use.

In a further embodiment, the shelter according to the invention has incorporated in its surface a pattern of electrically conducting material, the pattern being arranged for enhanced reflection of radar signals. This may be incorporated for easier recognition, especially, because radar is transmitted though clouds. For example, the electrically conducting material may be in the form of wires and/or metal strips. Alternatively, the conducting material may be conducting paint.

In a further embodiment, the shelter according to the invention may be provided with a radio responder, for example integrated in a thin microchip. Once, the area, where refuges are located, has been identified by the satellite system, radio responders may be used during an identification process and also be used in order to find other shelters in the nearby region, which by accident or due to other reasons have not been localised by the satellite system.

In a further embodiment, the material of the shelter may have a high absorption in the near infrared region such that for this wavelength region above 700 nm, the reflection is reduced - especially the region between 700 nm and 900 nm for the infrared wavelength band of the satellite camera. The absorbed near infrared radiation may be re- emitted at wavelengths longer than 900 nm. This gives another factor for differentiation against soil and vegetation, where the reflection for these wavelengths is high. Thus, a filter may be used for recognition of refugee shelters, where there is required a high intensity in the green band but a low intensity in the infrared band. This gives a high visibility contrast of the shelter against vegetation.

As disclosed in International patent application WO 2003/063587 by Vestergaard- Frandsen, the content of which is herewith incorporated, tarpaulins of this kind may contain insecticides, for example Pyrothroids and in particular Deltamethrin, to reduce the risk for diseases. Other insecticides or insect repellents that may be used in connection with the invention are also mentioned in WO 2003/063587

In addition, the shelter material, for example a tarpaulin, may contain a UV protective agent capable of reducing the UV radiation induced degradation of the pesticidal substance, such that the pesticidal substance may be optimally efficient as long as possible. Due to the reduced degradation of the pesticidal substance on the surface of the material, a relatively small amount of pesticidal substance has to be supplied to the surface by migration from inside the material to the outer surface of the material.

The shelter in combination with the insecticide are, in a further embodiment, constructed such that the migration speed of the pesticidal substance is fast enough to en- sure an effective level of the pesticidal substance on the outer surface of the material. On the other hand, in order to avoid overshooting of the necessary effective level of the pesticidal substance, the migration rate through the material may be controlled by migration moderation. Such moderation may be achieved by the physical properties of the shelter material, for example the density, the thickness or the distance from the insecticide reservoir inside the material to the outer surface. However, the migration rate for the pesticidal substance may also be controlled by migration inhibitors in the shelter material. Usable migration inhibitors are, for example, triazine derivatives, which at the same time have a fire resisting effect.

A shelter material, such as a tarpaulin, according to the invention may in a further development be constructed such that the release of the insecticide or the UV protecting agent or both is temperature dependent in a predetermined way. For example, if the shelter is to be used in tropic regions, a relatively high temperature can be expected, when the shelter material is exposed to sun light. In contrast, the temperature during storage and transport is usually much lower. This fact can be utilized by a temperature dependent migration speed such that the relatively low temperature during storage and transportation results in a slow migration or even negligible migration of the UV pro- tecting agent and/or the insecticide - allowing a long term storage of the shelter according to the invention - whereas the migration speed is increased when the shelter is exposed to sunlight or high temperature.

A temperature dependent release of the UV protecting agent is advantageous in tropi- cal regions because extensive exposure to sunlight with corresponding heating of the dispenser also increases the need for a relatively high amount of UV protecting agent. This way, the shelter material according to the invention functions as a self-regulating dispenser of a UV protecting agent.

If the material, for example a tarpaulin, of a shelter according to the invention contains a reservoir inside the material surrounded by outer wall elements, the UV protecting agent is preferably incorporated in the outer wall element, because it, this way, yields an efficient UV protection of the pesticide reservoir inside the material. Also, typically the migration speed is lower for UV protecting agents than for pesticidal agents, why the UV protecting agent preferably is located nearer to the material surface from the beginning than the pesticidal agent itself, for example in the outer layer.

By applying different migration inhibitors, where one inhibitor or one group of inhibitors are acting on the pesticidal substance and another inhibitor or group of inhibitors may act on the UV protecting agent, optimized migration rates may be achieved such that a perfectly match results between the amount of UV protecting agent and the pesticidal agent on the surface of the material of a shelter according to the invention.

In case that the pesticidal agent according to the invention is a mixture of a number of pesticidal substances, different migration inhibitors may be applied, where each inhibitor is directed towards rate migration control of at least one of the substances. Also, the insecticides by themselves will have different migration rates, and this can be exploited for a successive release of insecticides. Equivalently, in case that the UV protecting agent consists of a number of UV protection substances, a number of special migration rate reducing inhibitors may be applied for controlling the migration rate of the individual UV protecting substances. As above, the different migration rates of different UV filters or other types of protecting substances can be used to match with insecticides with different absorption spectra or to obtain a prolonged UV protection by combining fast and slow migrating filters. The different types of migration inhibitors may be chosen with substantial temperature dependence matching certain appliances, for example use in tropical regions with strong exposure to sunlight.

Possible migration inhibitors are metallic salts like bromides, which also may have some fire retardation effects. Substances as Carbon Black may be used as well, where Carbon Black has the additional property of being a UV protector. Substances as kaolin, stearates and migrating UV filters may, on the other hand, be used for increasing migration.

Optimization of the insect combating properties of the dispenser is furthermore achieved by matching the UV protecting agent or agents to the pesticidal agent in that the wavelength range for the most effective UV filter properties of the protecting agent overlaps with those wavelengths, where the pesticide is most sensitive for disintegra- tion.

hi order to prevent the polymer in a tarpaulin from UV induced disintegration, the different layers in the tarpaulin may contain additional UV protectors, for example Carbon Black, as far as the migrating UV protecting agent is not sufficiently effective.

Furthermore, the shelter material, such as a tarpaulin, may additionally contain HALS molecules. HALS is an abbreviation for Hindered Amine Light Stabilizers, which is a group of additives having a common chemical structure (a piperidine ring) as part of their molecule. These highly effective UV stabilizers protect the polymer by scaveng- ing free radicals that may occur due to UV irradiation. The effect of Hals molecules is also beneficially on the UV protecting agent as it prolongs the lifetime of this agent. The invention also concerns a system for tracing a refugee shelter, the system comprising a satellite configured to image the surface of the earth with an optical instrument, such as a camera, having multi colour wavelength bands for capturing light, for example reflected light, in these bands, wherein one of the bands is in a green colour wave- length region. The system also comprises an image analysis system using the information extracted from the optical instrument and having a first image resolution. The system is configured to analyse images with respect to objects having a green part and a white part, each part having a size equal to or larger than the first resolution.

Preferably, the system is configured to analyse objects, where the green part and the white part are of approximately the same size, for example in the order of at least 1.2 m, or preferably larger than 2 m, where the parts are adjacent to each other, hi a further embodiment, the system is configured to analyse objects, where the green part and the white part are of rectangular forms.

Furthermore, the satellite may also have a high resolution pan-chromatic instrument with a resolution of less than the resolution of the multiband instrument, and where the image analysis system is configured to combine the optical colour information form the multiband instrument with the optical high resolution information from the pan- chromatic instrument to achieve the first resolution as a fused resolution between the multiband instrument and the pan-chromatic instrument. The multiband instrument and the pan-chromatic instrument need not necessarily be two different instruments, but it may also refer to two different functioning modes of the same instrument.

By using a multispectral camera, for example with an optical resolution of 2.4 m, in combination with a panchromatic camera with a higher resolution, for example with a an optical resolution of 0.6 m, a colour image resolution can be achieved, which is higher than the resolution of the multiband camera. This may be achieved with software routines where digital multispectral data are fused with digital panchromatic data. In best cases, the resolution by the image analysis system including the multispectral data may be close to the resolution of the panchromatic camera. The enhanced coloured fused image data can successfully be used to identify objects, if the reflection of light from the object of interest can clearly be distinguished in the images. This rea- son is part of the decision to use the specific colour combination for refugee shelters according of the invention.

The system is intended for analysing images with respect to rectangular forms having a white surface with high reflection, and analysing the found rectangular forms with respect to green rectangular forms adjacent to the white rectangular form by using the multiband camera.

^"Preferably, the multiband colour bands comprises a near-infrared band, and the system comprises an optical filter, for example comprising software algorithms, for recognition of refugee shelters relatively to vegetation, the filter being programmed to require a high intensity of a searched object in the green band and a low intensity in the near- infrared band.

Advantageously, the system comprises a filter for recognition of refugee shelters relatively to vegetation. There are different possibilities for filters to be used for the analysis in order to differentiate between shelters according to the invention and vegetation.

One of the filters is programmed to require a relatively high Green-Red Processing Index (GRI) as compared to vegetation and a relatively low Normalised Difference

Vegetation Index (NDVI). The definitions of these two indices are GRI=(G-R)/(G+R) and NDVI= (NIR-R)/(NIR +R), where G refers to the total intensity captured in the green band of the camera from the object, R refers to the total intensity captured in the red band of the camera from the object, and NIR refers to the total intensity captured in the near infrared band of the optical instrument from the object.

Another filter is programmed to require a relatively low Near Infrared-Green Processing Index (NIGI) as compared to vegetation and a relatively low Normalised Difference Vegetation Index (NDVI). The definitions of these two indices are NIGI=(NIR- G)/(NIR+G) and NDVI= (NIR-R)/(NIR +R), where G refers to the total intensity captured in the green band of the camera from the object, R refers to the total intensity captured in the red band of the camera from the object, and NIR refers to the total intensity captured in the near infrared band of the optical instrument from the object. The applied optical filters, for example implemented as software algorithms, uses the indices in logical models for identifying a maximisation of the GRI and a minimisation of the NIGI and NDVI.

If plotted in a two-dimensional graph, with NDVI as abscissa and GRI or NIGI as or- dinate, the green part of the shelter according to the invention will be located in a different part of the graph than vegetation or bare soil. This implies an efficient differentiation between the green part of the shelter and the vegetation and the soil. For exam- pie, a filter defined as a set of algorithmic rules may be constructed to automatically perform a mathematical discrimination of the two dimensional values (GRI, NDVI) or (NIGI, NDVI) in order to separate objects from vegetation or from bare soil or both. In practice, this corresponds to only allowing objects to be imaged that are not ordinary vegetation or bare soil.

hi some cases, it may be useful that the infrared reflectivity is high, or even that the infrared or near-infrared part of the emitted radiation is enhanced, for example by fluorescence. This further enhances the visibility in regions with bare soil and scarce vegetation.

SHORT DESCRIPTION OF THE DRAWINGS

The invention will be explained in more detail with reference to the drawing, where FIG. 1 shows a shelter according to the invention as seen from above, FIG. 2 shows a) the spectral response from the panchromatic scanner and b) the spectral bands of the multispectral scanner,

FIG. 3 is a wavelength dependent reflection spectrum for typical vegetation, FIG. 4 shows a reflection spectrum for a white, a silver and a blue tarpaulin, FIG. 5 shows a reflection spectrum for a green coloured tarpaulin, FIG. 6 shows for different sources a graph with the green-red reflection index GRI relative to the Normalised Difference Vegetation Index (NDVI), FIG. 7 shows for different sources a graph with the Near infared - green index relative to the Normalised Difference Vegetation Index (NDVI), FIG. 8 shows reflection curves for three different cases.

DETAILED DESCRIPTION / PREFERRED EMBODIMENT

FIG. 1 shows a shelter according to the invention as seen from above and in perspective when correctly installed. One half of the shelter surface is white, whereas the other half is green. The white colour has a general high reflection, making it easy to discover, and the green half in combination with the white makes it possible to differentiate the shelter surface from other natural or man made objects. This is elucidated in more detail in the following.

ha order to understand the capabilities of satellites with respect to imaging of ground areas, the following is relevant. Satellites as the QuickBird satellite or the Ikonos satellite provides high resolution panchromatic images with a resolution of 0.6 m and 1 m, respectively, and images with a more coarse multi spectral resolution of 2.4 m and 4 m, respectively. FIG. 2a shows the spectral response of the panchromatic camera mode of the Ikonos satellite and FIG. 2b illustrates the response of the multispectral camera mode of the Ikonos satellite with the different wavelength bands. There are four bands, where the first band is located around the blue colour, the second band around the green colour, the third band around the red colour and the fourth band in the near- infrared region.

FIG. 3 shows a wavelength dependent reflection pattern for vegetation. The upper, almost straight, curve is the reflection from bare soil or reflection from soil, which is barely covered by vegetation. The lower curve corresponds to the reflection from vegetation with a minimum in the red region and a maximum in the infrared region. It should be noted that vegetation despite the green colour only has a small reflection in this region, which is due to the fact that during photosynthesis, also the green part of the spectrum is absorbed, hi turn, a large amount of infrared energy is released from the vegetation resulting in a rising of the curve at wavelengths between 700 nm and 750 nm. For wavelengths longer than 750nm the high reflection in the infrared region is due to the cell structure of the vegetation and called Foliar reflectance. The latter may be utilised in a further embodiment, where the material of the shelter may have a high absorption in the near infrared region such that the reflection is reduced especially the region between 700 nm and 900 nm, where the satellite camera has its infrared wavelength band. This gives another factor of differentiation against soil and vegetation, where the reflection for these wavelengths is high. Thus, a logical model may be used for recognition of refugee shelters, where there is required a high intensity in the green band but a low intensity in the infrared band. This gives a high visibility contrast of the shelter against vegetation.

In the background of the spectrum of FIG. 4, the multispectral band response of the satellite camera is shown.

FIG. 4 illustrates the wavelength dependent reflection for tarpaulins used as shelter for refugees. The white tarpaulin reflects little in the UV regime, but has a high reflection in the visible and infrared region. The silver tarpaulin reflects less in the visible but more in the UV part. The blue tarpaulin has a local maximum of reflection in the blue region and increased reflection in the infrared region. It should be noted that despite the blue colour, the reflection in the blue band is less than from the white tarpaulin. The white and the silver tarpaulin when used alone lack spectral characteristics in the corresponding bands against the soil apart from the fact that the reflection is higher.

The Blue tarpaulins have the disadvantage of a general low reflectivity together with the fact that the general scattering of the blue colour in the atmosphere blurs the blue band image. In addition, wet lowlands, marsh or mellow type land cover typically ap- pears as blue in satellite images.

The reflection spectrum of a possible green colour used for the invention is illustrated in FIG. 5. It is clearly seen that the reflection has a maximum in the green band, The optimal curve does have a high green peak in the green part of the satellite spectrum, whereas the reflection is substantially lower in the other bands. The infrared reflection is in the vicinity of 20%, which is approximately what can be realistically expected from possible green colours. When combining the white and the green colour, the following advantage is achieved. The white colour has a generally high reflection in all bands, and, especially, it has a high reflection for the panchromatic camera, which images with high resolution, where also the green part of the shelter contributes in the pan-chromatic spectrum due to its high reflectivity. Thus, the high resolution panchromatic camera may provide images, where high intensity spots may be candidates for refugee shelters. By making a fusion of the panchromatic and the multispectral data on these spots, the white and green combination can be used to identify whether the found object indeed is a refugee shelter. At the same time, the green and the white surfaces when viewed from a dis- tance or from the air, are not too different from the vegetation and soil background and do not attract attention as, for instance, a red surface would do. In addition, having a size of around 4 m, the size is just large enough to approximately match the resolution of both fused panchromatic data as well as the multispectral bands. Thus, the combination of a green and a white colour in the material of the shelter is a simple solution for a complex problem.

FIG. 6 is a graph having the Normalised Difference Vegetation Index (NDVI) as abscissa and the green-red reflection index GRI as ordinate. The NDVI index is defined as NDVI= (NIR-R)/(NIR +R), where NIR refers to the total radiation captured from the object in the near infrared band of the camera, and R refers to the total radiation captured in the red band of the instrument from the object. The GRI index is defined as GRI=(G-R)/(G+R), where G refers to the total radiation captured in the green band of the instrument from the object.

As can be seen in the graph, for a white tarpaulin, the GRI is low as well as the NDVI. For a laurel leaf, the GRI is relatively low, but the NDVI is high due to the relatively high near-infrared emission, see FIG. 3, which is partly due to reflection and partly due to conversion of light into the near-infrared spectrum because of photosynthesis. A high intensity of 40% in the near infrared band is shown for vegetation in FIG. 3.

A Blue tarpaulin is seen in the spectrum of FIG. 6 to be not far from the laurel leaf. Thus, blue tarpaulins are not easily differentiated from vegetation. In contrast, the white tarpaulin is easy to differentiate from vegetation, which is not surprising. Sur- prisingly however, green painted tarpaulin and thin green plastic are in entirely other parts of the graph and, therefore, easy to differentiate from vegetation by this optical filter. The optimal curve of FIG. 3 is seen to be far from the laurel leaf as well. The reflection curves for the shown green tarpaulin, the green plastic foil and the optimal green shelter material is illustrated in FIG. 8 as an overlay upon the band spectrum of FIG. 2b.

A discrimination of vegetation relative to the optimised green shelter is also facilitated by use of another index, namely the Near Infared - Green index relative to the Normal- ised Difference Vegetation Index (NDVI) as illustrated in FIG. 7. Again, the index of the blue tarpaulin is located in the spectrum near the laurel leaf, whereas the index of the white tarpaulin and the optimal green tarpaulin are at great distance in the graph.

With reference to the two-dimensional graphs of FIG. 6 and 7, it is obvious that the green part of the shelter according to the invention is located in a different part of the graphs than vegetation, the vegetation being represented by the laurel leave data. It should be pointed out that bare soil would occur in the graph very near to the vertical, ordinate axis, as the NDVI index is low for bare soil. This implies that an efficient differentiation between the green part of the shelter and the vegetation and the soil can be performed. For example, the filter may be constructed to automatically perform a mathematical discrimination of two dimensional values (GRI, NDVI) or (NIGI, NDVI) for vegetation and bare soil. In practice, this can be achieved by neglecting all measurements, for which the two dimensional values (GRI, NDVI) or (NIGI, DNVI) are within certain intervals. These intervals correspond to certain areas of the graphs in FIG. 6 or 7, respectively. The analysis, whether the scanned regions contain image data representative for refugee shelters, may include the requirement of only allowing data to be analysed further, if these data do not belong to those areas in the graphs that represent vegetation or bare soil. In other word, the analysed data do not represent ordinary vegetation or bare soil.

For instance, the requirement of an NDVI being less that 0.7 or rather 0.6 would discriminate vegetation. The additional requirement of an NDVI of more than 0.3 or rather 0.3 or possibly 0.4 would effectively discriminate against bare soil. Further- more, the requirement of a GRI of more that 0.4 or rather 0.5 would discriminate effectively against bath vegetation and soil. A combination of all these three requirements would increase the efficiency of the filter. Alternatively or in addition to the requirements for GRI, the filter may use the requirements for a NIGI or less than 0.3 or rather less than 0.1.

Another filter may be constructed alternatively or in addition to require at least one of the following criteria to be fulfilled:

- the GRI of the object is larger than 1.5 times the GRI of the background vegetation; - the NDVI of the object is less than 0.8 times the NDVI of the background vegetation;

- the NIGI of the object is less than 0.4 of the NIGI of the background vegetation. This filter takes into account the fact that the illumination of the object and the surrounding background vegetation may change the absolute values in the different indices. In addition, this filter would be relatively insensitive to dust particles in the at- mosphere which influences the colour of the reflected light and also the transmission of the light through the atmosphere.

Claims

1. A shelter for refugees having an upper surface of which one part is white and the other part is green.

2. A shelter according to claim 1, wherein the size of the white part and the size of the green part each is at least 1.2 m by 1.2 m.

3. A shelter according to claim 1, wherein the size of the white part and the size of the green part is at least 2 m by 4 m.

4. A shelter according to claim 1, 2, or 3, wherein about one half of the upper surface is white and the other half is green.

5. A shelter according to any preceding claim, wherein the green part of the shelter has a green reflectivity at a wavelength between 520 nm and 600 run.

6. A shelter according to claim 5, wherein the peak reflectivity is around 550 nm.

7. A shelter according to claim 4 or 5, wherein the FWHM of the green reflectivity curve is between 80 and 110 nm.

8. A shelter according to claim 6, wherein the FWHM of the green reflectivity is around 90 nm.

9. A shelter according to any preceding claim, wherein the peak reflectivity for the green part is at least 20%.

10. A shelter according to claim 8, wherein the reflectivity for the green part is at least 30%.

11. A shelter according to any preceding claim, wherein the material incorporated in its surface electrically conducting material for enhanced reflection of radar signals.

12. A shelter according to any preceding claim, wherein the material of the shel- ter has a high absorption in the near infrared wavelength region between 700 nm and

900 nm.

13. A shelter according to anyone of the claims 1-11, wherein the outer surface of the tarpaulin contains a fluorescence substance for enhancement of the near-infrared component from the surface of the shelter.

14. A shelter according to any preceding claim, wherein the upper surface of the shelter with the green and the white part is made of a polymer tarpaulin.

15. A shelter according to claim 14, wherein the tarpaulin contains an insecticide.

16. A shelter according to claim 14 or 15, wherein the insecticide is contained within the tarpaulin material and configured for migration from inside the material to the surface of the material.

17. A shelter according to claim 15 or 16, wherein the insecticide is contained in a reservoir inside the tarpaulin.

18. A system for tracing a refugee shelter according to any preceding claim, com- prising a satellite configured to image the surface of the earth with an optical instrument having multi colour wavelength bands for capturing light in these bands, wherein one of the bands has a green colour wavelength region, and comprising an image analysis system using the information extracted from the optical instrument and having a first image resolution, and wherein the system is configured to analyse images with respect to objects having a green part and a white part, each part having a size equal to or larger than the first resolution.

19. A system according to claim 18, wherein the system is configured to analyse objects, where the green part and the white part are of approximately the same size and adjacent to each other.

20. A system according to claim 18, wherein the system is configured to analyse objects, where the green part and the white part are of rectangular forms.

21. A system according to claim 18, wherein the system is configured to analyse objects with a green part and a white part with a size in the order of at least 1.2 m, preferably larger than 2 m.

22. A system according to any preceding of the claims 18-21, wherein the satellite also has a high resolution pan-chromatic instrument with a resolution of less than the first resolution of the multiband instrument, and where the image analysis system is configured to combine the optical colour information form the multiband instrument with the optical high resolution information from the panchromatic instrument

23. A system according to any preceding of the claims 18-22, wherein the multi- band colour bands comprise a near-infrared band, wherein the system comprises an optical filter for recognition of refugee shelters relatively to vegetation, the filter being programmed to require a relative high intensity of a searched object in the green band and a relative low intensity in the near-infrared band as compared to vegetation.

24. A system according to any preceding of the claims 18-23, wherein the multi- band colour bands comprise a red band and a near-infrared band, wherein the system comprises an optical filter for recognition of refugee shelters relatively to vegetation, the filter being programmed to combine a Green-Red Processing Index (GRI) and a Normalised Difference Vegetation Index (NDVI), where the definitions of these two indices are GRI=(G-R)/(G-R) and NDVI=(NIR-R)/(NIR +R), where G refers to the total intensity captured in the green band of the camera from the object, R refers to the total intensity captured in the red band of the camera from the object, and NIR refers to the total intensity captured in the near infrared band of the camera from the object.

25. A system according to any preceding of the claims 18-24, wherein the multi- band colour bands comprise a red band and a near-infrared band, wherein the system comprises an optical filter for recognition of refugee shelters relatively to vegetation, the filter being programmed to combine a Near Infrared-Green Processing Index (NIGI) and a Normalised Difference Vegetation Index (NDVI), where the definitions of these two indices are NIGI=(NIR-G)/(NIR+G) and NDVI= (NIR-R)/(NIR +R), where G refers to the total intensity captured in the green band of the camera from the object, R refers to the total intensity captured in the red band of the camera from the object, and NIR refers to the total intensity captured in the near infrared band of the optical instrument from the object.

26. A system according to claim 24 or 25, wherein filter comprises a mathematical discrimination of two dimensional values (GRI, NDVI) or (NIGI, NDVI) or both if they are within one or more predetermined two dimensional intervals, the intervals containing values typical for vegetation or for bare soil or both.

27. A system according to claim 24, 25, or 26, wherein the optical filter is programmed to discriminate objects for which at least one of the following requirements is valid - the GRI is larger than 0.3,

- the NIGI is less than 0.3,

- the NDVI is less that 0.7,

- the NDVI is larger than 0.1.

28. A system according to claim 24, 25, 26, or 27 wherein the optical filter is programmed to discriminate objects for which at least one of the following requirements is valid

- the GRI of the object is larger than 1.5 times the GRI of the background vegetation;

- the NDVI of the object is less than 0.8 times the NDVI of the background vegetation; - the NIGI of the object is less than 0.4 of the NIGI of the background vegetation.

29. Application of a shelter according to anyone of the claims 1 to 17 or of a system according to any one of the claims 18-27 for refugee tracing by satellite surveillance.