EP1362477A1

EP1362477A1 - Method and device for collimating a digital panoramic image

Info

Publication number: EP1362477A1
Application number: EP02704811A
Authority: EP
Inventors: Jean-Claude Artonne; Christophe Moustier
Original assignee: Immervision International Pte Ltd
Current assignee: 6115187 Canada Inc
Priority date: 2001-02-16
Filing date: 2002-02-13
Publication date: 2003-11-19
Also published as: US20040086271A1; FR2821172A1; FR2821172B1; WO2002067572A1; US6885817B2

Abstract

The invention concerns a method for collimating a digital panoramic image (PI1) derived from at least a wide-angle initial image whereof the image points are transferred into a three-dimensional co-ordinate system. The invention is characterised in that the method comprises: a step which consists in inserting a collimating mark on the initial image when the shot is made; and a step which consists in detecting the location (461) of the collimating mark in the digital panoramic image (PI1), using an image analysing algorithm. FIG. 10A: xyz SPHERICAL CO-ORDINATE SYSTEM OF AXES PI1 DIGITAL PANORAMIC IMAGE P1, P2, P12 PARALLEL LATITUDES M1 MERIDIAN

Description

METHOD AND DEVICE FOR ORIENTATION OF AN IMAGE

DIGITAL PANORAMIC

The present invention relates to digital photography and in particular the production of wide angle photographs, as well as the transformation of wide angle photographs into digital panoramic images. The present invention also relates to the presentation of digital panoramic images on a screen, and the virtual visit of places by means of panoramic images.

In recent years, rapid advances in microcomputer and digital camera manufacturing techniques have led to significant development in digital photography and its accessibility to the public.

Among the various applications offered by digital photography, the presentation of 360 ° panoramic images on computer screens is booming because this technique allows for virtual visits to places from a simple computer screen. offering a reduced angle of vision, the observer being able by means of a screen pointer to drag the image presented on the screen to the left, the right, the top or the bottom, until reaching the limits of 1 panoramic image. These panoramic images are generally of a spherical or cylindrical appearance, so that the observer can at least make a complete rotation in the horizontal plane of the image returning to his starting point. Spherical images also allow you to make a complete revolution in the vertical plane. Furthermore, the forecast of hyper-anchor type links between two panoramic images allows the observer to move from one image to another by a simple "click" of the mouse on an active zone present in the image, the zone active generally corresponding to an object present on the image, for example a door, a window, ... Various examples of panoramic images and virtual tours are presented on many websites. We can in particular refer to the site "http://www.panoguide.com"("the guide to panoramas and panoramic photography") which gives a comprehensive overview of all the products available to the public, from photographic equipment up to software enabling 360 ° panoramic images to be formed by assembling wide-angle photographs, correcting the tint of the images, producing active areas making it possible to chain panoramic images, etc. Such software, which implements mathematical algorithms for processing digital images are offered to the public in the form of programs downloadable from the Internet or commercially available CD-ROMs.

To date, these techniques for obtaining digital panoramic images and virtual tour, despite their increasing accessibility to the public and the enthusiasm they arouse, have various drawbacks which will be explained in the following.

Disadvantages relating to photography equipment It is recalled here that obtaining a 360 ° digital panoramic image generally requires the production of at least two 180 ° photographs (or N photographs made with an angle of 360 ° / N ) by means of a panoramic lens and a panoramic head, 360 ° lenses being expensive and having a reduced angle of view in the vertical plane. Such a panoramic head comprises a rotatably mounted part which receives the camera and which comprises means for adjusting the position of the camera, by means of which it is possible to obtain, after various adjustments, the alignment of the nodal plane of the lens and the axis of rotation of the panoramic head, which is essential to avoid parallax errors. However, such alignment is not easy to obtain and requires various adjustments and tests. In addition, panoramic heads are precision instruments at a considerable price. On the other hand, digital cameras of type SLR ("single Lens Reflex") can receive any type of lens but are expensive and not very accessible to the general public, which generally turns to digital cameras of compact type, c is to say fixed objective. To overcome the inconvenience of the irremovability of the lens of compact cameras, some manufacturers offer lenses called "adapters"("conversionlens") among which there are panoramic adapters ("fisheye conversion lens" or "fisheye converters ") and telephoto lens adapters. These adapters can be screwed directly onto the fixed objective ^' of the compact camera, the rear lens of the adapter then being located opposite the front lens of the fixed objective, and allow the holder of a compact camera to take wide angle photographs. Unfortunately, such adapters are not universal and many compact cameras cannot receive them, since they do not have the necessary threads.

Thus, an objective of the present invention is to provide a method and a camera support device making it possible to carry out wide angle photography by means of a digital camera of the compact type, including a compact camera not comprising means of fixing a panoramic adapter.

Another objective of the present invention is to provide a method and a device facilitating the production of wide-angle photographs without parallax error, without requiring the usual and delicate adjustments of the position of the camera aimed at obtaining a good alignment between the camera. axis of rotation of the camera and the nodal plane of the front lens of the panoramic objective. Another object of the present invention is to provide an apparatus support device which is of a simple structure and of a reduced cost price. Disadvantages of color disparities between panoramic images

Another drawback of the abovementioned techniques relates to the correction of the hues of the panoramic images obtained by assembling wide-angle photographs.

Recall here that after having taken at least two wide angle digital photographs, the photographic files delivered by the image sensor of the photographic camera must be transferred to a microcomputer equipped with software executing algorithms of conversion of picture. Such algorithms transfer the image points of each photograph into a three-dimensional coordinate system, of the - spherical, cubic, cylindrical, polyhedral type, etc. After the transfer, two semi-panoramic images are available, for example two images in hemispheres, which are assembled to obtain a total panoramic image, that is to say 360 °.

Although digital cameras perform white balance and brightness correction (gamma correction), the shooting conditions differ depending on whether you are facing the sun or back to the sun and, for photographs taken indoors, depending on the light sources present (neon lights, windows, etc.). Consequently, each of the semi-panoramic images has its own dominant hue, which appears clearly in the final panoramic image, for example in the form of an abrupt variation in hue between the first and the second half-sphere. in the case of a spherical panoramic image. A classic solution to this problem is to recalibrate the colors of a hemisphere by referring to the other hemisphere. Such registration includes a step of determining the gamma of the primary colors of the first hemisphere in the areas of overlap or of joining with the second hemisphere, made with reference to the intensity of the primary colors of the points of the second. hemisphere. The next step is to apply a gamma correction to all points in the first half sphere. This gives a constant dominant shade over the whole of the panoramic image.

Unfortunately, such a hue correction has only a relative value and the problem of disparity of hues reappears when comparing two panoramic images, each image having a general hue which, although being homogeneous thanks to the aforementioned process, is different from the following image. This problem clearly appears when several panoramic images are chained as part of a virtual visit to a place, and results in strong variations in hue when the observer moves from one panoramic image to another.

Thus, ^• yet another objective of the present invention is to provide a means and a method of hue correction making it possible to homogenize the hue of several digital panoramic images.

Disadvantages of Orienting Panoramic Images During a Virtual Tour

Another problem of the aforementioned techniques, appearing within the framework of a virtual visit, is that the observer is subject to a phenomenon of disorientation during a transition from one panoramic image to another, because he is deprived of '' a common reference between the different panoramas. This phenomenon is particularly sensitive in the context of a virtual visit to a place comprising several adjoining rooms, each represented by one or more panoramic images. Consider for example three adjoining rooms each comprising an access door to each of the other two rooms, and three panoramic images representing each of the rooms respectively and each comprising two active areas defined in the regions corresponding to the doors. The problem which arises is to define the portion of panoramic image to be displayed on the screen when the observer enters a panoramic image. A known solution consists in defining a default orientation angle which is constant whatever the entry point into the panoramic image. Using the example cited above, this means that the portion of room presented on the screen is constant whatever the door by which one entered there. It is therefore obvious that this solution has the drawback of confusing the observer. Another known solution consists in defining several default orientation angles, chosen dynamically as a function of the point of entry into the panoramic image, that is to say as a function of the active area selected in the image. previous pan. This solution has the drawback of being complex to implement. It requires the establishment of a cartography of the places and the determination of an angle of orientation for each planned link between 'two images.

Thus, yet another objective of the present invention is to provide a means and a method for orienting a digital panoramic image.

Yet another objective of the present invention is to provide a method for displaying a digital panoramic image in which the orientation of the image is determined dynamically without the need to chain the various panoramic images.

At least one objective of the present invention is achieved by providing a support device for a camera, comprising means for fixing a camera and means for rotation about an axis, characterized in that it comprises an orientation means arranged so as to appear in a shot when a camera equipped with a panoramic lens is fixed on the support device. According to one embodiment, the device comprises a compass having a round-shaped dial arranged concentrically with the axis of rotation of the support device.

According to ^• an embodiment, the device comprises a compass comprising a dial presenting an artificial color intended to be distinguished from the natural colors of a shot during an image analysis by computer aiming to find the location of compass. According to one embodiment, the compass comprises an orientation mark having a determined color offering a high contrast relative to the color of the dial, in order to be distinguished from the dial during an image analysis by computer aiming to find the location of the orientation marker.

According to one embodiment, the orientation marker is a needle comprising on a first north or south half a color having a high contrast relative to the color of the dial and on a second south or north half a color substantially identical to the color of the dial.

According to one embodiment, the compass has a central part of dark color, so that the dial of the compass forms a colored ring whose shape facilitates the detection of the dial during an image analysis by computer aimed at find the location of the dial.

According to one embodiment, the device comprises means for locating an arbitrary orientation the position of which can be adjusted manually, and means for fixing the locating means arranged so that the locating means can be arranged in a visible position where it appears in a shot when a panoramic lens is associated with the camera. According to one embodiment, the locating means is a tongue which extends above a compass dial and has a color offering a high contrast relative to the color of the dial in order to facilitate the detection of the tongue during computer image analysis to find the location of the tracking means.

According to one embodiment, the locating means is rotatably mounted around an axis coaxial with the axis of rotation of the support device. According to one embodiment, the device comprises a color calibration piece comprising at least three primary colors, and means for fixing the color calibration piece arranged so that the piece color calibration appears in a shot when a camera equipped with a panoramic lens is arranged on the support device.

According to one embodiment, the color calibration part is circular and concentric with the axis of rotation of the device.

According to one embodiment, the device comprises means for fixing a photographic camera equipped or capable of being equipped with a first lens, and means for fixing an adapter lens, in particular a panoramic adapter lens, arranged to maintain the adapter lens aligned with the first lens without the need to attach the adapter lens to the camera. According to one embodiment, the means for fixing the adapter lens comprise a moving part and means for fixing the adapter lens to the moving part, the moving part being slidably mounted and pushed by an elastic means in a corresponding direction at the location of the front lens of the first objective.

According to one embodiment, the device comprises an optical guide for bringing a front light to a photosensitive cell of the camera.

According to one embodiment, the means for fixing the adapter lens are arranged to maintain the adapter lens in a position such that the axis of rotation of the support device is substantially in the nodal plane of the front lens of the lens adapter.

The present invention also relates to a method for orienting a digital panoramic image obtained by transferring the image points of at least one initial image into a three-dimensional coordinate system, comprising a step of inserting a marker of orientation on the initial image at the time of shooting, and a step of detecting the location of the orientation marker in the digital panoramic image, made by means of an image analysis algorithm. According to one embodiment, the inscription of an orientation mark on the initial image comprises the fact of arranging a compass in the field of view of a panoramic lens, so that the compass appears on 1 'initial image.

According to one embodiment, the step of detecting the location of the orientation mark in the digital panoramic image comprises a step of searching for the location of the dial of the compass. According to one embodiment, the step of detecting the orientation mark in the digital panoramic image comprises a step of searching for the location of a hand in a dial area.

According to one embodiment, the image points of the initial image are transferred into a spherical coordinate system to form a spherical panoramic image, the compass dial occupies a sector of a sphere in the spherical panoramic image, and the step Finding the location of the dial in the spherical panoramic image includes an angular scan of the sphere following an oblique line or perpendicular to the sphere sector.

According to one embodiment, the inscription of an orientation mark on the initial image comprises the fact of placing a colored tab in the field of view of a panoramic lens, so that the tab appears on the 'initial image.

According to one embodiment, the tab is arranged above an artificial color zone intended to be distinguished from the natural colors of a shot, the step of finding the orientation mark in the image comprising a step of finding the location of the artificial color area and a step of detecting a tongue color in the artificial color area. According to one embodiment, the artificial color zone on which the tongue is arranged is the dial of a compass, and the step of searching for the orientation mark in the image comprises a step of searching for the tab in the colored area of the dial and, if the tab is not found, a step to search for a compass needle.

According to one embodiment, the initial image is a photograph.

According to one embodiment, the initial image is delivered by a video camera.

The present invention also relates to a method of interactive display on a screen of a digital panoramic image obtained by transferring the image points of at least one initial image into a three-dimensional coordinate system, comprising the operations consisting of displaying on the screen an initial image sector and to drag the displayed image sector as a function of an interactivity signal, method comprising a preliminary step of orientation of the panoramic image produced in accordance with the orientation method described herein above, the initial image sector displayed on the screen being determined with reference to the orientation mark detected. According to one embodiment, the initial image sector displayed on the screen is a sector of the panoramic image comprising a set of points whose coordinates in the three-dimensional coordinate system have an angle between a lower limit and an upper limit relative to a reference angle as a function of the orientation mark detected.

According to one embodiment, the initial image sector displayed on the screen is determined so that at least one image point of the initial image sector has an angle relative to the orientation mark equal to a determined reference angle depending on the orientation mark detected.

According to one embodiment, the image point having an angle equal to a reference angle relative to the orientation frame is displayed in the center of the screen.

According to one embodiment, the reference angle is determined during a step of displaying a previous panoramic image, when a user activates in the preceding panoramic image a selection zone of a following panoramic image, by calculating the angle that the selection zone presents relative to the orientation mark.

According to one embodiment, the reference angle is chosen equal to the angle presented by the selection area relative to the orientation mark.

According to one embodiment, the initial image is a photograph.

According to one embodiment, the initial image is delivered by a video camera.

These objects, characteristics and advantages as well as others of the present invention will be explained in more detail in the following description of a camera support device according to the invention, of a method according to the invention of orientation of a digital panoramic image, of a method according to the invention for correcting the hue of a digital panoramic image, and of a method according to the invention of displaying on a screen a digital panoramic image, made without limitation in relation to the attached figures, among which:

FIG. 1 is a sectional view of an embodiment of a camera support device according to the invention,

FIG. 2 is a top view of the support device of FIG. 1,

- Figure 3 is a front view of the support device of Figure 1, Figure 4 is a perspective view of the support device of Figure 1, - Figure 5 is an exploded view of an element of the support device of the figure 1,

FIG. 6 is a close-up view of the support device showing a compass and a color calibration piece,

FIG. 7 is a sectional view of the compass and of the color calibration part,

FIG. 8 is an example of wide angle photography produced by means of a photographic device arranged on a support device according to the invention, FIG. 9 is a flowchart describing steps for obtaining a digital panoramic image, for orienting the panoramic image and for correcting the hue of the digital panoramic image, FIGS. 10A and 10B schematically represent an image digital panoramic of spherical type and respectively illustrate a step of the image orientation method according to the invention and a step of the hue correction method according to the invention, - Figures 11A and 11B schematically represent an enclosed place and illustrate a method according to the invention for displaying panoramic images,

- Figure 12 is a flowchart describing an embodiment of the display method according to the invention, and - Figure 13 shows a video surveillance system and illustrates an application of the present invention.

I - Description of a camera support device according to the invention a - Main aspects of the support device Figures 1 to 4 represent respectively in a sectional view, a top view, a front view and a perspective view an exemplary embodiment of a device 20 according to the invention, intended to serve as a support for a compact photographic camera and shown here with such a camera.

The support device 20 comprises a body 21 rotatably mounted on a base 1 fixed on a tripod 2. The body 21 is produced here by welding or gluing two molded plastic shells, the assembly line 22 of the two parts appearing in FIG. 4. With reference to FIG. 1, the rear part of the body 21 has a housing 23 receiving here. a digital camera 10 of compact type, comprising a non-removable lens 11 or "fixed" lens. At the front of the housing 23 is a cylindrical cavity 24 receiving the objective 11 and opening onto another cylindrical cavity 25 of larger diameter, which opens at the front of the body 21. The camera 10 is locked in the housing 23 by means of a pin 12 screwed into a fixing orifice provided on the underside of the device 10, such an orifice being in itself conventional. The rotation of the device 20 on the base 1 is ensured by a tubular piece 3A integral with the base 1 and oriented upwards, receiving a cylindrical part 3B formed in the lower part of the body 21. The lower part of the body 21, which s extends around the tubular part 3A, has a lower face substantially parallel to the base 1 provided with a ball 26 mounted captively in a cavity, the ball 26 being pushed by a spring against the base 1. The ball 26 cooperates with a cavity 4 formed on the base 1, the assembly forming a system for blocking the body 21 in ^' a determined angular position around the axis 3B. At least two cavities 4 are formed on the base 1 on either side of the axis of rotation 3B to allow the body 21 to be locked in two angular positions offset by 180 °, for the production of two complementary panoramic photographs allowing to obtain, after scanning and assembling the photographs, a 360 ° digital panoramic image.

Furthermore, the cylindrical cavity 25 formed in the front part of the body 21 receives a part 27 allowing the attachment of a lens 15 of the panoramic converter lens type. Such a panoramic adapter 15 is provided to cooperate with the fixed objective 11 of the camera to form an optical group offering a shooting angle of the order of 360 °, preferably substantially greater than 360 ° and of the 'order of 363 °. As shown in the various figures, the fixing piece 27 keeps the panoramic adapter 15 facing the fixed lens 11 and in alignment with it without it being necessary to fix it to the camera .

The fixing part 27 is slidably mounted in the cavity 25 and is pushed by springs 27A, 27B in the direction of the fixed objective 11. As appears in the exploded view of FIG. 5, the part 27 is here a hollow cylinder forming a sheath in which the adapter panoramic 15, of corresponding shape, is arranged. The bottom of the part 27, located opposite the front lens of the fixed objective 11, has a wall in which an opening 28 has been made ensuring the passage of light between the panoramic adapter 15 and the fixed objective 11. The orifice 28 is surrounded by an annular piece 29 of small diameter, for example made of felt or rubber, fixed to the rear face of the wall. The annular part 29 comes into contact with the peripheral part of the fixed objective 10, which is made of plastic, and plays the role of shock absorber and spacer. Thus, when the panoramic adapter 15 is engaged at the bottom of the part 27 and the part 27 is pressed against the objective 11 by the springs 27A, 27B, the rear lens of the panoramic adapter 15 does not come into contact with the front lens of the objective 11, which avoids scratching the two lenses.

The room 27 and the panoramic adapter 15 are provided with a universal type locking system, here a bayonet system, making it possible to arrange in the room 27 other types of adapter, for example a tele-adapter

("tele converter lens"). Thus, we see in Figure 5 that the bottom of the part 27 has three openings 30A, 30B, 30C provided to receive three parts 31A, 31B, 31C forming hooks integral with the rear face of the panoramic adapter 15, the locking is operating conventionally by insertion and rotation of the panoramic adapter 15 in the part 27. The support device further comprises a part 32 forming a lever, one end of which cooperates with a notch 33 formed on an edge of the panoramic adapter 1, by via a slot in an edge of the workpiece ^'27. in figure 1, it appears that the workpiece 32 is held in a locking position by a spring 34 and is pivotable into a release position of the panoramic adapter 15 by action on a push button 35. The body 21 of the support device also includes a light pipe 36 provided with an optical fiber, opening onto the front face of the body 21 and making it possible to bring a frontal light up to a photocell 13 of the camera 10.

According to an optional but advantageous characteristic of the present invention, the nodal plane of the front lens 16 of the panoramic adapter 15 is naturally in alignment with the axis of rotation 3B of the device when the adapter 15 is locked in the piece 27 and that the latter is in abutment against the objective 11. Here we designate by "nodal plane" a plane comprising the nodal points of the lens, the alignment of which with the axis of rotation must be ensured to avoid errors parallax, as is well known to those skilled in the art. In practice, this result is obtained by an arrangement of the axis of rotation 3B at the front of the body 21, taking into account, when designing the body 21, the length of the panoramic adapter 15 and the length of the fixed lens 11.

As a result, the support device 20 according to the invention, associated with the panoramic adapter 15 and with a digital camera of the compact type, can be used by inexperienced people to take panoramic photographs, without adjustment of alignment or tests to detect parallax errors. This also results in a low cost price of the support device according to the invention, which does not have the expensive graduated mechanisms that are found on conventional panoramic heads. -Therefore, the assembly formed by the panoramic adapter and the support device can be marketed in the form of a kit with a reduced sale price, accessible to the majority of the public.

In addition, the support device according to the invention can be adapted to any type of compact digital device, including compact devices which are not designed to receive a panoramic adapter. The low cost price of the device according to the invention makes it possible to provide a different body 21 for each type of compact device present on the market, while retaining a panoramic adapter 15 common to all the embodiments. The support device which has just been described is of course capable of various variants within the reach of ordinary skill in the art, which can relate to most of the particular characteristics of the embodiment which has just been described, all remaining within the scope of the present invention. In particular, an alternative embodiment making it possible to further reduce the cost price of the device consists in fixing the panoramic adapter 15 definitively on the body 21, without providing for the part 27. An embodiment exclusively dedicated to large-scale photography angle can thus be provided. In such an embodiment, the panoramic adapter is slidably mounted and is pushed by a spring system into a rear stop position where it is opposite and in alignment with the fixed objective of the camera. photographic.

On the other hand, the stop position of the panoramic adapter can be obtained in various ways, other than contact with the fixed objective, in particular by means of a fixed stop at the bottom of the housing receiving the panoramic adapter.

It will be noted here that certain compact digital cameras have fixed but motorized lenses, the term "fixed" here designating the irremovable nature of the lens. As such motorized lenses are likely to advance during a focal length adjustment, the provision of an elastic means ensuring flexible contact between the panoramic adapter and the fixed lens makes it possible to avoid any deterioration of the lens or the adapter in case of unexpected movement towards the front of the fixed lens. On other compact cameras, the motorization ensuring the adjustment of the focal distance is applied to the lens system which moves inside the fixed objective, the external length of which thus remains constant.

The support device 20 shown in Figures 1 to 4 has other characteristics which will be described in the following. These additional features are in themselves independent of the previous ones and are thus capable of being applied to other camera supports, in particular conventional panoramic heads. These additional characteristics are provided in relation to aspects of the present invention relating to the processing of a digital image, in particular a method of orienting digital panoramic images and a method of hue correction which will be described later. b - Aspects of the support device for obtaining oriented panoramic images

Referring again to FIG. 1, it can be seen that the body 21 has, under the front lens 16 of the panoramic adapter, a region that is substantially set back lying vertically from the axis of rotation 3B, forming a sort of recess where additional elements are arranged. These additional elements include a compass 40 and a tongue 50 fixed to the end of a vertical rod 41 coaxial with the axis of rotation 3B, the rod 41 not being integral in rotation with the body 21. The rod 41, screwed here in the base 1, crosses the base 1 as well as the cylindrical axis 3B of the body 21 to reach the region located under the front lens 16.

These elements are shown in more detail in Figure 7. The compass 40 comprises a housing 42 covered by a glass 43 and carrying a magnetized needle 44. The tongue 50 is arranged horizontally and parallel to the glass 43, and extends above of the compass. The tongue 50 is carried by an arm 51 which runs along the edge of the housing 42. The lower part of the arm 51 is fixed to a disc 52 arranged under the housing 42 and rotatably mounted around the rod 41.

The bottom of the housing 42 preferably has an artificial color distinguished from the natural colors of a shot, for example fluorescent yellow. One half of the needle 44, for example the north half, has a color offering a high contrast with the color of the dial, for example red, while the other half of the needle preferably has the same color than the dial. The tongue 50 itself has a color offering a high contrast with respect to the color of the dial, while being different from that of the needle, for example blue, the needle 44 and the tongue 50 each constituting a mark d orientation intended to be photographed during a shooting. Finally, the central part of the dial has a dark color, preferably black, obtained here by gluing a disc of black paper 45 on the glass 43. Thus, as can be seen in FIG. 2, the dial 46 of the compass seen from above has the appearance of a colored ring, here a yellow ring, cut in the radial direction by a red line (northern half of the needle 44) and by a green line (tongue 50).

FIG. 8 schematically represents a panoramic photograph 65 produced by means of the support device according to the invention. The useful part of this photograph is conventionally circular in shape and the photograph has dark edges which will be removed later when the image is scanned. The dial 46 of the compass being coaxial with the axis of rotation 3B and in alignment with the nodal plane of the front lens 16 of the panoramic adapter, one half of the dial appears in each panoramic photograph taken, whatever the position angle of the body relative to the base. We thus see on the lower edge of the photograph a yellow ring

(dial 46). The yellow ring is here cut in the radial direction by a red line (north half of the needle 44) and by a green line (tongue 50), which means that the photograph was taken substantially in the north direction and / or that the user has not sought to conceal the tab.

With reference to FIG. 6, the recommended "user manual" of the device according to the invention is as follows: the user chooses the point of view from which he wants to take two complementary photographs, takes a first photograph, rotates the 180 ° camera and takes a second photograph, if the earth's magnetic field is present and the needle of the compass is naturally oriented to the North, the user should preferably rotate the tongue 50 around its axis so that it does not appear on the photograph. If, on the contrary, the user is in a place where the earth's magnetic field is attenuated and does not orient the compass needle correctly, the user chooses an arbitrary direction and maintains the tongue 50 in this direction each time shooting and each new group of two photographs, if he then wishes to make a virtual visit of the place without loss of orientation thanks to a process described below. c - Aspects of the support device concerning the control of the tint of a panoramic image

Referring to FIG. 6, it can be seen that the region of the support device located under the front lens 16 also includes a color calibration piece 60. The calibration piece 60 is advantageously annular and coaxial at 1 axis of rotation 3B so as to appear in the shots whatever the angular position of the camera, as can be seen in the photograph in FIG. 8. The calibration piece 60 is arranged here at the periphery of the dial 46 of the compass and is fixed directly to the glass 43, as appears in the sectional view of FIG. 7. The calibration piece is for example a plastic or paper ring glued to the glass 43.

The calibration piece 60 here comprises a plurality of colored sectors 61A, 61B, 61C each having a predetermined primary color. These primary colors are preferably green, red and blue. They are preferably chosen to be unsaturated, and for example have an intensity of 50%. The following values can be chosen with reference to the PANTONE standard:

Sectors 61A: Red 50% or Magenta50 + Yellow50 Sectors 61B: Green 50% or Cyan50 + Yellow50 Sectors 61C: Blue 50% or Cyan50 + Magenta50 The sectors 61A, 61B, 61C form sequences of primary colors which are repeated over the entire perimeter of the calibration piece 60 and thus reveal a succession of Red Green Blue sequences. In an alternative embodiment, the calibration piece 60 is a ring of gray color, for example having a 50% medium gray (Black 50) which corresponds to a color comprising an equal proportion of Red 50%, Green 50% and blue 50%. The calibration part may also include, between the sequences of three primary colors, areas of black or white or sequences of black and white, or alternatively sequences of black, white and gray. We can also predict the presence of an 18% gray (Noirl8) to possibly correct the luminance during a shade correction step described below.

Finally, the calibration piece 60 preferably has a thin black band at its periphery, forming a sort of black ring which surrounds the primary color sequences, the usefulness of which will appear below. The support device according to the invention is of course capable of various variants and embodiments coming within the scope of the present invention, in particular as regards the shape and arrangement of the compass, the structure of the compass, the shape and the arrangement of the color calibration part, its structure and the arrangement of colors on the calibration part.

On the other hand, although the device which has just been described was initially designed to allow the use of a panoramic adapter with compact cameras devoid of means for mounting such a panoramic objective, it should be noted that the support device according to the invention can also be used with SLR ("Single Lens Reflex") cameras. Such use of the support device with SLR devices is justified in particular by the fact that the support device comprises additional elements such as the compass, the orientation tongue, the color calibration part, which are likely to interest the owners of such SLR cameras, in general photography professionals. The usefulness of these additional elements will appear clearly on reading the following description of a method of orientation according to the invention of a panoramic image, and of a method of correction according to the invention of the tint of a digital panoramic image.

II - Description of a process for obtaining an oriented digital panoramic image with a constant hue The flow diagram of FIG. 9 describes the main steps of obtaining an oriented digital panoramic image with a constant hue. We distinguish on this flowchart an acquisition step SI, a digitization step S2, a step S3 of forming a digital panoramic image, a step S4 of orientation of the panoramic image and a step S5 of color correction of the panoramic image. The steps S1, S2 and S3 are in themselves conventional and will only be briefly described. Step S4 is carried out in accordance with an orientation method according to the invention. Step S5 is carried out in accordance with a shade correction method according to the invention. Steps S4 and S5 are inherently independent of each other and could be reversed. However, given the arrangement of the calibration piece 60 in the support device 20 described above, it is advantageous here to carry out step S5 after step S4 for reasons which will appear later.

Step S1 consists in taking at least two complementary panoramic photographs, by rotating the camera by an angle of 180 ° around an axis passing through the nodal plane of the panoramic lens. These steps are preferably performed with a digital camera, although a film camera can also be used if a scanner is available to scan the photographs to obtain photo files. The photo files delivered by the digital camera or by the scanner contain images whose image points are coded RVBA and are arranged in a two-dimensional table, "R" being the red pixel of the image point, "V" the green pixel, "B" the blue pixel, and "A" the Alpha or transparency parameter. The parameters R, G, B and A are generally coded in 8 bits and can thus have an intensity ranging from 0 to 255. Step S2 is a conventional step of transferring the two photo files into a computer, generally a microcomputer , with possible storage in the hard disk. The microcomputer, under the guidance of an appropriate program, transfers the image points of the two photographs into a three-dimensional mathematical space. Consider here and in what follows that this mathematical space is a spherical coordinate system of axes Oxyz, ^'which is the preferred solution for the implementation of the invention. However, it will be clear to those skilled in the art that the present invention is not limited to this example and can also be implemented with other three-dimensional coordinate systems, for example cylindrical, Cartesian, etc. ..

Thus, the RVBA image points of each photograph are transformed during step S2 into RVBA coded image points (φ, θ), φ being the latitude of a point calculated relative to the axis Ox in the vertical plane Oxz, and θ the longitude of a point calculated relative to the axis Ox in the horizontal plane Oxy. The angles φ and θ are coded for example on 4 to 8 bytes (IEEE standard). By convention, the axis Ox is calibrated on the center of the photograph, as illustrated in FIG. 8. At the end of the step S3, there are thus two images in hemispheres.

Step S3 of forming the total panoramic image consists in assembling the two hemispheres by adding the image points which constitute them, and possible merging of the overlap zones if the initial photographs were taken with a shooting angle. greater than 180 °. Before assembly, one of the two hemispheres is rotated 180 ° around the axis Oz by incrementing the angle θ of the image points by a value equal to π, so that a hemisphere comprises image points of longitude between -π / 2 and π / 2 while the other half-sphere includes image points of longitude between π / 2 and 3π / 2.

Conventionally, step S3 can also include the creation of active areas in the panoramic image obtained, and of links connecting the active areas to other spherical panoramic images. a - Orientation of the panoramic image (step S4) It is assumed here that the two initial panoramic photographs were taken by means of the support device described above equipped with its compass 40, or by means of a conventional panoramic head equipped in accordance with the invention of a compass coaxial with the axis of rotation of the nodal plane. Under these conditions, as illustrated in FIG. 10A, the spherical image PII obtained comprises, near its south pole, a dial area 460 which corresponds to the dial 46 described above. The dial area 460 occupies a sphere sector delimited by two parallels PI and P2, corresponding to the annular shape of the dial 46 transposed into the spherical space. The parallel PI has a latitude φl and the parallel P2 a latitude φ2. In the sector of sphere 460, which is for example of fluorescent yellow color as proposed above, there is an orientation mark 461 which must be detected and which has a longitude Θ _N in the horizontal plane Oxy. The determination of this angle Θ _N constitutes the essential objective of step S4 and of the orientation method according to the invention.

In practice, the orientation mark 461 may correspond to the northern half of the needle 44 of the compass and be red, or correspond to the tongue 50 and be green. If we refer to the "instructions for use" proposed above, the presence of the tab means that the user has decided to use the tab as an orientation guide and not to hide it. The detection of the tab must therefore have priority over the detection of the compass needle.

Step S4 can be implemented by means of various image analysis algorithms, the design of which is in itself within the reach of those skilled in the art. According to one aspect of The invention proposes a method of great simplicity comprising first of all a step of searching for the dial area 460, aiming to determine the angles φl and φ2, then a step of searching for the orientation frame 461 in the dial area 460. al - Search for the dial area

The detection of the angles φl and φ2 is ensured by scanning a quarter of a sphere in latitude, from the south pole

(φ = -π / 2) up to the equator (φ = 0). The scanning is performed by following a longitude reference meridian θ ₀ , for example the meridian Ml of zero longitude represented in FIG. 10A, which corresponds to the center of the photograph. The center of the compass dial having been chosen to be black, the detection of the angle φl consists in detecting a transition from black to fluorescent yellow and the detection of consiste2 consists in detecting a transition from fluorescent yellow to another color than the fluorescent yellow. A “color” function is thus defined which consists, for example, of a weighted combination of the colors R, G, B of each image point, the weighting parameters being chosen as a function of the color of the dial in order to obtain maximum detection sensitivity.

The algorithm 1 appearing in the appendix forms an integral part of the description and describes the implementation of this step of the method of the invention. The annotations in parentheses are explanatory comments and are not part of the algorithm. Note that this algorithm provides for the case where the compass dial is not found, so that software executing the algorithm can deactivate the "image orientation" function by itself if the user has performed photographs without a compass or other means of orientation. a.2 - Search for the orientation mark and orientation of the image The angles φl and φ2 being found, it remains to find the orientation mark 461 in the dial area between the parallels PI and P2 in order to determine the 'angle Θ _N. This research is done here by keeping constant the angle φ and going around the sphere in the direction of longitude, from -π to + π. The search is made on an intermediate parallel P12 located in the center of the dial area between the parallels PI and P2, of a latitude φl2 equal to (φl + φ2) / 2. As indicated above, priority is given to detecting the tab, the presence of which means that the user has chosen not to use the compass.

The algorithm 2 appearing in the appendix forms an integral part of the description and describes the implementation of this second step of the orientation method according to the invention. The subroutine called "another attempt" makes it possible to envisage the case where the reference color read at the point where the scanning begins (here the point of angle -π on the meridian M12) would be the color of the orientation mark, which would mean that the search has coincidentally started at the place where the tab or the needle of the compass is located.

Once the angle Θ _{N has been} found and memorized, the orientation of the axes Ox and Oz of the digital image is known and one thus has an oriented image.

In an alternative embodiment, all the image points of the sphere can then be readjusted by orienting the axis Ox on the reference frame. In this case, the angle Θ _N becomes equal to 0 after registration. This alternative embodiment which requires additional computation time is optional in practice, the memorization of the angle Θ _N being sufficient to give the panoramic image an orientation which is lacking in the prior art. The usefulness of such an orientation of a digital panoramic image will appear subsequently, when a method of displaying panoramic images involving the angle Θ _{N will be described} .

It will be clear to those skilled in the art that the process which has just been described is capable of various variants and embodiments, both as regards the steps of inserting an orientation mark in the initial image, that with regard to the method of detecting the orientation mark in the panoramic image obtained after scanning the initial image

Thus, in an alternative embodiment, the needle and the dial of the compass are replaced by a magnetic disc sensitive to the earth's magnetic field, having on its upper face a determined color and one or more graduations indicating one or more cardinal points.

If several graduations are provided for the inscription in the initial image of various orientation marks, such graduations can also be coded by their shape rather than by their color, for example by a number of parallel black lines which differ according to the graduation considered. In general, any means allowing an orientation mark to be written on an initial image intended to be digitized as described above can, according to the invention, be provided. b - Hue correction (step S5) It is assumed here that the two initial panoramic photographs were taken by means of the support device described above equipped with the color calibration part 60, or by means of a conventional panoramic head equipped in accordance with the invention with a color calibration part coaxial with the axis of rotation of the nodal plane. Under these conditions, as illustrated in FIG. 10B, the spherical image PII obtained comprises in the vicinity of its south pole a color calibration zone 600 which corresponds to the calibration piece 60. The calibration zone 600 occupies a sector sphere delimited by the parallel P2, of latitude φ2, and a parallel P3 of latitude φ3. Note that the angle φ2 is known and was determined during step S4, the color calibration part being arranged here at the periphery of the dial of the compass. Furthermore, the sphere sector 600 comprises sequences of primary colors R, G, B whose original intensity on the calibration piece 60 is known and will be designated Iref. The colors are coded on 8 bits, i.e. a color intensity scale ranging from 0 to 255, the original intensity Iref is here of the order of 127 since it was proposed above to provide semi-saturated primary colors (50%).

The shade correction method according to the invention comprises the following steps:

a step of detecting the color calibration zone 600 in the panoramic image,

a step for determining the gamma of the primary colors of the color calibration zone, made with reference to the reference color intensity Iref assigned to each primary color, here the value 127,

- a gamma correction step applied to all or part of the image points of the digital panoramic image, carried out by means of the gamma of the primary colors of the color calibration zone. b.l - Detection of the calibration zone

This step here consists in determining the angles φ2 and φ3 of the parallels P2 and P3, and here boils down to a detection of the angle φ3 since the angle φ2 is known. The angle φ3 is determined by a color transition detection algorithm based on the same principle as the algorithm 1, which will not be described here for the sake of simplicity. Since the calibration piece 60 is provided at its periphery with a thin black strip 62 (FIG. 6), the detection of the angle φ3 consists in detecting a drop in color intensity by scanning the sphere in latitude from l 'angle φ2, for example following the meridian Ml. b.2 - Gamma calculation and gamma correction

The step of calculating the gamma of the primary colors of the calibration zone 600 and the step of gamma correction are carried out by means of mathematical formulas per se conventional. What distinguishes the method according to the invention from the prior art is that these steps are carried out by means of a color reference common to all the photographs. The hue correction performed is thus constant from one panoramic image to another, so that the hue variations observed in the prior art are eliminated by the method of the invention. For the calculation of the gamma, a calculation is first made of the mean value "r", "v", "b" of the primary colors of the calibration zone. It is indeed necessary to take into account the variations in lighting on the various parts of the calibration part at the time when the initial photographs are taken. In addition, as the shooting conditions differ depending on whether you are facing the sun or back to the sun and, for indoor photographs, depending on the light sources present, a calculation of the average intensity of the primary colors over the entire calibration area would be imprecise. The sphere is thus divided into several sectors in the direction of the longitude and the hue correction is ensured sector by sector, by performing in each sector a calculation of the average value of the primary colors, a calculation of the gamma of the primary colors and a correction of gamma.

In an approximation sufficient to obtain a satisfactory and uniform shade correction, the sphere is divided into two hemispheres each corresponding to one of the initial panoramic photographs.

The algorithm 3 appearing in the appendix forms an integral part of the description and describes the implementation of the method according to the invention with a sectoring of the image limited to two hemispheres. The calibration zone 600 (fig. 10B) is read by following a parallel P23 situated halfway between the parallels P2 and P3 and having a latitude φ23 equal to (φ2 + φ3) / 2.

The application of this algorithm to various panoramic images makes it possible to standardize the hue of all the images, thus achieving the desired result.

It will be clear to those skilled in the art that the process which has just been described is capable of various variants and embodiments, both with regard to the method chosen for inserting in the initial image a calibration zone. of colors as regards the steps for detecting the calibration zone and correcting the hue. In general, any means allowing the insertion of a color calibration zone in an initial image intended to be transformed into a digital panoramic image can, according to the invention, be provided. Finally, although we chose higher, for the sake of simplicity, a reference color intensity Iref which is identical for each color, it goes without saying and it follows from the formulas appearing in the algorithm 3 described in the Annex that a given intensity Iref (R), Iref (V), Iref (B) can be chosen for each primary color R, G, B.

III - Description of a method for displaying an oriented image according to the invention a - General principles of the method according to the invention As explained in the preamble, the drawback of conventional virtual visit methods is that the sector initial of the panoramic image displayed on the screen is frozen, the "initial image sector" being the image sector presented on the screen when the observer enters the image. To overcome this drawback, it is necessary in one prior art to provide a complex chaining of panoramic images requiring in most cases a precise topography of places.

Thanks to the orientation method described above, here we have oriented panoramic images in which the angle Θ _N between the orientation frame and the axis Ox is known. This angle is determined for each panoramic image processed, so that the various processed images have a common orientation reference. According to the invention, this common orientation reference is used during a virtual visit to dynamically define, at the time of entering a panoramic image, an orientation which is not fixed as in the prior art but which depends on the position "gaze position" of the observer when he leaves the previous image. The method according to the invention will be better understood by referring to FIGS. 11A and 11B, which represent two examples of entry into a panoramic image SE1 from two different panoramic images SE2 and SE3. We consider here by way of example that the panoramic image SE1 represents a part RI contiguous to a part R2 and contiguous to a part R3. The part R2 is represented by the panoramic image SE2 and the part R3 is represented by the panoramic image SE3. The rooms RI and R2 communicate through a door D1 and the rooms RI and R3 communicate through a door D2. The panoramic images SE1, SE2, SE3 are shown flat in the horizontal plane, in the form of circles. In a region corresponding (after projection onto the circle) at gate D1, the image SE2 includes an active area associated with a link connecting it to the image SE1 (and vice versa). In a region corresponding to gate D2, the image SE3 comprises an active area associated with a link connecting it to the image SE1 (and vice versa). Consider now with reference to FIG. 11A that the observer is "in" the image SE2 and clicks, by means of a screen pointer, on the active zone corresponding to the gate D1. The next image displayed is thus the image SE1. According to the invention, a reference angle θref which represents the angle between the "gaze position" of the observer and the orientation frame is determined in the image SE2. The "gaze position" of the observer is the axis passing through the center 0 ₂ of the image SE2 and the image point Pi of the active area which has been selected by the observer to switch to the following image. The angle θref is given by the following relation:

in which Θ _N2 is the angle between the axis Ox of the image SE2 and the orientation frame, for example north N, θpi being the longitude of the point Pi whose coordinates are φpi and θpi.

According to the invention, an angle θpi 'is then calculated according to the following relation:

(2) θpi '= θref - θ in which Θ _N1 is the angle between the axis Ox of the image SEl and the orientation frame N. The angle found θpi 'defines in the image SEl a set of points of the same orientation.

We then choose an angle φO of arbitrary value, for example the zero angle, which defines with the angle θpi 'a point Pi' with coordinates φO, θpi 'in the image SEl.

When the image SEl is displayed on the screen (the screen being referenced SCR and marked by a thick line on the circle SEl), the initial sector presented on the screen is a sector of the image SEl whose central point is point Pi '. The central point Pi ′ of the initial sector SCR corresponds to the central point of the screen since the initial sector occupies the entire screen. The term “screen” here designates the window for displaying the panoramic image sector, this window being able in practice to occupy only part of the “real” screen that the observer has in front of him.

We see that the point Pi 'forms, with the center 0 ₁ of the image SEl, an axis having an angle θref with the orientation frame, so that the "gaze position" offered to the observer when he enters in the image SE1 is identical to the "gaze position" of the observer when he leaves the image SE2.

The angle φO of the central point Pi 'here being equal to 0, the switching from one image to the other brings the "gaze position" back to horizontal. In an alternative embodiment, the angle φO is chosen equal to the angle φpi of the point Pi of exit from the previous image, so that the observer enters the image SEl with an observation angle which corresponds, relative to the vertical plane, to that which it had in the image SE2.

In an alternative embodiment, an angular sector is defined centered on the 'angle θpi' and delimited by two values θpi '-θl / 2 and θpi' + θl / 2, the angle "θl" corresponding to the viewing angle offered by the screen in the horizontal plane. We then display on the screen 1 'all the points having an angle θ belonging to this angular sector, by defining as before an angle φO of penetration in the vertical plane and a corresponding sector between φO-φl / 2 and φO + φl / 2, "φl" being the viewing angle offered by the screen in the vertical plane.

FIG. 11B illustrates an entry into the image SE1 from the image SE3 and shows that the method of the invention has the effect of automatically modifying the initial sector displayed on the screen. Here, the "gaze position" in the horizontal plane is the axis [0 ₃ Pi) determined by the center 0 ₃ of the image SE3 and a point Pi selected by the screen pointer on the active area (belonging here at gate D2). The angle θref is therefore different from its previous value and the image sector SCR presented on the screen is oriented to the West whereas it was oriented substantially to the North in the example of FIG. 11A.

The method according to the invention is of course susceptible of various variant embodiments. In the foregoing, we have chosen by convention to enter an image while preserving the "gaze position" of the observer. Various other methods for dynamic determination of the initial sector can be provided while remaining within the scope of the present invention, that is to say by using the orientation mark as a reference means for determining the initial sector. b - Implementation of the method FIG. 12 is a flowchart describing the main steps of a virtual visit method according to the invention. Like the shade correction orientation methods described above, this method is executed by a computer or a microcomputer to which a program is provided comprising image processing algorithms, this program being for example recorded on a CD-ROM or downloadable from the Internet.

The virtual visit begins during a step 100 by the selection of a first panoramic image IP _N of rank N = i, which is loaded into a buffer memory of the microcomputer during a step 110. This first image may be imposed on the observer or be chosen by the latter. During a step 120, the microcomputer tests a "POS _± " flag which indicates to it whether an angle θref of "gaze position" has been defined for this image. If the flag POSi is equal to 1, the microcomputer proceeds to an oriented display of the image during a step 130A. If the POS flag is equal to 0, the microcomputer proceeds with an unoriented display of the image during a step 130B.

The oriented display of the image in step 130A firstly consists in calculating the angle θpi 'of the central point of the initial sector as a function of the reference angle θref and the angle Θ _N , in accordance with the relation (1) described above. Then, the microcomputer selects the image sector whose central point presents the coordinates (φO, θpi ') and displays it on the screen. the non-oriented display of step 130B is carried out in accordance with the prior art, the central point of the initial sector being a point of coordinates φO, θo whose angle θo is arbitrary.

After step 130A or 130B of determining the entry point into the panoramic image, the microcomputer remains inside a interactivity management loop in itself conventional, comprising steps 140, 150 and 160 , which allows the observer to move the image up, down, left or right using a screen cursor or his keyboard, the actions of the observer generating a signal interactivity which determines the movement of the image in the observation window.

Thus, during step 140, the microcomputer determines whether the interactivity signal is present. If the interactivity signal is present, the microcomputer switches to step 150 where it slides the image on the screen according to the sign and / or the value of the interactivity signal, then switches to l step 160 where it determines whether an active area has been selected or not. If the interactivity signal is not present, the microcomputer goes directly to test step 160. After the test step 160, and if no active zone is selected, the microcomputer returns to the step 140.

The 140-160 or 140-150-160 loop is broken when an active zone is selected. The microcomputer switches then in a step 170 where it calculates the reference angle θref in accordance with the relation (1) described above, the angle θref corresponding to the "gaze position" of the observer, then stores the angle θref. During a step 180, the microcomputer determines the image of rank j which is designated by the hyper-anchor link associated with the active zone selected, sets to 1 the flag POS _j of the image of rank j and returns to step 100 for loading the IP image _{N = j} . The POS flag having been set to 1 before loading the image, the initial sector of the new image is displayed in an oriented manner during step 130A. c - Application of the invention to video surveillance The display method which has just been described, as well as the image orientation method on which it is based, are susceptible of applications in other fields than that of digital photography.

By way of example, FIG. 13 represents a video surveillance system comprising video cameras VC1, VC2, VC3, ... VCn equipped with CCD type digital image sensors. The cameras are linked to a central computer 70 arranged in a surveillance center and provided with at least one screen.

It should be noted here that the disadvantage of conventional video surveillance systems is that the various cameras must be mounted on motorized axes controlled remotely, in order to widen the field of surveillance and to be able to scan the various corners of a place to be monitored.

Here, the video cameras VI, VC2, VC3..VCn are equipped with panoramic lenses PLI, PL2, PL3, ... PLn offering a viewing angle preferably equal to or greater than 180 °. The various images II, 12, 13, ... In delivered by the cameras are processed in real time by the central computer by applying the conventional scanning step S2 (FIG. 9) by transfer into a three-dimensional coordinate system. . The images are presented on the screen either simultaneously or by selecting a camera from the n-1 available cameras. The advantage of this video surveillance process is that it allows you to scan places by simply dragging the image sector presented on the screen. This method is equivalent to that which consists in rotating a camera around an axis but has the advantage of a significant economy of means since the motorized axes of the cameras and the means of remote control of the motorized axes are no longer necessary. . In addition, the maintenance operations of the camera fleet are considerably simplified.

According to the invention, each camera is further equipped with an orientation means 40.1, 40.2, 40.3 ... 40.n, for example a compass of the type described above, which is arranged in the field of view. PLI to PLn wide angle lenses. Each image received by the central computer 70 is oriented in real time in accordance with the step S4 described above, and the transitions from one image to another are processed taking account of the angle Θ _N according to the method illustrated in FIG. 12. In practice, the application of the method of the invention can be limited here to the transitions between two images delivered by two different cameras. Indeed, the panoramic images delivered by the same camera, although permanently refreshed, keep the same orientation. On the other hand, the application of the method of the invention to the transition between two images supplied by different cameras provides great comfort in use and makes it possible for example to "follow" without losing orientation a person crossing several surveillance fields , going from one camera to another. The method according to the invention can also be implemented when the images delivered by the cameras are displayed simultaneously on one or more screens, for example to orient all the image sectors simultaneously and in the same direction. In such a video surveillance system, the panoramic lenses of the cameras can also be equipped with a color calibration part, and the step of hue correction S5 described above can be applied to the images. panoramic images from video images delivered by cameras.

APPENDIX (Part of the description)

Algorithm 1:

Definitions: function "color (φ, θ)" = f (R, G, B) at point (φ, θ) ε = constant> 0 (scanning increment) continue = TRUE threshold SI = constant> 0 (SI .- detection threshold of a color variation) θ ₀ = constant (θ ₀ defines the search meridian Ml) θ = θ ₀ (the search will be carried out on the meridian Ml)

Crefl = color (-π / 2, θ ₀ ) (Crefl is the reference color at the south pole of the image (here black)., On the meridian Ml) φ = -π / 2 (search for φl for φ ranging from -π / 2 to 0 .-)

As long as continue = TRUE do c = color (φ, θ ₀ ) If difference between, c and Crefl> SI

Then φl = φ and continue = FALSE

Otherwise φ = φ + ε

End if

If φ> 0 (scanning of φ = -π / 2 to 0 finished, φl not found) - Then go to <result 1>

End if

End As long as (search for φ2 for φ ranging from φl to 0:)

Cref2 = color (φl, θ ₀ ) (reference color of the dial in φl (here yellow)) continue ≈ TRUE

As long as continue ≈ TRUE do c = color (φ, θ ₀ )

If difference between c and p2> IF Then φ2 = φ and continue = FALSE

Otherwise φ = φ + ε

End if

If φ> 0 Then go to <result 1>

End if

End while

Go to <result 2> <result 1>

"Compass dial not found"

Go to <end>

Memorization of φl and φ2 <end>

***

Algorithm 2

Definitions:

Function "color (φ, θ)" = f (R, G, B) at point (φ, θ)) ε = constant> 0

"tongue color" = constant threshold S3 = constant> 0 (tongue or needle detection threshold) threshold S4 = constant> 0 (tongue detection threshold) θl = -π (initial angle for choosing the reference color

Cref) θ = -π (initial angle of tongue or needle search) φl2 = (φl + φ2) / 2 (parallel P12 of tongue or needle search)

<loop 1> (search for θ ranging from -π to + π)

Cref ≈ color (φl2, θl) (reference color to θl) c = color (φl2, θ) (color tested) if difference between c and Cref> S3 then go to <determination> (something was found) otherwise θ = θ + ε if θ> π go to <other attempt> otherwise return to <loop 1>

<determination> (tongue or compass needle determination) θ _N = θ (angle Θ _N of the orientation mark found) if difference between color (φl2, Θ _N ) and "tongue color"<

S4 then go to <result 2> otherwise go to <result 3>

Θl = θl + ε (other choice of reference color) if θl> π go to <result 1> otherwise θ = -π and go to <loop 1> <result 1>

"no orientation mark found, image not oriented"

"tab found"

<result 3>"compass needle found" memorizing Θ _N

<End>

***

Algorithm 3

Definitions:

R (D (φ, θ)) = red component of the image point D (φ, θ) V (D (φ, θ)) = green component of the image point D (φ, θ)

B (D (φ, θ)) = blue component of the image point D (φ, θ)

Threshold S5 = constant> 0 (detection threshold of red, green or blue) φ = φ23 = (φ2 + φ3) / 2 (reading of the calibration zone according to the parallel P23) ε = constant> 0 (increment of latitude reading θ)

Iref = 127 (reference intensity of primary colors on the calibration piece)

<start> FUNCTION CALL "HALF-SPHERE TREATMENT", with: start = -π / 2 end = π / 2

FUNCTION CALL "HALF-SPHERE TREATMENT", with: start = π / 2 end = 3π / 2

<end>

"HALF-SPHERE PROCESSING" FUNCTION (parameters = start, end) θ = start (initial scan angle of the calibration zone) r = 0, v = 0, b = 0 ("r", "v" and " b "average values of red, green and blue on the calibration area)

NR = 0, NV = 0, NB = 0 NR, NV, NB; parameters for calculating mean values r, v, b) <reading of the area of ¹ calibration>

If R (D (φ, θ))> S5 go to <addition of red dots>

If V (D (φ, θ))> S5 go to <addition of green dots> If B (D (φ, θ))> S5 go to <addition of blue dots>

<addition of red dots> r = r + R (D (φ, θ))

ΝR = ΝR + 1 go to <increment>

<addition of green dots> v = v + V (D (φ, θ))

NV = ΝV + 1 go to <increment> <addition of blue dots> b = b + B (D (φ, θ))

ΝB = ΝB + 1 go to <increment>

<increment> θ = θ + ε if θ> end go to <calculation of the average intensity> otherwise go to <reading of the calibration zone>

<calculation of average intensity> (calculation of average intensity r, v, b of each primary color) r = r / ΝR v = v / ΝV b ≈ b / NB go to <gamma calculation>

(calculation of the gamma γr, γv, γb of each primary color) γr = [log (r / 255] / [log (Iref / 255)] γv = [log (v / 255] / [log (Iref / 255)] γb = [log (b / 255] / [log (Iref / 255)] go to <hue correction>

<hue correction> (gamma correction over the entire image sector) for θ going from start to end for γ going from -π / 2 to + π / 2 do:

R (D (φ, θ)) ≈ 255 [R (D (φ, θ)) / 255] ^γr V (D (φ, θ)) = 255 [V (D (φ, θ)) / 255] ^γv

B (D (φ, θ)) = 255 [B (D (φ, θ)) / 255] ^γb

Claims

1. Support device (20, 21) for a camera, comprising means for fixing a camera (10) and means (3A, 3B) for rotation about an axis, characterized in that it comprises a orientation means (40, 50) arranged to appear in a shot when a camera equipped with a panoramic lens (15) is attached to the support device.

2. Support device according to claim 1, comprising a compass having a dial (46) of round shape arranged concentrically with the axis of rotation (3B) of the support device.

3. Support device according to one of claims 1 and 2, comprising a compass comprising a dial having an artificial color intended to be distinguished from the natural colors of a shot during an image analysis by computer aimed at find the location of the compass.

4. Support device according to claim 3, in which the compass comprises an orientation mark (44) having a determined color offering a high contrast relative to the color of the dial, so as to be distinguished from the dial during an analysis computer image to find the location of the orientation mark.

5. Support device according to claim 4, in which the orientation mark is a needle (44) comprising on a first north or south half a color having a high contrast relative to the color of the dial and on a second south or north half a color substantially identical to the color of the dial.

6. Support device according to one of claims 4 and 5, in which the compass (40) has a dark central part (45), so that the dial (46) of the compass forms a colored ring whose -form facilitates the detection of the dial during a computer image analysis aimed at finding the location of the dial.

7. Support device according to one of claims 1 to 6, comprising means (50) for locating an arbitrary orientation, the position of which can be adjusted manually, and means (41, 51, 52) for fixing the means of tracking arranged so that the tracking means can be arranged in a visible position where it appears in a shot when a panoramic lens (15) is associated with the camera.

8. Device according to claim 7, wherein the locating means is a tongue (50) which extends above a dial (46) of compass (40) and has a color offering a high contrast relative to the color of the dial in order to facilitate the detection of the tab (50) during an image analysis by computer aiming to find the location of the locating means.

9. Support device according to one of claims 7 and 8, wherein the locating means (50) is rotatably mounted about an axis (41) coaxial with the axis of rotation (3B) of the support device.

10. Support device according to one of claims 1 to 9, characterized in that it comprises a color calibration part (60) comprising at least three primary colors, and means (40, 41) for fixing the color calibration part arranged so that the color calibration part appears in a shot when a camera equipped with a panoramic lens is arranged on the support device.

11. Support device according to claim 10, in which the color calibration piece (60) is circular and concentric with the axis of rotation of the device (3B).

12. Support device according to one of claims 1 to 11, comprising means for fixing a camera (10) equipped or capable of being equipped with a first lens (11), and means for fixing a lens adapter (15), in particular a panoramic adapter lens, arranged to hold the adapter lens in alignment with the first lens without the need to attach the adapter lens to the camera.

13. Support device according to claim 12, in which the means for fixing the adapter lens comprise a moving part (27) and means (30A, 30B, 30C) for fixing the adapter lens to the moving part, the movable part (27) being slidably mounted and pushed by an elastic means (27A, 27B) in a direction corresponding to the location of the front lens of the first objective.

14. Support device according to one of claims 12 and 13, comprising an optical guide (36) for bringing a front light to a photosensitive cell (13) of one camera.

15. Support device according to one of claims 11 to 14 wherein the means for fixing the adapter lens are arranged to maintain the adapter lens in a position such that the axis of rotation (3B) of the support device is substantially in the nodal plane of the front lens (16) of the adapter lens.

16. Method for orienting a digital panoramic image (PII) obtained by transferring the image points of at least one initial image (65, II, 12, 13, etc.) into a three-dimensional coordinate system (Oxyz). . In), characterized in that it comprises:

a step of inserting an orientation mark (44, 50) on the initial image (65, II-In) at the time of the shooting, and

a step of detecting the location (461) of the orientation mark (44, 50) in the digital panoramic image

(PII), made using an image analysis algorithm.

17. The method of claim 16, wherein the registration of an orientation mark on the initial image comprises arranging a compass (40, 40.1-40.n) in the field of view d 'a panoramic objective (15, PL1-PLn), so that the compass appears on the initial image.

18. The method of claim 17, wherein the step of detecting the location (461) of the orientation mark (44) in the digital panoramic image (PII) comprises a step of finding the location (460 ) of the dial (46) of the compass.

19. The method of claim 18, wherein the step of detecting the orientation mark in the digital panoramic image comprises a step of finding the location (461) of a hand (44) in a dial area (460).

20. Method according to one of claims 18 and 19, in which:

- the image points of the initial image are transferred into a spherical coordinate system to form a spherical panoramic image,

- the compass dial occupies a sphere sector (460) in the spherical panoramic image, and The step of finding the location of the dial in the spherical panoramic image comprises an angular scanning of the sphere following a line (MI) oblique or perpendicular to the sector of the sphere.

21. Method according to one of claims 16 to 20, wherein the registration of an orientation mark on the initial image comprises the fact of having a colored tab (50) in the shooting field of a panoramic lens, so that the tab appears on the initial image.

22. The method of claim 21, wherein the tongue is arranged above an artificial color area (46) intended to be distinguished from the natural colors of a shot, the step of searching for the mark of orientation in the image comprising a step of finding the location of the artificial color area and a step of detecting a tongue color in the artificial color area.

23. The method according to claim 22, in which:

- the artificial color zone on which the tongue is arranged is the dial (46) of a compass, and - the step of searching for the orientation mark in the image includes a step of searching for the tongue in the zone color of the dial and, if the tab is not found, a step to find a compass needle.

24. Method according to one of claims 16 to 23, wherein the initial image is a photograph (65).

25. Method according to one of claims 16 to 23, wherein the initial image (II, 12, 13, ... In) is delivered by a video camera (VC1, VC2, VC3, ... VCn).

26. Process for interactive display on a screen of a digital panoramic image (PII) obtained by transferring into a three-dimensional coordinate system (Oxyz) the image points of at least one initial image (65, II, 12, 13, ... In), comprising the operations of displaying an image sector on the screen initial and to drag the displayed image sector as a function of an interactivity signal, method characterized in that it comprises a preliminary step (S4) of orientation of the panoramic image produced in accordance with the method according to one of claims 16 to 23, and in that the initial image sector displayed on the screen is determined with reference (θ _N ) to the detected orientation mark.

27. The method of claim 26, wherein the initial image sector displayed on the screen is a sector of the panoramic image comprising a set of points whose coordinates in the three-dimensional coordinate system have an angle between a lower limit and an upper limit relative to a reference angle

(θref) function of the orientation mark detected.

28. The method of claim 27, wherein the initial image sector displayed on the screen is determined so that at least one image point (Pi ¹ ) of the initial image sector has, relative to the orientation mark, a angle equal to a reference angle (θref) determined according to the orientation mark detected.

29. The method of claim 28, wherein the image point (Pi) having relative to the orientation mark an angle equal to a reference angle (θref) is displayed in the center of the screen.

30. Method according to one of claims 27 to 29, wherein the reference angle (θref ') is determined during a step (170) of displaying a previous panoramic image, when an active user in the previous panoramic image, a selection zone for a next panoramic image, by calculating the angle (θpi + θ _N ) that shows the selection area relative to the orientation mark.

31. The method of claim 30, wherein the reference angle (θref) is chosen equal to the angle presented by the selection area relative to the orientation frame.

32. Method according to one of claims 26 to 31, wherein the initial image is a photograph (65).

33. Method according to one of claims 26 to 31, wherein the initial image (II, 12, 13, ... In) is delivered by a video camera (VC1, VC2, VC3, ... VCn).