FR2515910A2 - METHOD FOR TRANSMITTING AND REPRODUCING VIDEO SCENES, ESPECIALLY AIR VIEW SCENES, WITH REDUCED FREQUENCY OF IMAGE RECURRENCE - Google Patents

Abstract

THE INVENTION RELATES TO A PROCESS FOR TRANSMITTING AND REPRODUCING VIDEO SCENES, IN PARTICULAR SCENES FROM AERIAL VIEWS, WITH A REDUCED FREQUENCY OF IMAGE RECURRENCE. IN THIS PROCESS, ACCORDING TO WHICH IS INTRODUCED IN THE SUCCESSION OF THE TRANSMITTED O, N IMAGES, INTERCALAR N IMAGES PRODUCED BY INTERPOLATION ANDOR EXTRAPOLATION FROM TWO IMAGES AND WE DETERMINE THE POSITIONS, TO BE ASSOCIATED WITH POINTS IMAGES A, B, C. .. A, B, C OF THE TWO TRANSMITTED IMAGES, THE FIRST OF THESE IMAGES IS REPRESENTED IN THE FORM OF A STILL IMAGE FOR A CERTAIN DURATION, FOLLOWING WHICH THE REPRESENTATION OF INTERLACER IMAGES TAKES PLACE. APPLICATION IN PARTICULAR TO THE TELEVISION RETRANSMISSION OF SCENES RECORDED BY AERIAL SHOOTING.

Description

The invention relates to a method for transmitting and reproducing

  video scenes, especially scenes of

  with a reduced frequency of recurrence of images.

  The invention relates to a method for transmitting and re-

  produce video scenes, including aerial view scenes, with a reduced frequency of recurrence

  the case of the relative displacement of the recorded scene by

  port to the sensor of shooting, according to which one introduces in the continuation of the transmitted images, infill images

  produced on the reproduction side by interpolation and / or

  polation adapted to the formation of the image, taking as a

  two transmitted images and determine the

  of image points, corresponding to the same object and

  to be respectively associated for the interpolation or V

  extrapolation, of the two transmitted images, from the

  determined, as being known, of the positions of the points

  images for at least two object points, having coordinates

  different horizontal and vertical lines in each of the two transmitted images, according to the main patent No. 81 08972, and constitutes an improvement and a different form of

  of the process according to this main patent.

  The quality of reproduction of the interlayer images,

  in the following images transmitted according to the de-

  written in the main patent, depends in particular on the accuracy of the discrete representation of the images, the accuracy of the

  image formation model for the production of

  tercalaires and the precision of calculation operations Depending on the amount eligible for implementation costs, different limits are obtained which result in a flattening

  strong transitions of gray values (limitation of

  details) and random variations of the gray values at the edges of the image (edge instability) and can thus alter the identification capacity and by

  therefore the possibility of exploiting image details.

  The present invention aims to develop even more

  the process according to the main patent from the point of view of perfec-

  in the ability to exploit the details of

  without the expense being significantly increased.

  This problem is solved according to the invention by

  the first of two transmitted images is respectively

  This image is represented as a still image for a certain period of time and after this representation in the form of a still image a reproduction of interlayer images is produced which is produced from the still image - an image and an image transmitted later, by means of interpolation and / or extrapolation adapted to the training

of the image.

  According to an appropriate mode of implementation, one can choose

  the duration of the representation in the form of an image

  fixed in a manner relative to the duration of the re-

presentation of moving images.

  Hereinafter, the process according to

  main patent with reference to the drawings.

  By way of example, an embodiment of the invention has been described below and diagrammatically illustrated in the accompanying drawings.

process according to the invention.

  FIG. 1 represents the geometrical arrangement for the recording of an aerial view scene, taken at height H above the ground, in the case where the camera (video camera) is directed obliquely towards the camera. before (the x and x 'axes are perpendicular to the plane of the drawing;

  Figure 2 shows the registration plan for

  camera management as well as the trajectory of two points of the object plane when moving the camera in the direction of flight according to Figure 1; FIG. 3 represents the temporal arrangement of two transmitted images (O-image and N-image) and an interpolated interpolated-image (n-image) on the reproduction side, with the image-dot positions A 0, AN and B 0, BN determined by the measurement and corresponding to the same object, as well as the association of the CNCO and Cn 'CN image points determined according to a linear model from said positions of the

picture points.

  In the case of the geometric arrangement of the

  re 1, there are shown lines, which are parallel to the y axis at the scene (object plane), in the shooting plane (Figure 2) according to the image forming perspective in the form oblique lines with respect to the axis y 'These lines represent simultaneously the locus of the positions of the image points for an object point (here: the trajectory traveled on the image) during the displacement of the next camera

  the direction of flight (see Figure 1).

  The dependence of the position of the image point of an object point as a function of time, that is to say its shift from one image to another, is determined by the laws of the geometry of the conjugated points. as shown in FIG. 3, it can be approximated by a linear form When the association of the positions for a pair of object points is known

  in two reference images, for example for the posi-

  (image points) A, B O in image 0 and AN, BN in

  image N in figure 3, we can determine the associations

  for the rest of the image field, for example for the image points Co, CN from the ratio of the distances of the object point according to the law of propagation of the radiation The association between the image points CO and CN relating to the same object and having the coordinates x C 0, Y 5 and x 5 N, YCN 'and the corresponding image point Cn

  having for coordinates x 5 n, 5 Yn in an image N is

  managed by an association right, specific for this

  point between the O and N images and can be determined from

  firing of the linear relations between the coordinates x 5 and Y Cn Yo = h 1 Y 5 + h 2 '(1) O nx C = gx x' + 92 (2) and y = N h 3 yn + h 4, (3 ) x CN g 3 XC + g 4 (4) The coefficients h 1 to h 4 and gl to g are obtained from the measured associations of the image points Ao, AN and Bo O BN, hi (i = 1,2,3 , 4) being constant for each n (i.e., within each image) and gi being constant for each y'Cn (i.e., within

  each line of image) (see below) The algorithm

  the equations allows a sequential treatment from the point of view of the lines and from the point of view of the images at

  meaning of the video signal to be delivered and is therefore

  Particularly suitable for the technical implementation of the process From the SCO and SCN gray values for COO and CN image points having the coordinates x 5 O, Y'CO and x CN, YCN determined from the relationships (1 at (4), the gray value S Cn, to be regenerated, is obtained for the image point Cn by interpolation according to the relation: i Sc = 1 (Nn) SC + NS (5) N1 -n N 2 O either by extrapolation according to the relation: SC S (6 a) or S = SC (6)

C C C

  n ONON (6) - Extrapolation is used in image parts in which one or more of the coordinate values determined from relationships (1) to (4) are outside the range of values : b <yb

CO N 2, (

a <<a (8)

2 <

CN CN (8)

  that is to say, outside the shot plane possessed by

  width a and height b in accordance with figure 2.

  An extrapolation made in place of

  may be appropriate in other cases, for example when the association between the reference images can not be

  be transiently determined only unexpectedly

  or when the superposition of information from

images is relatively small.

  Displacements of objects in the scene do not appear

  not feel in a succession of images (or partial images)

  regenerated by extrapolation, while in the case of

  terpolation they are generally characterized by a decrease in

  gradual increase and decrease in light intensity

  This case, in which we do not take image points corresponding to the same object for the interpolation, corresponds in its effects to a fade-in between the positions of the moving object in successive reference images.

  The conditions are similar in the case of disturbances

  channel variations Variations in the gray value

  by parasites in a reference image, increase

  tentatively in the sequence of interpolated images,

  that is to say within N periods of images, and

  then appear again correspondingly.

  The association, necessary for the method, of the reference points A 0, AN and B 0, BN can be determined in particular by means of an image correlation method or from the known on-board measurement data. reception In order to increase decision security, the association measurement for other image points and a subsequent selection according to predetermined criteria (such as reliability or majority criteria) is

  possible If we measure associations of reference points

  especially for the area of a moving object within the

  In such areas, it is also possible to determine image points relating to the same object for the interpolation and it is possible to reproduce these object points directly, and not with the decrease.

  and the increase described above of the luminous intensity.

  Hereinafter we will define the coefficients hi and gi (i = 1,2,3,4) used in equations (1) to (4). The selected decomposition of equations provides advantages for the execution of calculation operations in a technical equipment apparatus N hLN-n 11 -n 12 2 = Nn 1 N (i-ii) * 1

h 3 = N-n -

-

  (N-n) 12h = N-n 4 N-nil (9a) (9b) (9c)

(9d) -

  1 k 1 1 Y Co 1 = 1 k 1 Y 5 nk 2 (y C YC O_N g 2 = 1 k 1 Y 5 n 1 k 1 y '1 CN g 3 = 1 k 1 Y 5 nk 2 (yc - Y 5 n) 2 N Cn g 4 = 1 k Y '1 Cn n The quantities hi and gi have denominators

  which are identical The values of hi must be determined

  one time for each vertical period (duration of an image) and the values for gi must be determined once for each horizontal period (duration of a line) of the television signal In accordance with equation (1) or (3 )

gi depends on hi.

  The following auxiliary quantities are used:

Y AN YB

1 = 1 N N

y -Y B.

AO B O

(lila) (lib)

12 = B-YB (1 1)

N O

  (l Oa) (l Ob) (l Oc) (10 d) and d A-d B = + XAB (12 a)

AYANB BYN BN AN

  k 2 = d A (1-k 1 Y'A) + k 1 x (12 b) with I d A =; N X A (12 c) y 'A x BN BO _O N 2 d d B = y y (12 d)

N B O

  The values for 11, 12, k 1, k 2 must be determined

  Once all the N vertical periods have been created The necessary values of the coordinates are known by measuring the association of the positions of the image points, relative to the same object, AO (x AO 0, y 0) and AN (X 2 N, YAN) and BO (x 0, y 0) and BN (XBN, YBN) The method described is in principle also valid when the associations of the image points within the temporal sequence are described by the exact relationships of the geometry of the points conjugates, instead of being by the linear model more favorable from the point of view

Cost.

  As a supplement or variant of the interpolator

  described in the picture, the corresponding interpolation or extrapolation can be carried out for predetermined image areas, for example for objects moving by relationship to the background of the image and for which it is necessary to establish in a particular way the points

  corresponding coordinates, or in a general way

  to increase the accuracy of reproduction

  and / or highlighting individual parts of images.

  In principle, it is also possible to use the process for

  other applications with relative displacement between the

reil of shooting and the scene.

  Depending on the number of intercalated images introduced, the reproduction can be performed with the initial speed, with

  a compression of time or with a dilation of time.

  The reproduction quality of the interlayer images,

  in the following images transmitted by the process

  described in the main patent, depends in particular on the pre-

  the representation of the discrete representation of images, of

  of the imaging model for the production of

  infill images and the accuracy of cal-

  According to the expenditure of realization, one obtains different limits which result in a flattening of strong transitions of the values of gray (limitation of the contrast of the details) and random variations of the values of gray at the level of the edges of the image (instability of the edges) and can

  alter the identification capacity and therefore the pos-

  exploitability of image details.

  It is possible to improve the ability to exploit the details of the images without a large increase in the expense, by representing respectively the first of two images transmitted for a certain duration in the form of a still image and following this representation under

  the shape of a still image is a representation of

  spacers, which are produced from the still image

  considered and an image subsequently transmitted by the

  terpolation and / or extrapolation adapted to the formation of the image.

  Under the term "still image", it is necessary to understand a reproduction

  producing an image without variation of the image elements The representation in the form of a still image is typically performed by repeating the reproduction of the transmitted image under consideration with a repetition rate of image

  usual for the reproduction apparatus.

  During reproduction in the form of a still image, which occurs for a predetermined duration, the image is

  free from limitations that affect the reproduc-

  Interim images regenerated by interpolation

  and / or extrapolation An observer of the image can therefore

  to use respectively the phase of the still image for

  focus on details of the image (for example,

  The subsequent representation of the moving images is used in this case to bring the observer, in a large measure without fatigue, to the observation of the next image 'transmitted and to allow him also to detect potential

  movements of the object in relation to the background of the image.

  Servant can then not lose sight of the areas of the image that interest him, and focus again immediately

  about the interesting details in the following still image

  in this case, in the case of a series of still images

  succeeding without transition, a certain minimum period of

  search is usually necessary to find in the next image the corresponding interesting area, -particularly in the case of a relatively large displacement of an image to the next As a result of the possibility of obtaining a particular information on the details in the 1 'phase

  image, the limitations, appearing as a result of the

  generation of images, the reproduction accuracy in the phase of the moving image are significantly less important than

  in the case of an exclusive representation of moving images.

  In the case of using the process in real time,

  the phase of the moving image is reproduced with a compres-

  time, since the time available for the representation of the transition from one image to the next is shorter than the duration of the period of the transmitted image sequence and therefore less than the corresponding time interval between the images during the recording For example in the case of a sequence transmitted images having a period of a duration of four seconds (corresponding to a recurrence frequency of 0.25 images per second), and in the in the case of choosing a phase of the still image of 3 seconds, the phase of the moving image is not equal to 4 seconds (time interval between the images during the recording),

but only at 1 second.

  In the case of a greater compression of the time during the phase of the moving image, it may be appropriate to provide, instead of a variation with a constant speed,

  a variation with a speed first gradually increasing

  then gradually decreasing relative to the sudden implementation and stopping, intervening in the case of a constant speed, the displacement of the image during the transition from the still existing still image to the following fixed image, it appears, in the case of the increase or the progressive reduction of the speed, acceleration or

  slowdowns, which provide an enhanced effect

  from the subjective point of view during the observation of the image The speed can vary according to the time for example according to a function in the form of triangle, according to a function of

  angle (for example sine, cosine) or according to an ex-

  or even according to an approximation using poly-

  nomes. The duration and / or the moment of the representation in the form of a still image can be conceived as variable. It is thus possible to produce the reproduction device in such a way that the observer can insert, as desired, sequences of still images. an appropriate duration in the representation of the moving images, for example when the observer of a regenerated series of moving images falls on image areas, in which he could identify details more precisely The switching can then be realized by hand But it is also possible to have an automatic switching, for example according to predetermined elements

  image or as a function of time In the case of an overlap

  enough of the contents of successive images, it is possible to

  It is also possible to extend the phase of the still image at will over the duration of the period of the transmitted image sequence, in which case we do not take into account each of the images, but we work with an image deletion. individual

  (eg every other image), with a longer interval

laughing, from one image to another.

  The ratio of the duration of the phase of the still image to the

  duration of the phase of the moving image may be chosen differently

  In particular, for example, the insertion of a representation in the form of a fixed image can be limited to an area in the center of the image.

  while other parts of the image remain, for example,

  permanently in the representation in the form of

mobile phones.

  Additional provisions known to improve

  the reproduction of the image, especially

  digitalization, in which limitations in terms of the technique of realization in the phase of the moving image may lead to limitations similar to the case occurring during the regeneration of the images according to the main patent, may be advantageously used in

the phase of the still image.

  If necessary, it is possible to represent the

  seems of the scene in a shorter or longer time interval, unlike the real-time operation in

  particular, for example by means of an extension of the overall

  the duration of the representation, proportional to the

  sent in the form of a still image, it can be concluded that in the representation in the form of moving images, the displacement of the scene is reproduced with the speed of the operation in real time, that is to say say without

time compression.

Claims (3)

REVENDI CAT IONS   1 Method for achieving transmission and reproduction   video scenes, in particular aerial scene scenes, with a reduced frequency of recurrence of the images in the case of the relative displacement of the recorded scene with respect to the image sensor, according to which images are introduced hereinafter. transmitted, intermediate images   produced on the reproduction side by interpolation and / or   polation adapted to the formation of the image, taking as basis two images respectively transmitted, and the positions of the image points, which are relative to the same object and must be associated for the interpolation or 1 '   extrapolation of the two images transmitted from the   determined, as being known, the positions of the image points for at least two object points, having   horizontal and vertical coordinates, in each   one of the two images transmitted, according to the main patent, characterized in that the first of two transmitted images is respectively represented for a certain duration in the form of a fixed image and to this representation in the form of an image stationary representation, which is produced from the still image under consideration and from an image transmitted later, by means of   interpolation and / or extrapolation adapted to the image.   Process according to Claim 1, characterized in that the duration of the reproduction in the form of an image fixed is variable.   3. Method according to one of claims 1 or 2,   characterized in that partial areas of the image are   respectively reproduced as a still image.