FR3048841A1 - METHOD AND DEVICE FOR CAPTURING AND RESTITUTING THERMAL IMAGE - Google Patents

Abstract

The thermal image capture device (10) comprises: - a thermal camera (22) successively supplying uncompensated thermal images of the sensor pixel response variations and the temperature differences of the camera, - a circuit (24) of determining differences, pixel by pixel, between two images successively provided by the camera, providing an image of differences and - a circuit (26) for remote transmission of the difference image. In embodiments, the capture device further includes a difference image compression circuit (29), for example to binarize the difference image. Preferably, the device comprises a camera support (20) surrounded by a tube (14) protecting the camera support from the wind.

Description

TECHNICAL FIELD OF THE INVENTION

The present invention relates to a method and a device for capturing and restoring the thermal image. It applies, in particular, to the surveillance of protected sites, parking, tolls and car traffic routes.

STATE OF THE ART

It is recalled here that a thermal camera records the various infrared radiation (heat waves) emitted by the bodies and which vary according to their temperature. It reproduces the heat stored by a body, or shows the heat flow of a wall due to a fireplace at the back.

Although the wavelength of infrared radiation depends on the temperature, thermal imaging cameras usually have only one channel, and the cameras only produce an image of the intensity of the radiation, which also makes it possible to appreciate the temperature of the source. The color produced by the camera is sometimes a false color, obtained by associating a color with the received intensity, in order to facilitate the direct reading of the temperature: to each color of the image corresponds a temperature.

There are mainly two types of thermal imaging cameras: - cameras with an uncooled infrared sensor. The sensor works by measuring the variation of a quantity (current, voltage) as a function of the temperature at each point of the sensor. This temperature varies depending on the amount of infrared radiation received. As this type of camera does not need a cryogenic enclosure, it is cheaper than the other type, but suffers from lower performance and - cameras with a cooled infrared sensor. This type of camera uses a container cooled by cryogenic techniques, the sensor being enclosed in a vacuum chamber or in a Dewar vessel. The sensor used is a photographic sensor but thanks to the use of materials different from those of cameras, it allows the acquisition in the field of infrared. Without a cooling system, the sensor would be dazzled by its own infrared emission.

Thermal cameras or infrared cameras are defined by their spatial resolution (the smallest visible object) and their thermal resolution (the smallest discernible difference in temperature). These two resolutions are not independent and the cameras are generally characterized by the curve giving revolution of the thermal resolution according to the spatial resolution.

Thermal image capture systems are very expensive. In particular, when a set of thermal cameras must be deployed on a site to be protected, the cabling necessary for feeding and communicating the captured images is complex and expensive.

In thermal cameras, the electrical signal from the pixels that make up the thermal sensor is not identical for an identical temperature of the parts of an object seen by a camera. There is nonlinearity of pixels to adjacent pixels. This is why the raw image that comes from a thermal sensor is unusable in the state. It is necessary for each temperature environment and each variation of this environment to make a fast image acquisition with a mechanical shutter at uniform temperature in front of the sensor, then to generate a compensation matrix pixel by pixel so that after compensation, one can get an image truly representative of the temperature of each point that composes the scene. The atmosphere includes the scene but also the temperature of the camera itself, which is not necessarily the same everywhere (for example, wind, or sun ...). However, this pixel-by-pixel processing of a succession of compensated images has two major drawbacks. It requires a complex and powerful electronics consuming energy. Moreover, during the mechanical shutter the camera becomes blind and does not perceive a fast simultaneous movement.

OBJECT OF THE INVENTION

The present invention aims to remedy all or part of these disadvantages. For this purpose, according to a first aspect, the present invention is directed to a thermal image capture device, which comprises: a thermal camera successively supplying uncompensated thermal images of the sensor pixel response variations or the temperature differences of the sensor; the camera; a pixel-by-pixel difference determining circuit between two images successively provided by the camera, providing a difference image; and a remote transmission circuit of the difference image.

Thanks to these provisions, on the one hand, the energy consumption of the device is reduced, the amount of data to be transmitted is low, no pixel response variation compensation being achieved and the thermal image capture can have a low resolution and not be cooled and, on the other hand, the bandwidth used for transmission is reduced.

In embodiments, the device further comprises: a difference image receiving circuit, a memory of an image taken in the visible spectrum by means of an electronic image sensor, and whose content of the scene is preferably as large as that observed by the thermal camera and - a chip of the image of the differences received, in the image stored in memory.

Thanks to these provisions, it is possible, during the visualization of the image resulting from the incrustation, to situate an event in a visual context. Indeed, one of the drawbacks of a conventional thermal imaging camera, even after linearity compensation processing, is that there is very little difference in temperature in an outdoor natural scene and that situating an event in the context is difficult.

In these embodiments, an acquisition of a color reference image is performed, very easy to interpret for a human being, with a field similar to the thermal sensor of the camera. This acquisition can be done, for example, at the time of commissioning or regularly if the context of the scene evolves. The context reference image can be memorized in a centralized manner, which makes it possible to transmit only a difference image, very low in bandwidth, before embedding on reception, in order to have the binary profile in a context image , preferably in color.

In embodiments, the device includes a distance estimation circuit, at the camera, of a dimension form represented by the image of the differences, as a function of the lowest image line of the represented shape. by the image of the differences.

Thanks to these provisions, the operator is informed on the position of the moving form. Since the distance estimation circuit estimates the distance as a function of the lowest image line of the shape represented by the difference image, the determination of the distance is fast.

When an intruder object enters the optical field of the camera, it generates a variable electrical level difference, which is simplified to detect by image differences, possibly binarized beyond a predetermined value. We thus obtain a simplified image part of the surface, the profile, the shape of the object in motion, but this image is sufficient for an operator to analyze and interpret the current event in the field of the camera. By knowing the angle of vision of the camera, its height of implantation and the height of position in the image of the moving profile, one calculates rather precisely the distance between the camera and the moving form, as well as its real dimension . These are very often necessary but sufficient elements in the interpretation of an event.

In embodiments, the device further includes a difference image compression circuit that receives the difference image and provides a compressed difference image to the remote transmission circuit of the difference image.

Thanks to these arrangements, the amount of data to be transmitted, the bandwidth used to transmit them and the electrical energy consumed to transmit them can be reduced.

In embodiments, the compression circuit is configured to binarize the difference image.

Recall here that binarize consists of comparing the value of the electrical signal from each pixel of the image sensor with a predetermined value, each pixel of the difference image thus taking a binary value, "0" or "1". Thanks to these provisions, the image of moving elements is more contrasted and the amount of information to be transmitted is greatly reduced.

In embodiments, the device further includes a camera mount surrounded by a windproof tube supporting the camera mount.

Thanks to these arrangements, the camera support is protected from the wind and does not swing in the wind. The camera and the images it provides are therefore more stable. When determining the difference in images, each pixel of the successive images corresponds to the same optical field.

In embodiments, the image memory is configured to provide a color image.

Thanks to these provisions, the visualization of the moving form in its context is easier.

In embodiments, the device includes a displacement speed estimation circuit of the shape represented by the difference image.

Thanks to these provisions, the operator is informed on the speed of the moving form.

According to a second aspect, the present invention aims at a method of thermal image capture, which comprises: a step of capturing a succession of uncompensated thermal images of the sensor pixel response variations and the temperature differences of the sensor; the camera, by a thermal camera, a step of determining differences, pixel by pixel, between two images successively provided by the camera, providing an image of differences and a step of remote transmission of the image of differences.

In embodiments, the method further comprises: a step of receiving the difference image, a step of providing, by a memory, an image of the scene observed by the camera and a step of of incrustation of the image of the differences in the stored image of the scene observed by the camera.

Since the advantages, aims and characteristics of this method are similar to those of the capture device and the reproduction device that are the subject of the present invention, they are not recalled here.

BRIEF DESCRIPTION OF THE FIGURES Other advantages, aims and features of the invention will emerge from the following description of at least one particular embodiment of the seat fixing device, with reference to the appended drawings, in which: FIG. 1 represents, schematically and in sectional view along a vertical plane, a particular embodiment of the thermal image capture device that is the subject of the present invention; FIG. 2 represents, in a block diagram, a particular embodiment of the thermal imaging device object of the present invention, - Figure 3 shows a top view of the optical field on the ground of a camera, - Figure 4 shows schematically an image displayed by a rendering device. FIG. 5 represents, in the form of a logic diagram, steps of a particular embodiment of the capture methods. and thermal image rendering, objects of the present invention. DESCRIPTION OF EXAMPLES OF EMBODIMENT OF THE INVENTION The present description is given in a non-limiting manner.

As of now, we note that the figures are not to scale.

The thermal camera used in the present invention is preferably a microbolometer camera.

In a bolometer, an incident signal P is absorbed by the surface C of the bolometer which is made of an absorbent material, such as metal. This material heats up to the temperature T. Its surface is connected by a thermal conductor G to a heat sink, maintained at a constant temperature. The temperature difference ΔΤ = P / G can be measured using a thermistor. The temporal thermal constant (t = C / G), defined by the heat transfer rate between the surface of the bolometer and the heat sink, gives the sensitivity of the detector.

In general, metal bolometers do not require cooling, as the heat of the thin layer used dissipates rapidly. The metal is increasingly replaced by a semiconductor or even a superconductor that must be refrigerated at very low temperatures but have a very high sensitivity.

In order to increase their sensitivity and reduce their intrinsic "noise", ie the parasitic radiation they emit, bolometers are sometimes cooled to very low temperatures (around a few Kelvin, that is, ie at temperatures below -269 ° C). This cooling is done, for example, by liquid helium (4.2K) and which can sometimes be pumped to have superfluid helium (1.4K) according to the ranges studied. However, this cooling is expensive in terms of camera complexity and energy consumption. The present invention implements a sensor camera neither cooled nor compensated in temperature variation nor compensated for variation in the response of the sensor pixels.

As seen in FIG. 1, the thermal image capture device 10 comprises a photovoltaic panel 12 mounted on an outer tube 14 connected to a ground housing 16 which contains a battery 18. An inner tube 20, to the inside the outer tube 14, carries a thermal camera 22 below the photovoltaic panel 12. An image processing circuit 24 receives the electrical signals representative of the images captured by the thermal camera 22. A compression circuit 29 compresses the data from the An image processing circuit 24. A radio transmitter 26 remotely wirelessly transmits the image signals provided by the image processing circuit 24.

These provisions have many advantages. The photovoltaic panel 12 supplies electricity to the battery 18 during the day. The photovoltaic panel 12 also protects the thermal camera 22 from direct sunlight. The battery 18 supplies the thermal camera 22, the circuit 24 and the transmitter 26 with electricity. Because of its weight, the battery 18 ensures the mechanical stability of the device 10, which limits the civil engineering operations necessary for the implantation. of the device 10. The inner tube 20, being protected from the wind by the outer tube 14, is not subject to oscillations under the effect of wind. As a result, the thermal camera 22 is stable, as well as the images it provides.

Preferably, the thermal camera 22 is an uncooled microbolometer camera, which reduces its cost price and eliminates the need for cooling. Preferably, the thermal camera 22 has a low resolution, for example 80 lines of 80 points, which reduces its power supply requirement and the amount of image data to be processed by the circuit 24.

The image processing circuit 24 makes a difference, in absolute value, for each pixel of the image sensor, between two successive images supplied by the thermal camera 22. Due to the mechanical stability of the thermal camera 22 carried by the internal tube 20, the difference between two successive images represents only the parts in which there has been a movement in the optical field of the thermal camera 22. In fact, the pixels of the images which represent fixed parts of the scene exhibit the same illumination values. The difference between these successive values is therefore zero, with the noise close. To eliminate this noise, the image processing circuit eliminates the difference values less than twice the noise.

The image compression circuit 29 then performs compression of the image calculated by difference of images. For example, the absolute values of the differences are binarized, each pixel of the difference image being able to take only one of two values, depending on whether the difference is less than a predetermined limit value or greater than or equal to this limit value. The amount of image data is thus reduced. In other embodiments, known type image compressions are performed instead of or in addition to the binarization.

In the remainder of the description, the image obtained by calculating the difference, for each pixel of the sensor, in absolute value, of the illumination values between two images successively supplied by the thermal camera 22 is called "difference image". , compressed or binarized.

The compressed signal is transmitted remotely by the transmitter 26. As can be understood, as long as there is no movement in the scene, the compressed image of the difference has very little data. A single bit can represent an image in which all the pixels have presented, between two successive images, a difference less than twice the noise level. As soon as a movement occurs in the scene observed by the thermal camera 22, pixels of two successive images have different signal levels. The difference for each of these pixels is therefore greater than twice the noise. However, the number of these pixels is usually very limited. It depends on the size of the object or the person moving the distance, the camera 22, this object or this person, the focal length of the camera 22, the width of the photosensitive portion of the sensor. the camera 22 and the speed of the movement.

Thus, the transmitter 26 consumes extremely little energy and occupies an extremely narrow bandwidth as long as no movement occurs in the scene observed by the thermal camera 22. The inventor has determined that a transmitter 26 can operate at 19200 bauds and consume less than 25 mW.

As illustrated in FIG. 2, a device 10 also comprises a thermal image rendering part 30 comprising a receiver 32 receiving the signal transmitted by the transmitter 26. A decompression circuit 34 supplies, from the received signals, the image difference obtained by the image processing circuit 24.

An image memory 36 provides a photograph, preferably in color, whose point of view is the same as that of the thermal camera 22. This photograph is first taken by a visible image sensor, for example a camera digital. Preferably, the photograph represents a larger optical field than that of the thermal camera 22, in order to give a more intelligible context to the movements represented by the difference of images provided by the decompression circuit 34.

An overlay circuit 38 embeds the image of the differences in the photograph, instead of matching the scenes represented by the photograph and the difference in images.

A screen 40 receives and displays the image with the incrustation of the differences in the photograph.

Thus, a security operator who observes the screen 38 easily observes the moving object or person and the part of the site that is monitored by the thermal camera 22, as in daylight, even though very little data is available. image have been transmitted and it is dark. The operator can quickly estimate the risk and its situation in the protected site. He can quickly send an intervention team to the right place, telling him what to intercept, single person, group of people, vehicle, animal or other object.

Preferably, during the embedding, the circuit 38 embosses in bright color, for example red, the image of the differences coming from the decompression circuit 34. Thus, the attention of the operator is more strongly called on the screen 40 What is even more important if the observed scene is an undeveloped part, trees and vegetation having green, brown or black colors, on which the red color is in strong contrast. The image of the differences provided by the decompression circuit 34 is permanently stored in a memory 42 in order to facilitate subsequent investigations.

Preferably, a distance estimation circuit 44 and, possibly, a speed of the shape (object or person) in motion, processes the image of the differences and supplies the distance data and possibly the speed data, in a digital manner to the circuit. inlay 38, which inserts these data into the stored photograph. Thus, the display on the screen includes an estimated distance data and possibly an estimated speed data near the object or the person in motion. These data can help the safety operator to estimate the risk and the nature of the object or person in motion.

The distance is preferentially estimated by first determining the shape of the larger dimension represented by the image of the differences. The largest dimension is, for example, the largest area, in pixels, or the greatest height, in pixels. The extraction of such a form is known to those skilled in the art. The estimation of the distance, at the camera, of the larger dimension represented by the image of the differences, is preferentially carried out as a function of the lowest image line of this form represented by the image of the differences .

We therefore determine the lowest point of the decompressed image where a non-zero image difference is represented. Then, by triangulation, the estimation circuit 44 determines to what point on the ground, in the scene observed by the thermal camera 22, corresponds this lowest point. For this purpose, the circuit 44 is preliminary informed on the height of the camera 22 relative to the ground and the angle 28 (Figure 1) of inclination of the thermal camera 22 relative to the average plane of the scene observed.

Once the estimated distance, preferably, the estimation circuit 44 estimates the speed of the object or the person in motion by measuring the number of pixels traveled during a sequence of images and the width of the optical field of the camera 22, in meters, at the estimated distance. The width of the optical field is the product of the distance estimated by the width of the image sensor, divided by the focal length of the camera 22, the focal length and the width of the sensitive portion of the image sensor being in the same units and the field width and the estimated distance being in the same units. For example, a camera with a focal length of 20 mm and a sensor width of 10 mm observes a field width of 50 meters at 100 meters distance.

The speed of the moving object or person is obtained by multiplying the number of pixels traveled per time unit, the optical field width of the thermal camera 22 to the estimated distance and dividing this product by the number of pixels. pixels on each line of the image sensor of the thermal camera 22. For example, if this optical field measures 50 meters at 100 meters distance from the camera 22 and a line of the image sensor of the camera 22 presents 80 pixels, three pixels of displacement per second correspond to a speed of 1.875 meters per second, or 6.75 km / h (because 1.875 = 3 x 50/80).

It is observed that when the lowest point of the shape detected by difference of image changes line of the image, it is possible to re-evaluate the distance of the object or of the corresponding person as well as its speed. Alternatively, the speed is decomposed into a component in the direction of the camera 22 and a component in a plane perpendicular to the axis of vision of the camera 22.

Alternatively, the height of the moving object or person is estimated and displayed on the screen 40. This estimated height is proportional to the estimated distance of the object or person and to the number of lines of the image. of difference where a non-zero difference in values between two successive images, for the same pixel of the image sensor appears.

In FIG. 3, the lower half of the ground optical field 50 of the camera 22 is observed. This optical field is the intersection of the plane of the ground and the optical field of the camera, which is conical. Since the camera 22 generally has an optical field that covers the horizon, the upper part of the optical field on the ground, not shown, comprises part of this horizon. As understood, the width of the optical field, at a given distance, is proportional to this distance. If, as is generally the case, the sensor of the camera 22 has horizontal lines and the ground is generally horizontal, each line of the camera represents a line on the ground perpendicular to the optical axis of the camera 22.

By knowing the inclination of the optical axis of the camera relative to the mean plane of the ground in the half-optical field 50, a distance is assigned to each line of the sensor of the camera 22 and therefore of the image provided by the camera 22. As explained above, a field width on the ground is jointly assigned to this line.

FIG. 4 shows an image 60 displayed on the screen 40. This image 60 includes, in incrustation on the photograph stored in the optical field of the camera, the pixels of the image difference which correspond to a difference in value. greater than the limit value of binarization. These pixels are assembled into a shape 70.

By searching for the low point of this shape 70, a line 62 of the image is obtained. This line 62 is, as explained above, associated with a distance (association 64 shown on the left of FIG. 4) which gives the estimate of the distance of the moving object or person in the field of the camera 22. This line 62 is also associated with an optical field width on the ground (association 66 shown on the right of FIG. 4). By measuring the displacement, in number of pixels, of the shape 70 from its previous state 72, to a previous unit of time, an estimated speed, in number of pixels per unit of time, is obtained. Each pixel representing a distance equal to the ground optical field width divided by the number of pixels per line of the sensor of the camera 22, the speed is converted into the traveled pixels per unit of time in meters per second or, as illustrated in FIG. 4, in kilometers per hour.

The estimated distance and the estimated speed are encrusted at 74 in image 60.

FIG. 5 shows steps of a particular embodiment 80 of the thermal image capture methods (steps 82 to 88) and of a thermal image recovery method (steps 90 to 102).

During a step 82, the thermal camera provides a thermal image. During a step 84, the absolute value of the difference between this thermal image and the previous thermal image, pixel by pixel, is determined. That is, for each pixel of the image sensor, the value for that pixel is subtracted from the current image of the value for that pixel in the previous image. The absolute value of this difference is thus obtained.

During a step 86, the image of the absolute values of the differences is compressed, for example, on a binarise. During a step 88, a radio signal is emitted representative of the compressed image of the absolute values of the differences.

During a step 90, the radio signal is received and the received data are stored. During a step 92, the received image is decompressed, in the case where the compression does not involve a simple binarization.

During a step 94, the distance of the shape represented by the image of the differences is estimated as explained with reference to FIGS. 3 and 4. During a step 96, the speed of the shape represented by the image of the differences, as explained with reference to FIGS. 3 and 4. During a step 98, a height of the shape represented by the image of the differences is estimated, as explained above.

During a step 100, a digital photograph of the scene observed by the thermal camera 22 is read in memory. This photograph, preferably in color, is produced in visible light, for example with a digital camera, by day, when installing the image capture device.

During a step 102, in a photograph memorized preliminary of the optical field of the camera, it is preferentially embedded in color: - the shape represented by the image of the absolute values of the differences and - the estimated values, in particular of distance and of speed.

During a step 104, the image resulting from this overlay is displayed.

As understood from the foregoing description, the thermal image capture method consumes little power and low bandwidth and provides a stable image. In addition, the thermal image rendering method provides an image calling the attention of the operator, including the context in which the object or person in motion is located and distance and velocity data. This rich image is easy to interpret by the operator.

Claims (10)

1. Device (10, 30) for thermal image capture, characterized in that it comprises: - a thermal camera (22) successively supplying uncompensated thermal images of the sensor pixel response variations and temperature differences of the camera, - a pixel-by-pixel difference determining circuit (24) between two images successively provided by the camera, providing an image of differences and - a circuit (26) for remote transmission of the difference image. . 2. Device (10, 30) according to claim 1, which further comprises: - a circuit (32) for receiving the image of differences, - a memory (36) of an image taken in the visible spectrum through an electronic image sensor, and whose content of the scene is preferably as large as that observed by the thermal camera and - a circuit (38) for embedding the image of the differences received, in the image stored in memory. 3. Device (10, 30) according to one of claims 1 or 2, which comprises a circuit (44) for estimating the distance, to the camera, of the shape represented by the image of the differences, as a function of the lowest image line of the shape represented by the image of the differences. 4. Device (10, 30) according to one of claims 1 to 3, which further comprises a circuit (29) for compressing the image of the differences and providing a compressed difference image to the circuit (26) of remote transmission of the image of differences. The device (10, 30) of claim 4, wherein the compression circuit (29) is configured to binarize the difference image. 6. Device (10, 30) according to one of claims 1 to 5, which further comprises a camera support (20) (22) surrounded by a tube (14) windproof the camera support. The device (10, 30) according to claim 2 or one of claims 3 to 6 when dependent on claim 2, wherein the image memory (36) is configured to provide a color image. 8. Device (10, 30) according to one of claims 2 or 7 or one of claims 3 to 6 when dependent on claim 2, which comprises a circuit (44) for estimating the travel speed of the shape represented by the image of the differences. 9. Method (80) for capturing a thermal image, characterized in that it comprises: a step (82) of capturing a succession of uncompensated thermal images of the variations of responses of the pixels of the sensor and of the differences camera temperature sensor, by a thermal camera (22), - a pixel-by-pixel difference determining step (84), between two images successively provided by the camera, providing a difference image and - a step (88) remote transmission of the image of differences. The method (80) of claim 9 further comprising: - a step (90) of receiving the difference image; - a step (100) of providing, by a memory, a picture of the scene observed by the camera and - a step (102) of incrustation of the image of the differences in the stored image of the scene observed by the camera.