A décamouflage method of an object in a scene observed by a plurality of devices comprising a device for acquiring multi-spectral images, and a thermal image acquisition device, and a device implementing said method.

One method used since ancient times to conduct surveillance is to assign an observing role to a human being. The man then uses his visual system and the auditory system to detect objects or people. The monitoring methods using the human visual system may be in default when objects or people to detect blend into their environment, using, for example, camouflage techniques. Objects or people and then become camouflaged stealth within the meaning of the human visual system, that is to say, they are invisible or hardly visible to the human eye.

Recent years have seen the emergence of surveillance systems based on a variety of devices capable of capturing information that can reveal the presence of an object or a human. Such devices include image acquisition devices operating in different spectral bands ranging from the visible range to the infrared range. These spectral bands are located in particular:

• in the visible (VIS) having wavelengths ranging from 0.38 to 0.78 micrometers (μτη)

• in the near infrared ( "Near Infra Red (NIR)" in English terminology) (0.78 Ιμπι)

· In the infrared wavelengths shortwave ( "Short-wavelength infrared (SWIR)" in English terminology (1 to 2.5 μτη)

• in the infrared at medium wavelengths ( "Medium-wavelength Infrared (MWIR)" in English terminology) or mid-infrared (2.5 to Βμπι)

· Infrared wavelength-long wave ( "Long-wavelength

Infrared (LWIR) "in English terminology) (5 to 14μπι).

The image acquisition devices operating in the visible range such as VDO image acquisition devices (Direct Way Optical) and DCV (Route Day Colors) provide photographs tell VDO images respectively

VJC pictures, close to what would be a human being. One easily understands that the VDO image acquisition devices and VJC bring little or no relevant information on an object in a scene when this object blends into its environment.

It is known that some quasi-invisible objects in the visible, appear more clearly in certain areas of the infrared. It is then common to couple the image acquisition devices operating in the visible range with acquiring infrared images devices. It is also possible to use image capture devices across a range of wavelengths (or spectral band) larger comprising a plurality of spectral bands in the visible range and / or the field of infrared. This type of image acquisition devices, called IMS image acquisition devices subsequently, is capable of picking up multi-spectral images (IMS), comprising a plurality of components, each component corresponding to a spectral band IMS acquired by the image acquisition device.

Among the image acquisition devices operating in the infrared, the image acquisition devices is known operating in the medium infrared wavelengths and / or wavelengths (variously referred devices acquiring thermal image or VTH (thermally) in the following) capable of capturing a thermal signature of an object or of a human being. The acquisition of thermal imaging devices suffer from certain limitations in a ground surveillance context. Indeed, when a monitoring area is the floor, the thermal image acquisition devices may be sensitive to thermal clutter effects ( "Thermal clutter" in English terminology) caused by hot objects corresponding not to search objects, such as for example stones heated by the sun. This sensitivity to thermal clutter can then cause false alarms. Moreover, due to the thermal clutter, a search object can be found drowned in very noisy information. Furthermore, it is known that thermal image acquisition devices are very effective in detecting static objects on the ground during the day.

Although improvements images from the DCV image acquisition devices, VDO, IMS and thermal (called DCV pictures respectively, VDO, IMS and thermal (ie VTH) thereafter) are possible by methods of

image processing, these improvements are generally deemed unsatisfactory. Thus it is possible to improve the IMS or DCV image by image processing methods highlighting contrasts in said images. However these processes, we call processes of contrast enhancement thereafter are relatively effective in bringing out a silhouette of an object or a human being, but do not allow to highlight internal details to that figure. But it may be interesting to obtain internal details to the outline of an object in order to better identify said object.

It may be noted that while noisy, thermal image can provide valuable information about internal details to the outline of an object.

Furthermore, it is known to more than a combination or an alternate information from VJC image display, VDO, IMS and heat are not satisfactory.

It is desirable to overcome these disadvantages of the prior art.

It is particularly desirable to provide a method and a device enabling effective décamouflage of an object or of a human being in a scene. It is more desirable that said method and said apparatus facilitating an identification of said object or said human. In other words, it is desirable that said method and said device are adapted to provide, for example to an operator to monitor a scene, an image including an outline of a search object and details of said object in the figure.

According to a first aspect of the present invention, the present invention relates to a décamouflage method of an object in a scene observed by a plurality of devices comprising an image acquisition device, said multi-spectral, comprising a plurality of components each representative of a spectral band within a visible and / or near infrared and / or infrared shorter wavelengths and an image acquisition device, said thermal, comprising at least one component representative of a spectral band comprised in the mid-infrared and / or infrared long wavelengths. The method comprises: obtaining a multi-spectral image and a thermal image, each component of the multi-spectral image and each component of the thermal image being aligned spatially and temporally therebetween; obtaining at least a position of a sub-part of an image, said window, and for each position obtained: extracting one of each multi-spectral image window

and heat to said position; applying a contrast enhancement process to the extracted at least one window comprising a window derived from the multi-spectral image, said procedure, when applied to a window to obtain a window, said window improved in wherein a contrast between pixels corresponding to the object and pixels not corresponding to the object is increased; forming a multi-component window, each window obtained improved and extracted each window that has not been applied said procedure providing at least one component of the multicomponent window; and, applying said process to the multi-component window; generating an image, said image reproduction, by inserting each improved window obtained by applying said process to each multi-component window formed in a representative receiving image of the scene.

Said method, thanks to the coupling between information from multi-spectral images, and information from thermal imaging is used to provide an operator to monitor a scene, an image including an outline of the search object and details said object in the figure. The visualization of the object and its details is improved.

In one embodiment, the contrast emphasizing process comprises when applied to a window: obtaining at least one position of a first mask adapted to contain pixels corresponding to said object in said window and for each position: Position said mask to said position in said window; defining a second mask including pixels of said window not included in the first mask; and, applying a projection Fisher to the pixels of said window to provide improved window wherein a contrast between pixels of the first and second mask is increased.

In one embodiment, the first mask is adapted so that each pixel of the object is contained in the first mask.

In one embodiment, the first mask is adapted to contain each pixel of a detail of said object of interest to identify said object.

In one embodiment, the method comprises for the window extracted from the multi-spectral image and the window derived from the thermal image: applying the contrast enhancement procedure for a plurality of positions of the first mask in each of said windows , the plurality of positions to fully cover the object; forming a first single improved window from

each improved window obtained during each application of the contrast enhancement process to the extracted window of the multi-spectral image and a second single improved window from each improved window obtained during each application of the sharpening procedure contrast to the window derived from the thermal image; and form the multi-window components from the first and second improved single window.

In one embodiment, the method comprises for the multi-component window formed: applying the contrast enhancement procedure for a plurality of the first mask positions in the multi-window components, the plurality of positions to fully cover the object; forming a third improved single window from each improved window obtained during each application of the contrast enhancement process to the multi-component window; use the third single window to generate improved the restitution image.

In one embodiment, the plurality of components of the multi-spectral image comprises at least one spectral band within the visible region corresponding to a primary color red and / or blue and / or green and for each position of said window obtained the method comprises: applying the contrast enhancement process to the window derived from the multi-spectral image, each component corresponding to a spectral band within the near infrared and / or infrared wavelengths shorter n 'being taken into account; calculating a contrast value, said visible contrast value between the pixels corresponding to the first mask and the pixels corresponding to the second mask of the improved window obtained following the application of the contrast enhancement procedure; and terminate the implementation of décamouflage method of an object to the position of said window obtained when said visible contrast value is greater than a predefined threshold, said threshold visible.

In one embodiment, the thermal image comprises at least two components and for each position of said window obtained the method comprises: applying the contrast enhancement process to the window derived from the thermal image; calculating a contrast value, said thermal contrast value between the pixels corresponding to the first mask and the pixels corresponding to the second mask of the improved window obtained following the application of the window to the contrast enhancement procedure derived from the thermal image; and terminate the implementation of décamouflage method of an object to the position of said window obtained when the thermal contrast value is greater than a predefined threshold, said threshold temperature.

In one embodiment, the multi-spectral images are representative of spectral bands in a spectral band ranging from "0.4" to "1" μτη or "0.6" to "1" μπι or "0.9 "to" 2.5 "μτη and thermal images are representative of a spectral band between" 3 "and" 5 "μτη or between" 8 "and" 12 "μπι.

According to a second aspect of the invention, the invention relates to a décamouflage device of an object in a scene observed by a plurality of devices comprising an image acquisition device, said multi-spectral, comprising a plurality of components each representative of a spectral band within a visible and / or near infrared and / or infrared shorter wavelengths and an image acquisition device, said thermal, comprising at least one component representative a spectral band included in the mid-infrared and / or infrared long wavelengths. The apparatus comprises: obtaining means to obtain a multi-spectral image and a thermal image, each component of the multi-spectral image and each component of the thermal image being aligned spatially and temporally therebetween; obtaining means for obtaining at least a position of a sub-part of an image, said window, and for each position obtained: extracting means for extracting a window of each of the multi-spectral images to said thermal position; applying means for applying a contrast enhancement process to the extracted at least one window comprising a window derived from the multi-spectral image, said procedure, when applied to a window to obtain a window , said improved window, wherein a contrast between pixels corresponding to the object and pixels not corresponding to the object is increased; forming means to form a multi-component window, each window obtained improved and extracted each window that has not been applied said procedure providing at least one component of the multicomponent window; and, application means for applying said process to the multi-component window; generating means for generating an image by inserting each improved window obtained by applying said procedure

each multi-component window formed in a representative receiving image of the scene.

According to a third aspect of the invention, the invention relates to a computer program comprising instructions for implementing, by a device, the method of the first aspect when said program is executed by a processor of said device.

According to a fourth aspect of the invention, the invention relates to storage means storing a computer program comprising instructions for implementing, by a device, the method of the first aspect when said program is executed by a processor of said device.

The characteristics of the invention mentioned above, as well as others, will emerge more clearly on reading the following description of an exemplary embodiment, said description being given in conjunction with the accompanying drawings, in which:

- Fig. 1 schematically illustrates an exemplary environment in which the invention may be implemented;

- Fig. 2A schematically illustrates an exemplary IMS image acquisition device included in a viewing system;

- Fig. 2B schematically illustrates an exemplary hardware architecture of a processing module included in a display system;

- Fig. 3 schematically illustrates a mono-component sample image produced by an image sensor of the IMS image acquisition device;

- Fig. 4 schematically illustrates a décamouflage method of an object in a scene according to the invention;

- Fig. 5 illustrates schematically a stealth verification procedure included in the décamouflage method of an object in a scene according to the invention;

- Fig. 6 schematically illustrates a contrast enhancement procedure included in the décamouflage method of an object in a scene according to the invention;

- Fig. 7A schematically illustrates a step of extracting a window in a picture; and,

- Fig. 7B schematically illustrates a mask setting step used in the contrast enhancement procedure.

The invention is described hereafter in the context of a display system comprising a housing incorporating an IMS image acquisition device, a thermal image acquisition device, a processing module and a display device images such as a screen. The invention also applies in a broader context. The invention is particularly applicable when the IMS image acquisition device, the thermal image acquisition device, the image display device and the display system processing module are separate and remote elements , each device may be fixed or mobile and handled by different operators.

Moreover, we note that a human being in an image is considered an object.

Fig. 1 schematically illustrates an exemplary environment in which the invention may be implemented. An operator (not shown) observes a scene 1 comprising a search object 6 (here a helicopter hidden under branches) from a viewing system 5. The display system 5 comprises an acquisition device for thermal images 50 , an IMS image acquisition device 51, a processing module 52 and an image display device 53. The thermal image acquisition device 50 is for example of video capture system and allows acquire a thermal image sequence 3 representative of an optical field 7 with a first frame rate. The image acquisition device 51 is IMS e.g. video capture system type and makes it possible to acquire a sequence of representative IMS four images of the same optical field 7 with a second frame rate. In one embodiment, the first and second frame rates are equal and equal to "25" to "30" frames per second. Each IMS 4 image provided by the image acquisition device 51 is a multi-spectral image which we detail the characteristics in relation to FIG. 3. We detail the image acquisition device IMS 51 in relation to FIG. 2A.

The processing module 52 receives the thermal images 3 and 4 images IMS respectively the thermal image acquisition device 50 and the image acquisition device IMS 51 and applies a treatment that we describe in connection with FIG . 4. We detail the processing module 52 in connection with FIG. 2B. From a pair of images comprising a thermal image 3 and image IMS 4, the processing module 52 produces an image, called image reproduction, wherein the search object 6 is identifiable and provides this image to the device viewing images 53 that the poster. The image display device 53 is for example a screen or eyepiece of the display system 5.

In one embodiment, the first frame rate is lower than the second frame rate. For example, the first frame rate is equal to "15" images per second and the second frame rate is equal to "30" frames per second.

Fig. 2A schematically illustrates an exemplary IMS image acquisition device included in a viewing system.

The IMS image acquisition device 51 receives a light beam 519 that redirects to an image sensor 517 to create a sequence of multi-spectral images 4. To do this the image acquisition device 51 includes a primary lens 512, a field stop 518, a second lens 513, a filter array 514 and an array of mini lenses 516. the primary lens 512, the field stop 518, secondary lens 513, the filter array 514, the matrix of mini lens 516, and the image sensor 517 are perpendicular to an optical axis 511. the primary lens assembly 512, diaphragm 518, secondary lens 513 generates a collimated light beam from the light beam 519. the light beam 519 is representative of the optical field 7 having a small angle of the order of "2.5 °" equally distributed around the optical axis 511. in the example of FIG. 2 A, there is a zoom ratio of "2" between the primary lens 512 and secondary lens 513 so as to obtain a magnification information from the two optical field 519. The collimated light beam is received by the filter array 514 514. the filter matrix is ​​composed of a plurality of filters decomposing the light beam 519 into a plurality of spectral bands. For example, the 514 filter array includes six filters capable of splitting the light beam into six spectral bands. Each of the six spectral band is in the visible range and / or in the field of near infrared and / or infrared wavelength waves. For example, six spectral bands are in a spectral band ranging from "0.4" to "1" μπι or "0.6" to "1" μπι, or "0.9" to "2.5" μτη . In one embodiment, three of the six spectral bands are in the visible range so as to capture the three primary colors red, green and blue, other spectral bands being located in the near infrared and / or infrared lengths shortwave. A plurality of sub-beams

light 515 is generated at the output of the filter array 514 each corresponding to one of the spectral bands of the plurality of spectral bands. In the example described in relation to FIG. 2A, six spectral bands are generated. Each sub light beams of the plurality of sub light beam 515 is then directed to an area of ​​the image sensor 517 by a lens mini mini lens array 516. The mini lens array 516 thus includes as many mini lenses of spectral bands generated by the 514 filter array (ie mini six lenses). The image sensor 517 is for example a CCD ( "Charge-Coupled Device" in English terminology, charge-coupled device) or a CMOS sensor ( "Complementarity metal-oxide-semiconductor" in English terminology , semiconductor complementary metal oxide) comprising an array of photosites capable of converting incident light photons into an electrical signal. Sampling the electrical signal at the second frame rate is used to form a pixel for each photosite. In one embodiment, the image sensor 517 is a matrix (3x500) x (2x500) photosites capable of producing images comprising (3 x 500) x (2x500) pixels. The image from the image sensor 517 is an image, said mono-component having a component, ie each pixel of the image has a component.

Fig. 3 schematically illustrates a mono-component sample image produced by an image sensor 517 of the acquisition device 51 images IMS.

The image mono-component takes the form of a matrix of thumbnails 31 to 36. Each thumbnail result from a focus on the image sensor 517 by a lens mini mini lens 516 of a sub beam array light of the plurality of sub light beams 515 provided by the filter array 514. Each thumbnail 31-36 corresponds to a spectral band of the plurality of spectral bands and is representative of the optical field 7. depending on properties of the object 6 sought, the search object 6 may be visible in six to zero spectral bands, that is, in zero to six thumbnails 31 to 36. in the example described in relation to FIG. 3, the search object 6 is displayed in the thumbnail image 31 and the thumbnail image 36. By against, the search object 6 is inconspicuous or invisible in the thumbnail images 32, 33, 34 and 35. The thumbnails matrix comprises three columns of two thumbnails size of 500x500 pixels.

In one embodiment, the image acquisition device IMS 51 includes a processing unit retrieving the mono-component image captured by the

image sensor 517 and converting this image into a picture IMS IMS 4 4. The image thus obtained has a number of pixels equal to the number of mono-component image pixels divided by the number of spectral bands provided by the matrix 514. Each pixel of the IMS 4 picture filters has a number of components equal to the number of spectral bands provided by the matrix of filters 514. in the example of FIG. 3, the IMS 4 image is of size 500x500 pixels where each pixel has six components. It is assumed here that the thumbnails 31 to 36 of the matrix of thumbnails are aligned spatially, The. thumbnails are recalibrated together so that all the pixels located at the same spatial position in the thumbnail 31-36 correspond to the same spatial position in the scene 1. In addition, it is noted that each component of a pixel the multi-spectral image 3 corresponds to a same time instant as all thumbnails provided having a component were acquired by the same image sensor 517 at the same time. In other words, the thumbnails of the thumbnail matrix are aligned in time.

Each image produced by the image acquisition device IMS 51 is provided to the processing module 52.

The thermal image acquisition device 50 is of type thermal camera and comprises for example an uncooled infrared sensor. In one embodiment, each thermal image 3 provided by the thermal image acquisition device 50 is identical in size to the image IMS 4. The thermal images are mono-component images representative of a spectral band within the field of mid-infrared or infrared long wavelengths. In one embodiment, the thermal images are representative of a spectral band between "3" and "5" μπι or between "8" and "12" μπι or between "7" and "14" μπι.

Each image produced by the thermal image acquisition device 50 is provided to the processing module 52.

The processing module 52 uses image pairs comprising a thermal image 3 and image IMS 4 wherein the thermal images 3 and 4 are aligned IMS spatially and temporally. If the image acquisition device IMS 51 and the thermal image acquisition device 50 do not directly generate harmonized image spatially and temporally, ie if there is no calibration (relative or absolute) between the IMS image acquisition device 51 and the device for acquiring thermal image 50, the processing module 52 generates from the IMS 4 and thermal images 3 provided respectively by the image acquisition device IMS 51 and the thermal image acquisition device 50, image pairs comprising thermal images 3 and 4 harmonized IMS.

In one embodiment, the thermal images 3 provided by the 50 thermal image acquisition device are of larger dimensions (respectively less) images IMS 4 provided by the IMS image acquisition device 51. In this case prior to their use by the processing module 52, a spatial alignment is applied between the thermal image 3 and image IMS 4 so as to map each pixel in the thermal image 3 with a pixel of the image IMS 4, The. there is a bijective relationship between the pixels of the thermal image 3 and the image pixels IMS 4. To do this, each thermal image 3 is subsampled (respectively interpolated) by the processing module 52 to the dimensions of the image IMS 4. in this manner, the thermal images and the IMS images corresponding to a same time used by the processing module 52 are aligned spatially.

In one embodiment, when the first frame rate is lower than the second frame rate, the thermal images 3 are temporally interpolated by the processing module 52 to reach the second frame rate. Temporal interpolation may for example consist of a repeating image. In this manner, the thermal images 3 and 4 IMS images are temporally aligned.

Fig. 2B schematically illustrates an exemplary hardware architecture of a processing module included in a display system.

According to the exemplary hardware architecture shown in Fig. 2B, the processing module 52 then comprises, connected by a communication bus 520: a processor or CPU ( "Central Processing Unit") 521; a RAM ( "Random Access Memory" in English) 522; a ROM ( "Read Only Memory" in English) 523; a storage unit such as a hard disk or storage media player, such as an SD card reader ( "Secure Digital" in English) 524; at least one communication interface 525 to the processing module 52 to communicate with the thermal image acquisition device 50, the image acquisition device IMS 51 and / or the image display device 53.

In one embodiment wherein the acquisition device of thermal images 50, the image acquisition device IMS 51, the processing module 52 and the display device 53 are separate and apart, the acquisition device thermal image 50, the image acquisition device IMS 51 and the display device 53 also includes a communication interface capable of communicating with the communication interface 525 via a network such as a network without son.

The processor 521 is capable of executing instructions loaded into RAM

522 from the ROM 523, an external memory (not shown), a storage medium (such as an SD card), or a communication network. When the processing module 52 is turned on, the processor 521 is capable of reading the RAM 522 instructions and execute them. These instructions form a computer program causing the implementation, by the processor 521, all or part of the process described below in connection with Figs. 4, 5 and 6.

The method described below in relation to FIGS. 4, 5 and 6 can be implemented as software by executing a set of instructions for a programmable machine, for example a DSP ( "Digital Signal Processor" in English) or a microcontroller, or be implemented in hardware by a machine or a dedicated component, such as an FPGA ( "Field-Programmable Gate Array" in English) or an ASIC ( "application-specific integrated circuit" in English).

Fig. 4 schematically illustrates a décamouflage method of an object in a scene according to the invention.

One objective of the method described in relation with Fig. 4, is to provide an operator watching the image display device 53, an image in which pixels corresponding to the object 6 can be clearly distinguished with respect to a background image, the background in an image being considered here as any pixel of said image not corresponding to the object 6. in addition, the method allows to highlight the internal contrasts to the object 6. for this purpose, said method is based on two successive implementations of a contrast enhancement procedure.

In a step 40, the processing module 52 obtains a pair of images comprising an image IMS 4 and a thermal image 3. The thermal image 3 and 4 IMS said torque image are spatially and temporally aligned, ie

each component of the multi-spectral image 4 and each component of the thermal image 3 are aligned spatially and temporally therebetween.

In a step 41, the processing module 52 obtains a position of a sub part of an image, called window thereafter. In one embodiment, the position, shape and size of the window defined by an operator using a control device connected to the processing module 52.

In one embodiment, the shape and size of the window are adapted to the shape and size of the search object 6.

In one embodiment, the operator defines a square window hundred pixels per side.

For each position obtained, the processing module 52 implements steps 42, 45, 46 and 47. Optionally, the processing module 52 implements steps 43 and 44 between steps 42 and 45.

In step 42, the processing module 52 extracts a window of each of the 4 images IMS thermal and 3 to said position. Fig. 7A schematically illustrates a step of extracting a window in an image. Fig. 7A takes the example of the image IMS 4 in which is positioned a window 300 including the search object 6.

Each extracted window is used thereafter by the processing module 52 to provide to an operator an image in which a contrast of pixels belonging to the search object 6 and pixels not belonging to the specific item 6 is highlighted .

In step 45, the processing module 52 applies a contrast enhancement process to at least one of the extracted windows. contrast enhancement procedure, when applied to a window, allows to obtain a window, said improved window, wherein a contrast between pixels corresponding to the object and pixels not corresponding to the object is accentuated. In step 45, the contrast enhancement procedure is systematically applied to the extracted image window IMS 4. In one embodiment, during step 45, the contrast enhancement procedure is also applied to the extracted window of the thermal image 3.

Fig. 6 schematically illustrates a contrast enhancement procedure included in the décamouflage method of an object in a scene according to the invention.

At step 450, the processing module 52 obtains a position of a pixel mask adapted to contain the pixels of the window 300 corresponding to the search object 6, called mask target T. In one embodiment, the position of the target mask T is preset in the window 300. in another embodiment, the position of T in the mask window 300 is defined by an operator. Knowing the characteristics of the search object 6, it is possible to adapt the shape and / or size of the target mask T to the shape and size of the search object 6. In one embodiment, the target mask T is square and the size of the target mask T depends on the size of the search object 6. in one embodiment, three target T masks are available to the processing module 52, a square mask of pixels of three side, a mask square of five pixels per side and a square mask of seven pixels per side. The processing module 52 then selects the target mask T the smallest may integrally contain the search object 6.

In step 451, the processing module 52 sets the target mask T at position 300 obtained in the window.

In step 452, the processing module 52 defines a pixel mask corresponding to the background in the window 300 (ie a pixel mask does not match the search object 6), called background mask B. In one mode of embodiment, the background mask B is a complementary mask the target mask T, ie all the pixels of the window 300 that do not belong to the target mask T belong to the background mask B.

In one embodiment, a G region corresponding to a band of a few pixels around the target mask T separates the target mask T mask from the bottom B. The G box to avoid taking into account, when a contrast enhancement , poorly defined pixels, ie the pixels that can not be clearly identified as belonging to the object 6 or the bottom. The background mask B corresponds to all the pixels of the window 300 not belonging to the target mask T nor G. FIG area. 7B schematically illustrates a target mask setting step T and bottom B used in a contrast enhancement procedure. A rectangular mask target T is placed in the window 300. The target T mask is surrounded by a background area G. A mask B corresponds to all the pixels of the window 300 not belonging to the target mask T, or to the area G.

In step 453, the processing module 52 applies a Fisher projection of the pixels of the window 300. An application of a method of projection

Fisher described in the article "some pracîical resulting in anomaly detection and exploitation of regions of interest in Hyperspectral Images" Goudail F. et al., Applied Optics, Vols. 45, No. 21, pp. 5223-5236 is used. Fisher the application process of a projection makes it possible to accentuate the contrast between the pixels in the target mask T and the pixels belonging to the background mask B. The method includes projecting each pixel in the window 300 in an optimal direction a one-dimensional or multi-dimensional space where each dimension of the space corresponds to a component of the window 300 on which is applied the Fisher projection. In the example of the window 300 extracted from the IMS image 4 described above, each pixel of the image IMS 4 includes six components each representative of an intensity value within a spectral band. The space is then a multidimensional six-dimensional space. It is assumed here that the values of each component of each pixel corresponding to T mask (respectively the bottom mask B), are random variables, spatially uncorrelated and having a Gaussian probability density function with mean m T (m respectively B ) and covariance matrix Γ. The article cited above mentions of estimating the mean m processes T (respectively m B ) and the covariance matrix Γ.

The optimal projection direction, represented by a vector u, can be determined from the covariance matrix Γ according to the following formula:

u = Γ 1 (mT— mB)

wherein m T (respectively m?) is a representative average of the pixels corresponding to mask T (respectively to the mask B).

where ni [(respectively m B ), with k G [1; K] is a value of a component of the average pixel m T (m respectively?) In a spectral band k, and K is the number of components of a pixel (here K = 6 for the extracted image window IMS

where f c (i) (p respectively B (i)) is a value of a k-th component of z ' th pixel p T (i) corresponding to the target mask T (respectively the bottom mask

B), and N is a number of pixels corresponding to the target mask T (respectively the bottom mask B).

The projection according to the vector u is the projection of Fisher and returns to search for a maximum correlation between variations in component values.

Each pixel p (i) of the window 300 is projected according to the Fisher projection:

u where t is the transpose of the vector u, f (i) is a pixel of an improved window (also called Fisher projection window F) corresponding to a result of an application of the projection on the Fisher 300 window. the improved window is a form of mono-component and window size identical to the window 300.

We considered until then that all spectral bands of the plurality of spectral bands were considered for the projection of Fisher. In one embodiment, the Fisher projection takes into account for each pixel a subset of the components of said pixel, ie a subset of the spectral bands of the plurality of spectral bands. For example, the Fisher projection could take into account that two or three spectral bands in which the contrast between the pixels of the target T and the mask background mask of the pixel B is the highest. The contrast in a spectral band can be defined as follows:

r _ (ml - m*)2

k {σξ)2

which is a standard deviation of the component values corresponding to the spectral band k pixels corresponding to mask B. Fisher projection takes into account the two or three spectral bands associated with the contrast values C k highest.

Back in FIG. 4, in step 46, the processing module 52 forms a multi-component window. Each improved window obtained by the implementation of the contrast enhancement procedure and each window extracted on which has not been used the contrast enhancement procedure provides at least one component of the multicomponent window. For example, when the contrast enhancement procedure was applied only to the window extracted from the image IMS 4, the multi-component window includes a component corresponding to the window obtained by the improved contrast enhancement procedure and component corresponding to the window derived from the thermal image 4. in one embodiment, prior to forming the multi-component window, the

processing module 52 performs a scaling of the components of each pixel of each window improved values ​​obtained by the implementation of the contrast enhancement procedure and each extracted window that has not been applied the contrast enhancement procedure. One objective of this scaling is that all windows used to create the multi-window components have pixel component values ​​spread over the same range of values. For example, a scaling is applied to the component value of each pixel of the improved window obtained upon application of the contrast enhancement process to the extracted image window IMS 4 (respectively the component value of each pixel of the window derived from the thermal image 3) so that the component value of each pixel of the improved window (respectively of the window derived from the thermal image 3) is distributed in a range of predetermined values ​​[MIN; MAX]. In one embodiment, MIN and MAX = 0 = 255.

In step 47, the processing module 52 applies the contrast enhancement procedure described in connection with Figs. 6 and 7B to the multi-component thus formed window.

In a step 48, the processing module 52 generates a reproduction image to be displayed by the image display device 53. To do this, the processing module 52 inserts each improved window obtained by applying the sharpening procedure contrast to each multi-component window formed in a representative receiving image of the scene. We call for the return window following an improved window obtained by applying the contrast enhancement procedure in a multi-component window.

To do this, for each rendering window, the processing module 52 retrieves the position of the window 300 obtained in step 41 and sets the playback window in a representative receiving image of the scene 1 in said position. The processing module 52 thus generates an output image in which the pixel values ​​located in a playback window are the pixel values ​​from the projection Fisher applied to the corresponding multicomponent window and the values ​​of pixels outside a playback window are the pixel values ​​of the reception image.

In one embodiment, the receiving image is a thumbnail of the thumbnail matrix.

In one embodiment, the processing module 52 reconstructs a receiving image from a subset of the spectral bands of the plurality of spectral bands. For example, the processing module 52 uses three spectral bands in the visible region corresponding to the three primary colors red, green and blue and creates a representative image receptor is a human visual system would be of the stage 1.

The playback image is then displayed to an operator through the display device 53.

In one embodiment, said automatic mode, it is not an operator who defines the position of the window 300 and the position of the target mask T. Several positions of the window 300 are tested successively by the processing module 52. For example, the window 300 is moved in the image IMS 4 (respectively in the thermal image 3) so that each pixel of the image IMS 4 appears at least once in the window 300. for each position of the window 300 tested, the processing module 52 implements the steps 42, 45, 46 and 47. in this embodiment, in step 451, the target mask T is defined automatically so that it is positioned at center of the window 300. Further to the implementations of the stages 42, 45, 46 and 47, the processing module 52 selects at least one restitution windows obtained and applies step 48 to return each selected window. For example, the processing module 52 selects the playback window to display the strongest contrast between pixels corresponding to the target T and the mask pixels corresponding to mask background B. In this case, it is considered that the playback window to display the strongest contrast between pixels corresponding to the target T and the mask pixels corresponding to the background mask B provides a good reproduction of image.

In one embodiment can be combined with the automatic mode between the steps 42 and 45, the processing module 52 implements the steps 43 and 44. Step 43 which are described in more detail in relation to FIG. 5 for testing whether an object in a window 300 is stealthy in the sense of the human visual system. An object is not in stealth way the human visual system if it is clear in at least one spectral band within the visible range. It is not necessary to try to improve the visualization of an object, if this item is not stealth, that is to say, clearly visible and identifiable in a scene. When in step 43, an object is considered not stealth by the processing module 52, the processing module 52 implements step 44 during which it terminates the implementation of the method décamouflage of an object to the position of the window 300 obtained in step 41. Otherwise, if the object is considered stealth by the processing module 52, the processing module 52 continues to implement the method décamouflage an object 45 with the previously explained step.

In this embodiment, we consider that the plurality of spectral bands comprises three spectral bands in the visible range and corresponding to the three primary colors red, green and blue. The image acquisition device IMS 51 is therefore able to provide spectral bands that provide a DCV sensor. The image acquisition device IMS 51 thus acts as a device comprising an image acquisition device capable of providing DCV images and an image acquisition device adapted to acquire spectral bands in the near infrared and / or infrared wavelength waves. In one embodiment, the image acquisition device IMS 51 is replaced by a device comprising an image acquisition device capable of providing DCV images and an image acquisition device adapted to acquire bands spectral located in the near infrared and / or infrared wavelength waves.

Fig. 5 illustrates schematically a stealth verification procedure included in the décamouflage method of an object in a scene according to the invention corresponds to the optional step 43.

In a step 431, the processing module 52 applies the contrast enhancement procedure described in connection with Fig. 6 to the window 300 extracted from the image IMS 4 taking into account at least one of the three components corresponding to the spectral bands in the visible range, that is to say at least one of the spectral bands corresponding to the three primary colors red, green and blue.

In a step 432, the processing module 52 calculates a contrast value

C between the pixels corresponding to mask pixels T and B corresponding to the mask of the improved window obtained following the implementation of the contrast enhancement procedure in step 431.

_ (mT - mB )2

c = M2

wherein m T (respectively m B ) is an average value of pixels corresponding to mask T (B respectively), and σ beta is a standard deviation of the pixels corresponding to mask B.

In a step 433, the processing module 52 determines whether the window 300 extracted from the IMS 4 image comprises a non-stealth object. To do this, the processing module 52 compares the contrast value C at a C predefined threshold contrast value s (e.g. C s = 2.3). When C> C s , the processing module 52 considers that the window 300 includes a non-stealth object. In this case, step 433 is followed by step 44. When C≤ C s , the processing module 52 considers that the window 300 does not include a non-stealth object. In this case, step 433 is followed by step 45.

In one embodiment, the thermal image 3 is a multi-component image. For example, the thermal image 3 comprises a component located in the infrared at medium wavelengths (MWIR) and a component located in the infrared at long wavelengths (LWIR). In this embodiment, the stealth verification process corresponding to the optional step 43 described in connection with Fig. 5 is made on the window derived from the thermal image 3 in step 42.

In this case, in step 431, the processing module 52 applies the contrast enhancement procedure described in connection with Fig. 6 to the window derived from the thermal image 3 by taking into account each component of the thermal image 3.

In step 432, the processing module 52 calculates a contrast value

C between the pixels corresponding to mask pixels T and B corresponding to the mask of the improved window obtained following the implementation of the contrast enhancement procedure in step 431.

In step 433, the processing module 52 determines whether the window derived from the thermal image 3 comprises a non-stealth object. To do this, the processing module 52 compares the contrast value C to the contrast value C predefined threshold s . When C> C s , the processing module 52 considers that the window derived from the thermal image 3 comprises a non-stealth object. In this case, step 433 is followed by step 44. When C≤ C s , the processing module 52 considers that the window derived from the thermal image 3 does not include a non-stealth object. In this case, step 433 is followed by step 45.

It is noted that the two embodiments of FIG. 5 may be combined so that a verification of the stealth of an object is done on least one of the spectral bands corresponding to the three primary colors red, green and blue and / or components of the thermal image 3.

In another embodiment of step 43, the processing module 52 defines a T mask and a mask B directly in each of the components of the image IMS 4 and each of the components of the thermal image 3 then calculates a value contrast C between the pixels corresponding to mask pixels T and corresponding to the mask B independently for each of the components of the image IMS 4 and each of the components of the thermal image 3. If, for at least one of said components C> C s , step 43 is followed by step 44. Otherwise, step 43 is followed by step 45.

We have seen that, in one embodiment, the contrast enhancement procedure is applied to the extracted window of the thermal image 3. In this case it is best to better bring out the details of the search object 6, not to use a T mask adapted to the size and shape of the search object 6, but a T mask having a shape and size adapted to the shape and the details of the size of interest for identify the search object 6. in one embodiment, the mask T used in the contrast enhancement procedure described in connection with Fig. 6 has a shape and size adapted to the shape and size of the details of interest to identify the search object 6. In steps 45 and 47, said contrast enhancement process is implemented for a plurality of positions T mask in the window 300 on which the procedure is applied. The plurality of positions fully cover the shape and size of the object 6.

In step 45, a plurality of windows is improved then obtained for the extracted image window IMS 4 (respectively for the extracted window of the thermal image 3). Improved windows of the plurality of windows obtained for the improved window extracted from the image IMS 4 are combined to form a single window for the improved window extracted from the image IMS 4. Improved windows of the plurality of windows obtained for improved window derived from the thermal image are combined to form a single window for the improved window derived from the thermal image 3. the two windows improved thus obtained are then used in step 46 to form the multi-component window.

In step 47, a plurality of windows is improved obtained for the multi-component window. Improved windows of the plurality of windows obtained for the improved multi-window components are combined to form a single window for the improved multi-component window. The single window improved thus obtained is used in step 48 to generate the output image.

CLAIMS

1) A method of décamouflage of an object in a scene observed by a plurality of devices comprising a device (51) for acquiring images, called multi-spectral, comprising a plurality of components each representative of a spectral band of in a visible and / or near infrared and / or infrared short wavelengths and means (50) for acquiring images, called thermal, comprising at least one component representative of a spectral band included in the mid-infrared and / or infrared wavelengths, characterized in that the method comprises:

obtaining (40) a multi-spectral image (4) and a thermal image (3), each component of the multi-spectral image (4) and each component of the thermal image (3) being aligned spatially and temporally therebetween ;

obtaining (41) at least a position of a sub-part of an image, said window, and for each position obtained:

• extracting (42) a window of each of the multi-spectral image (4) and thermal (3) to said position;

• Applying (45) a contrast enhancement process to at least one of the extracted window comprising a window derived from the multi-spectral image (4), said procedure, when applied to a window for obtaining a window, said improved window, wherein a contrast between pixels corresponding to the object and pixels not corresponding to the object is increased;

• forming (46) a multi-component window, each window obtained improved and extracted each window that has not been applied said procedure providing at least one component of the multicomponent window; and,

• applying (47) said procedure in the multi-component window;

generating (48) an image, called image reproduction, by inserting each improved window obtained by applying said process to each multi-component window formed in a representative receiving image of the scene.

2) Process according to claim 1, characterized in that the contrast enhancement procedure comprises when applied to a window:

obtaining (450) at least a position of a first mask (T) adapted to contain pixels corresponding to said object in said window and for each position:

positioning (451) said mask to said position in said window;

define (452) a second mask (B) comprising pixels of said window not included in the first mask; and,

applying (453) a Fisher projection to the pixels of said window to provide improved window wherein a contrast between pixels of the first and second mask is increased.

3) Method according to claim 2, characterized in that the first mask (T) is adapted so that each pixel of the object is contained in the first mask (T).

4) Method according to claim 2, characterized in that the first mask (T) is adapted to contain each pixel of a detail of said object of interest to identify said object.

5) Method according to claim 4, characterized in that the method comprises for the window extracted from the multi-spectral image (4) and the window derived from the thermal image (3):

applying (45) the contrast enhancement procedure for a plurality of positions of the first mask in each of said windows, the plurality of positions to fully cover the object;

forming (45) a first single improved window from each improved window obtained during each application of the contrast enhancement procedure in the window extracted from the multi-spectral image (4) and a second single improved window from each improved window obtained during each application of the contrast enhancement procedure in the window derived from the thermal image (3); and,

forming (46) the multi-window components from the first and second unique improved windows.

6) Method according to claim 5, characterized in that the method comprises for the multi-component window formed:

applying (47) the contrast emphasizing process for a plurality of the first mask positions in the multi-window components, the plurality of positions to fully cover the object;

forming (47) a third single improved window from each improved window obtained during each application of the contrast enhancement process to the multi-component window;

using (48) the third single window to generate improved the restitution image.

7) A method according to any one of claims 2 to 6, characterized in that the plurality of components of the multi-spectral image comprises at least one spectral band within the visible region corresponding to a primary color red and / or blue and / or green and in that for each position of said window obtained the method comprises:

applying (431) the contrast enhancement process to the window derived from the multi-spectral image, each component corresponding to a spectral band within the near infrared and / or infrared wavelengths shorter n ' being taken into account;

calculating (432) a contrast value, said visible contrast value between the pixels corresponding to the first mask and the pixels corresponding to the second mask of the improved window obtained following the application of the contrast enhancement procedure; and,

to (433, 44) end to the implementation of décamouflage method of an object to the position of said window obtained when said visible contrast value is greater than a predefined threshold, said threshold visible.

8) A method according to any one of claims 2 to 7, characterized in that the thermal image comprises at least two components and in that for each position of said window obtained the method comprises:

applying (431) the contrast enhancement process to the window derived from the thermal image;

calculating (432) a contrast value, said thermal contrast value between the pixels corresponding to the first mask and the pixels corresponding to the second

mask of the improved window obtained following the application of the contrast enhancement procedure in the window derived from the thermal image; and,

to (433, 44) end to the implementation of décamouflage method of an object to the position of said window obtained when the thermal contrast value is greater than a predefined threshold, said threshold temperature.

9) A method according to any one of the preceding claims, characterized in that the multi-spectral images are representative of spectral bands in a spectral band ranging from "0.4" to "1" μπι or "0.6" to "1" μπι, or "0.9" to "2.5" μτη and thermal images are representative of a spectral band between "3" and "5" μτη or between "8" and "12" μπι.

10) A décamouflage of an object in a scene observed by a plurality of devices comprising a device (51) for acquiring images, called multi-spectral, comprising a plurality of components each representative of a spectral band within a visible and / or near infrared and / or infrared short wavelengths and means (50) for acquiring images, called thermal, comprising at least one component representative of a spectral band of mid-infrared and / or infrared long wavelengths, characterized in that the device comprises:

obtaining means (40) for a multi-spectral image (4) and a thermal image (3), each component of the multi-spectral image (4) and each component of the thermal image (3) being harmonized spatially and temporally therebetween;

obtaining means (41) for obtaining at least a position of a sub-part of an image, said window, and for each position obtained:

• extraction means (42) for extracting a window of each of the multi-spectral image (4) and thermal (3) to said position;

• applying means (45) for applying a contrast enhancement process to the extracted at least one window comprising a window derived from the multi-spectral image (4), said procedure, when applied to a window to obtain a window, said improved window, wherein a contrast between

pixels corresponding to the object and pixels not corresponding to the object is highlighted;

• forming means (46) to form a multi-component window, each window obtained improved and extracted each window that has not been applied said procedure providing at least one component of the multicomponent window; and,

• applying means (47) for applying said process to the multi-component window;

generating means (48) for generating an image by inserting each improved window obtained by applying said process to each multi-component window formed in a representative receiving image of the scene;

11) A computer program, characterized in that it comprises instructions for implementing, by a device (52), the method according to any one of claims 1 to 9, when said program is executed by a processor ( 521) of said device (52).

12) storage means, characterized in that they store a computer program comprising instructions for implementing, by a device (52), the method according to any one of claims 1 to 9, when said program is executed by a processor (521) of said device (52).