EP3105925A1

EP3105925A1 - Method of enhanced alignment of two means of projection

Info

Publication number: EP3105925A1
Application number: EP14711288.2A
Authority: EP
Inventors: François HELT
Original assignee: Highlands Technologies Solutions
Current assignee: Highlands Technologies Solutions
Priority date: 2014-02-13
Filing date: 2014-02-13
Publication date: 2016-12-21
Also published as: KR20160145545A; US20170180714A1; CN106416242A; WO2015121542A1; CA2939474A1; JP2017511038A; US9992486B2

Abstract

The invention relates to an alignment, on a first projector, of at least one second projector, with the steps: activating the first (PI) and second (P2) projectors so that each projects onto a screen an image pattern comprising a circle on a uniform contrasted background; performing a picture taking of the projections by a sensor (CAP), and transmitting the projected image data to an analysis device; at the analysis device (DIS), identifying centres of respective circles of the projections and transmitting commands for adjusting the second projector so as to make the centre of the circle of the image projected by the second projector coincide with the centre of the circle of the image projected by the first projector. In particular, the determination of respective positions of the centres of circles comprises for each circle: identifying at least one first and one second pair of projected-image zones, each zone of a pair including an arc of a circle, and the zones, pairwise, being spaced a maximum distance apart; determining, for each zone, a profile of spatial distribution of luminous intensity in a direction radial to the circle, so as to identify in said profile an extremum of luminous intensity, and to deduce therefrom spatial co-ordinates of a precise point corresponding to said intensity extremum; defining spatial co-ordinates of the medium between the two precise points of the first pair and of the medium between the two precise points of the second pair, so as to deduce therefrom spatial co-ordinates of the centre of the circle.

Description

Improved alignment method of two projection means

The present invention relates to a technique of aligning a second projection means on a first projection means.

One can search for the alignment of two or more cinematography room projectors, so as to obtain the best projection overlay on the screen. Such an alignment is implemented to increase the level of light on large screens, or for a simultaneous stereoscopic projection "left eye-right eye" in which the images for each eye are assigned to different projectors.

Alignment can also be used for correct overlay of red, green, and blue images for a given projector. The high-end three-component electronic projectors allow adjustment of the superposition of the three primary (or convergence) by adjusting the orientation of the three imaging systems. The collimation of the three primaries also requires the alignment of the three projected primary images. This adjustment is not motorized in the current state of the art.

Aligning two projectors is usually a tedious task that requires the assistance of two people: the projectionist and a person in the room near the screen. This adjustment is currently done manually, as follows: the projectionist projects the same pattern on the two projectors, while the person in the room observes the gap on the screen between the two projected images, and communicates by signs the correction to bring. The projectionist uses this information to modify the position of the second projector (the first one used as a position reference), until an alignment appears satisfactory to the observer. If the projector has a remote control that includes moving the projected image horizontally and vertically, the projectionist can be placed near the screen and manually control the operation. Convergence adjustment is even more tedious because the operator acts on screws that move the planes of imaging systems in three dimensions (rotation and inclination on two axes). He must constantly look alternately on the alignment screws of the systems, strongly lit by a lantern, then on the screen, whose brightness is much lower. The present invention improves the situation. It proposes for this purpose a method implemented by computer means for aligning, on a first projection means, at least a second projection means, the method comprising the steps of:

Activating the first and second projection means to project each on a screen an image pattern comprising a circle on a uniform contrasted background,

Shooting the projections by a sensor, and transmitting the projected image data to an analysis device,

With the analysis device, identifying centers of respective circles projections and transmit control commands of the second projection means to align the center of the circle of the image projected by the second means with the center of the circle of the image projected by the first means.

In particular, the method comprises a determination of respective positions of said centers of circles, comprising, for each circle, the steps implemented by the analysis device:

Identifying at least a first and a second pair of projected image areas, each area of a pair including an arc, and the areas, in pairs, spaced a maximum distance apart,

Determine, for each zone, a profile of spatial distribution of luminous intensity in a radial direction to the circle, to identify in said profile an extremum of luminous intensity, and to deduce therefrom spatial coordinates of a precise point corresponding to said extremum of intensity,

Define spatial coordinates of the medium between the two precise points of the first pair and the middle between the two precise points of the second pair, to deduce spatial coordinates of the center of the circle. Thanks to these arrangements, it is possible to obtain the coordinates of the center of the circle in a much more precise manner (as will be seen below) than by a simple circle recognition, such as a Hough transform recognition. Thus, the alignment of the second projection means on the first projection means is much thinner and precise. In one embodiment, the circle of the projected images is specifically black (or at least dark) on a uniform colored background, and a minimum in light intensity is sought in each of the aforementioned profiles. Such an arrangement makes it possible to be less dependent in particular on the noise in the image acquired by a sensor if a light circle on a dark background was chosen. For example, the aforementioned uniform colored background may be of green primary color, because this color, which has the greatest brightness, serves as a geometric reference in the trichromatic systems. In one embodiment, the two images of the two projection means can be projected simultaneously, in particular if the two circles of the respective projections can be identified by, for example, different colors. However, in a preferred option in which the same specific pattern is projected (black circle on a green background for example), the method, alternatively, rather comprises the steps:

- Activation of the first projection means, with memorization of the coordinates of the center of the circle identified in the image projected by the first means,

- Then activation of the second projection means (the first projection means being extinguished), and comparison of the coordinates of the center of the circle identified in the image projected by the second means with the stored coordinates of the center of the circle of the image projected by the first way, to define the setting controls of the second means.

To identify the arcs of the aforementioned zones, it is preferentially carried out a preliminary step of rough recognition of circle, for the identification of the pairs of zones. For example, a rough Hough transform recognition gave good preliminary results.

Nevertheless, the determination of extremums in each zone gave much more precise results. One possible embodiment for this purpose consists in obtaining the luminous intensity profiles respectively by Radon projections, followed for example by Lagrange interpolations to obtain an extremum point for each profile (as detailed below with reference to FIG. 5).

In a practical implementation:

the zones of the first pair are situated at the left and right ends of the circle and the sum divided by two of the abscissas of the respective precise points of the zones of the first pair gives the abscissa of the center of the circle in a chosen reference point, and

the zones of the second pair are situated at the lower and upper ends of the circle and the sum divided by two of the ordinates of the respective specific points of the zones of the second pair gives the ordinate of the center of the circle in the chosen reference point. Of course, the reference chosen is the same for the two images respectively projected by the first and the second projection means (for example a pixel (0,0) of lower left edge of the captured image which encompasses the entire projected pattern, the means of shooting remaining fixed between the two acquisitions of projected images). In a particular embodiment of the aforementioned Radon projection, the luminous intensities of pixels of each zone, these pixels being taken in a direction perpendicular to the radial direction of the circle, are averaged to construct the profile associated with the zone. For example, if the zones are rectangular, the luminous intensities of pixels of each zone, taken in the large dimension of the rectangle forming the zone, are summed to build the profile associated with the zone (without necessarily having to calculate an average of these intensities, in the case of rectangular areas).

In one application, the first and second projection means may be two projectors in stereoscopic cinematography. In a variant where the second projection means is arranged to project a coinciding image and superimposed with an image projected by the first projector especially for a brightness enhancement (or for any other artistic or technical effect), the method of the invention may be implemented for collimation of projections by said first and second projection means. In general, it is possible, by implementing the method according to the invention, to control the alignment of a plurality of projection means (typically, more than two projectors). Thus, the method may comprise the alignment of at least a third projection means (or more) on the first and / or second projection means, with a reiteration of the steps of the method to do this.

For example, in the context of stereoscopic cinematography again, for example, four projection means used for stereoscopic cinematography can be considered, with two projection means aligned with each other for each eye, and the pairs of projection means for each eye. being further aligned with each other for stereoscopic vision.

The present invention also relates to a system for adjusting a second projection means to be aligned on a first projection means, comprising, for the implementation of the method according to the invention:

a device for controlling at least the second projection means for projecting onto a screen an image pattern comprising a circle,

a sensor for acquiring a digital image of the projected image, and

an analyzing device connected to the sensor for determining the position of the center of the circle in the projected image and comparing this position with a determined circle position on an image projected by the first projection means, for delivering adjustment controls in alignment second means according to said comparison. Of course, the present invention also relates to the analysis device executing the precise determination of the center of the circle according to the invention.

The present invention also relates to a computer program (and / or a readable memory medium comprising the instruction data of such a program), this program including in particular instructions for the implementation of the method, when this program is executed by a processor (for example from the above-mentioned analysis device). An example of a general algorithm of such a program is described below, with reference to FIG.

Other features and advantages of the invention will appear on examining the detailed description below, and the attached drawings in which:

FIG. 1 illustrates an exemplary system according to the invention capable of aligning two projectors PI and P2;

FIG. 2 illustrates the main steps of an exemplary method within the meaning of the invention;

Fig. 3 illustrates an example of a projected pattern having a circle;

Figure 4 illustrates the aforementioned areas each including an arc; FIG. 5 illustrates the light intensity profiles associated with the zones and each having an extremum.

For the purposes of the invention, a specific projection screen is used, associated with an elaborate procedure for very accurately locating the geometric alignment of the projected image. The registration element (a circle according to a preferred option), constituting the projected image, is associated with a determined position in a digital image taken by a sensor, with a precision better than the tenth of a pixel of this sensor, and thanks to the procedure elaborated above.

In the case of the alignment of multiple projectors, the obtaining of precise positions of the projection makes it possible to automatically control each additional projector to make the projected images coincide as well as possible.

In a first general step illustrated by dashed lines in FIG. 1, a device MIR for projecting projection control the projectors, and in particular the first projector PI for projecting onto a ECR screen a specific reference pattern, for example a black circle on a green background. Thus, the MIR device activates the reference projector, or "first projector" PI, the other projectors being extinguished. The sensor CAP acquires an image corresponding to that of FIG. 3 and transmits the digital data of the projected image (arrow DAT) to a processor PROC of a device DIS within the meaning of the invention which calculates the precise position of the center of the circle. This position is then stored in the MEM memory of the device. The CAP sensor is for example a digital camera, a digital camera or other.

Then, the projector PI (shown in dotted lines) is off, and the MIR control device controls the additional projector P2 to align ("second projector" above shown in solid lines) to project on the ECR screen the same image (circle black on a green background).

The acquisition of this second image is performed by the CAP sensor and the same determination calculation of the center of the circle is executed.

The projectors are not aligned at the beginning, the new position of the center of the circle is different. These differences are received and interpreted by the processor PROC of the analysis device DIS to deliver, for example in the form of signals addressed to a processing unit of the projector P2, control commands moving the projector P2 (arrow COM).

Step by step, it is done again the acquisition of a translated image as well as the geometric calculation of position of the center of the circle until that the distance between the obtained position and the reference position of the center of the circle of the image projected by the PI reference projector coincide at a tolerance threshold close (typically one pixel).

For the convergence of the primaries which concerns only one projector at the same time, one projects a black circle on a white background. We have the superposition of a black circle on a green background, a black circle on a red background and a black circle on a blue background. When the convergences are not good, we typically have a circle with colored fringes or even three colored circles on a white background.

In this case, one can acquire in one image the three components of the image: red, green and blue.

The calculation of the circle centers then gives the geometric positions of the respective centers of the three circles. The green used as a reference, because it has the greatest brightness, the operator is shown a square representing the center of the green circle and red and blue squares representing the respective positions of the centers of the red and blue circles on a considerably enlarged scale . The operator makes the necessary adjustments to match the three primary images. More sophisticated imaging system settings can be proposed to approach good convergence at several locations in the image. For example, it is possible to project circles in the center and at the four corners of the screen, and possibly at the position of the subtitles. that is, in the middle of the bottom of the screen. The method within the meaning of the invention then proposes a calculation and a display of the best conditions for each of the five positions.

Moreover, if control is not possible due to projector interface problems, you can display an enlarged image of the position of the additional projector relative to the reference projector and also display, at the end of the alignment operation, the pointing accuracy obtained. Positioning systems do not make it possible to generally obtain alignments as accurate as the measurements provided in the sense of the invention. Referring now to FIG. 2, a method is described for the optimum alignment of several projectors. We can use only the green channel of the image, that is to say that we obtain a black circle on a green background as shown in Figure 3. This avoids the colored fringes in the case where the convergence of the projectors is not optimum. In addition, a more accurate calculation is obtained according to the tests carried out. In addition, green is the brightest color according to visual perception, which also makes it easier for an operator to acquire the image.

Unlike visual alignment, with such a pattern, we do not use light lines on a dark background, but rather dark lines on a light background. If the first case allows an easy and immediate visual interpretation for human visual perception, it does not give enough brightness for digital image processing. The identification of dark spots on a light background is easier and more accurate in image processing, given the particular electronic noise.

Tests have been conducted with various forms and in particular a black square on a light background. Locating the edges of the square is simpler and gives immediate results. However, these results can be skewed in many ways:

the sensor CAP can be rotated little perceptible, in which case a horizontal and a vertical image do not translate obliques detected on the sensor,

the projected image is never perfect geometrically and we observe trapezoidal, cushion or barrel effects that distort the vertical and horizontal lines,

- perfect uniformity of light is rarely achieved: parts around the square are of different brightness, which distorts the position measurement of the edges.

The shape of the circle has been retained because it can be detected very robustly, especially in image processing by a Hough transform.

Thus, in a first step of the method of FIG. 2, after acquisition of the image by the sensor CAP in step S1, a rough processing of identification of the circle is implemented in step S2. Missing points, because of value too close to the light background, do not disturb the detection of the form. The approximate knowledge of the radius of the circle further facilitates its detection. In step S2, after Hough transform, a first position (xa, ya) of the center of the circle whose accuracy depends on the parameters of the algorithm but first of all on the pixel size of the sensor is then obtained. This position is retained as an integer address of one pixel on the sensor. The accuracy of the detection is then improved as follows.

With reference in particular to FIG. 4, from the knowledge of the radius and the position of the center of the circle deduced from the image stored in the memory MEM, four zones Z1, Z2, Z3 and Z4 are deduced at step S3, corresponding to the (almost) vertical left and right zones of the circle, as well as to the (quasi) horizontal zones of the top and the bottom, as represented in FIG. 4.

The luminous intensity values in each zone are accumulated, by vertical projection for ZI and Z2, or horizontal for Z3 and Z4 (Radon transform), in step S4. A profile modeled by a curve (in dashed lines in FIG. 5) obtained by Lagrange interpolation in particular at the values of the pixels having the lowest intensity sums (dotted line curves of FIG. 5) is obtained.

More particularly, the sum SL of the light intensities / pixels along a column is calculated for the left zone ZI of length L (large dimension of the rectangle formed by zone ZI), and for each column until reaching the width of the zone ZI (small dimension of the rectangle ZI), along the axis x of FIG.

Thus, SL x) = corresponds to the variation in stairs illustrated in Figure 5. Here, x corresponds to a pixel abscissa and takes an integer value (like pixel addresses in general).

These variations are "smoothed" by Lagrange interpolation to obtain the two curves in dashed lines of FIG. 5, in zone ZI and in zone Z2.

Let SL {x = ¾i {') be this interpolation, with x E ZI and x G Z2. Here, x takes on the other hand real values, not necessarily whole. Then, in step S5, the position of the lowest point of each dotted line curve of FIG. 5, rounded to the nearest tenth of a pixel, is sought.

Let: xl = mWx zi {SLm {x) i = Σ _j .i (> \ on the left ZI zone and

x2 = n.i¾ _{€ 2S} (SLm {x) = J on the right zone 72.

The middle of the two optimum points for the right and left areas gives a more accurate value xc of the abscissa of the center of the circle. The medium is used because each side calculated on an arc area does not give a precise position of the edge of the circle.

Thus, in step S6 of FIG. 2, the abscissa xc of the center of the circle is given by xc = ■ (x 1 + x2) f2

The same operation is repeated with the high and low zones by calculating a horizontal projection of Radon. A more accurate value yc is obtained from the ordinate of the center of the circle to the nearest tenth of a pixel taking the middle of the optimum positions found.

Thus, yc = yl + y2) / 2, with

yi = mi¾, g _{£ _Z:} 5LM ^ {y) = Σ | ί (χ>) and yl = | SLm y) = J course, the present invention is not limited to the embodiments described by way -before example; it extends to other variants.

For example, the term "circle" used above generally encompasses any concave closed curve whose center is to be determined. It may be a square (as indicated above, even if the measurement accuracy of the center is less efficient), or an ellipse or an oval whose center is to be determined.

In addition, rectangular zones have been described above as a possible embodiment. Nevertheless, the zones may have other shapes (for example elliptical or circular), in which case it is appropriate to average (and not just sum) the luminous intensities of pixels in directions perpendicular to the radial direction (because the number of pixels varies from one average calculation to another). CLAIMS

1. A method implemented by computer means for aligning, on a first projection means, at least one second projection means, the method comprising the steps of:

- With the analysis device, identify centers of respective circles projections and transmit control commands of the second projection means to align the center of the circle of the image projected by the second means with the center of the circle of the image projected by the first means,

the method being characterized in that it comprises a determination of respective positions of said centers of circles, comprising, for each circle, the steps implemented by the analysis device:

- Determine, for each zone, a profile of spatial distribution of luminous intensity in a radial direction to the circle, to identify in said profile an extremum of luminous intensity, and to deduce spatial coordinates of a precise point corresponding to said extremum of 'intensity,

Define spatial coordinates of the medium between the two precise points of the first pair and the middle between the two precise points of the second pair, to deduce spatial coordinates of the center of the circle.

2. Method according to claim 1, characterized in that the circle of projected images is black on a uniform colored background, and in that a minimum light intensity is sought in each of said profiles.

3. Method according to claim 2, characterized in that the bottom is green primary color.

4. Method according to one of the preceding claims, characterized in that it comprises the steps:

Claims

- Successive activation of the first projection means, with memorization of the coordinates of the center of the circle identified in the image projected by the first means,

- Then activation of the second projection means, and comparison of the coordinates of the center of the circle identified in the image projected by the second means with the stored coordinates of the center of the circle of the image projected by the first means, to define the commands of setting of the second means.

5. Method according to one of the preceding claims, characterized in that it comprises a rough recognition step (S2) of a circle, for the identification of the pairs of zones.

6. Method according to claim 5, characterized in that the coarse recognition is performed by Hough transform.

7. Method according to one of the preceding claims, characterized in that the light intensity profiles are obtained by Radon projections.

8. Method according to claim 7, characterized in that the Radon projections are followed by Lagrange interpolations to obtain for each profile an extremum point.

9. Method according to one of the preceding claims, characterized in that:

the zones of the second pair are situated at the lower and upper ends of the circle and the sum divided by two of the ordinates of the respective specific points of the zones of the second pair gives the ordinate of the center of the circle in the chosen reference point.

10. Method according to one of the preceding claims, characterized in that the light intensities of pixels of each zone, taken in a direction perpendicular to the radial direction, are averaged to build said profile associated with the zone.

11. Method according to one of the preceding claims, characterized in that the zones are rectangular and the light intensities of pixels of each zone, taken in the large dimension of the rectangle forming the zone, are summed to build said profile associated with the zone. .

12. Method according to one of the preceding claims, characterized in that the first and second projection means are two projectors in stereoscopic cinematography.

13. Method according to one of claims 1 to 11, wherein the second projection means is arranged to project a coinciding image and superimposed with an image projected by the first projector especially for a brightness enhancement, characterized in that the method is implemented for collimation of projections by said first and second projection means.

14. Method according to one of the preceding claims, characterized in that it comprises the alignment of at least a third projection means on the first and / or second projection means, with a reiteration of the process steps.

15. Process according to claims 12 and 14 in combination, characterized in that four projection means are used for stereoscopic cinematography, with two projection means aligned with each other for each eye, and the pairs of projection means for each eye being further aligned for stereoscopic vision.

16. Computer program, characterized in that it comprises instructions for the implementation of the method according to one of claims 1 to 15, when the program is executed by a processor.

17. System for adjusting a second projection means to be aligned on a first projection means, comprising, for implementing the method according to one of claims 1 to 15:

a control device (MIR) at least of the second projection means (P2) for projecting onto a screen (ECR) an image pattern comprising a circle,

a sensor (CAP) for acquiring a digital image of the projected image, and

an analysis device (DIS) connected to the sensor for determining the position of the center of the circle in the projected image and comparing this position with a determined circle position on an image projected by the first projection means, for issuing commands from adjustment in alignment of the second means according to said comparison.