WO2004017262A2

WO2004017262A2 - Method and device for reproducing an image with the correct brightness

Info

Publication number: WO2004017262A2
Application number: PCT/DE2003/002459
Authority: WO
Inventors: Hans Strasburger; Torsten Wüstenberg
Original assignee: Hans Strasburger; Wuestenberg Torsten
Priority date: 2002-07-22
Filing date: 2003-07-22
Publication date: 2004-02-26
Also published as: AU2003250798A1; WO2004017262A3; DE10233320B4; DE10233320A1

Abstract

The invention relates to a method for displaying an image on a display device comprising a display surface, said display device being characterised in accordance with a position of the display surface by a respective curve, which represents a functional relationship between a grey value of the image that is to be displayed and a reproduced luminosity in the position. According to the invention, said method enables the image to be reproduced with the correct brightness, irrespective of the display device used, by means of the following steps: (a) measurement of a first plurality of discrete values of curves in a second plurality of positions of the display device for a plurality of grey values; (b) measurement of a third plurality of luminosity values for a predefined grey value in a plurality of positions of the display device; (c) determination from the measurements in steps (a) and (b) of an approximated functional relationship between a grey value of an image to be displayed and a reproduced luminosity for each position of a display device, the functional relationship having position-dependent parameters; (d) calculation of a modified, luminosity-calibrated image from an image to be displayed, by the determination of corresponding luminosity values from the grey values of the image using one of the curves measured in step (a), the modified image with modified grey values being calculated from the luminosity values using the functional relationship determined in step (c); and (e) display of the modified image on the display device. The invention also relates to a display device for carrying out said method.

Description

Method and device for the luminance-correct representation of an image

The present invention relates to a method for luminance-correct display of an image on a display device with a display surface, the display device depending on a position of the display surface in each case by a curve, which indicates a functional relationship between a gray value to be displayed of the image and a displayed luminance represents the position being characterized. Furthermore, the device relates to a corresponding display device for carrying out the method.

Conventional techniques in visual psychophysics for presenting well-calibrated stimuli or stimuli (see Cornelissen et al., "A binocular fiberscope for presenting visual stimuli during fMRI" in Spatial Vision, Vol. 11, No. 1, p. 75 -81, 1997) are not directly applicable to functional magnetic imaging studies due to the disruptive effect of the MRI's high magnetic field on electronic devices in its immediate vicinity. Early attempts have used fiber optic displays to transmit a subject's visual stimulus, but in recent years a conventional presentation method has been to directly stimulate a matte from an LCD projector located outside the MRI booth Project a screen at the entrance of the MRI tube, which is viewed through a mirror attached in front of the subject. With a constant gray value, the luminance on a standard CRT screen (cathode ray tube screen) varies markedly with the position of the screen at which the pixel is displayed. These variations are even more drastic with LCD technology. The inventors of the present patent application became aware of the problem when they experienced deliminal lateral activation in the primary visual cortex by stimulation by Gabor grating presented at 10 degrees eccentricity on the left and right horizontal meridians at one Standard LCD structure have found (see. Fig. 1). The lateralization turned out to be a complete artifact and was effectively caused by the non-uniformity of the stimulus visible to the test subject. The non-uniformity of the luminance across the screen in an image with a constant gray value can be described by the variation in the steepness of the transfer function, which links (local) luminance with the (local) gray value and is often referred to as a gamma function. 3, which will be described more fully below, shows the luminance distribution on our screen at a constant gray value of 128. The non-uniform, non-linear gamma function not only leads to a non-uniform stimulus or stimulus background, but also becomes what It is equally important to distort every image, every image processing (rendering) on lighter parts of the screen, the image being even brighter than to be expected from the mean luminance at this position, and vice versa even darker on dark parts. This has led the inventors to derive a general model of the gantry function for all screen positions, which allows the (actual) luminance to be predicted for the full pixel parameter space, ie at any given screen position and any gray value. On this basis, a compensation technique was developed that allows the experimenter to specify the target image as a matrix of the required luminance values instead of the local gray value.

With regard to the prior art, reference is also made to the European publication EP 1 233 610 A2, from which a method for calculating gamma correction tables is known. This method has the following steps: (a) determining a nonlinear transformation between input values in a first range and output values in a second range, the nonlinear transformation having at least one adjustable parameter; (b) determining at least one approximation equation for the nonlinear transformation, the at least one approximation equation having coefficient values ci; (c) for each coefficient value c; Grouping together all the coefficient values over a range of the at least one adjustable parameter and starting an approximation to each grouping in order to obtain a respective coefficient approximation equation; (d) computing coefficients for a given value of the at least one adjustable parameter using the coefficient approximation equations; (e) Calculate conversion transformations using the calculated coefficients and a simplified mathematical formula; and (f) storing the calculated conversion transformations in a conversion table. ^'

Reference is also made to the European publication EP 0 933 924 A2, from which a method for processing an image is known, which comprises the following steps: obtaining data relating to the brightness of an original image; Dividing the data on the brightness into a plurality of groups for each brightness range based on the histogram of the data on the brightness; Preparing correction information for correcting the brightness data for each of the plurality of groups, and performing image processing on the original image based on the correction information.

The invention is therefore based on the object of avoiding the disadvantages of the prior art and, in particular, of developing a method of the type mentioned at the outset in such a way that an image is corrected for luminance, in particular in the case of fMRI applications, regardless of the display device used can, so that an improved evaluation possibility compared to the known methods is possible.

This object is achieved in a method of the type mentioned at the outset in that the method has the following steps:

(a) measuring a first plurality of discrete values of curves at a second plurality of positions of the display device for a plurality of gray values;

(b) measuring a third plurality of luminance values at a predetermined gray value at a plurality of positions of the display device; (c) determining from the measurements of steps (a) and (b) an approximate functional relationship, specific to the display device, between a gray value of an image to be displayed and a displayed luminance for each position of the display device, the functional relationship having position-dependent parameters; (d) Calculating a modified, luminance-calibrated image from the image to be displayed by determining corresponding luminance values from the gray values of the image on the basis of a curve measured in step (a), with the luminance values the modified image with modified gray values is calculated using the functional relationship determined in step (c); and (e) displaying the modified image on the display device.

Further preferred embodiments of the invention are disclosed in the dependent claims.

The invention, as well as further features, objectives, advantages and possible uses thereof, will be explained in more detail below on the basis of a description of preferred exemplary embodiments with reference to the attached drawings. In the drawings, the same reference numerals designate the same or corresponding elements. All of the described and / or illustrated features, alone or in any meaningful combination, form the subject matter of the present invention, regardless of how they are summarized in the patent claims or their relationship. The drawings show:

1 shows an example of an fMRI pattern, which represents an artifact, caused by an uncalibrated, visual stimulation with a standard LCD / TFT projector system (Sharp);

2 shows a luminance attenuation or attenuation of the fiber optics used to measure the local luminance on the projection screen;

3 a shows an uncalibrated luminance distribution on the matt screen, visible to the test person, for a constant gray value of 128, specifically color-coded;

3b shows an uncalibrated luminance distribution on the matt screen, visible to the test person, for a constant gray value of 128, with isoluminance configurations;

4 gamma curves at five representative screen positions; 5 shows an interpolation of the term (a) of the second order in order to obtain a unified model for the gamma curve;

6 shows the efficiency of the compensation; the upper and lower surfaces show the uncalibrated luminance distribution at a gray value of 128 and the calibrated distribution at 500 cd / m ² ;

7 shows a measured luminance profile of a Gabor grating, specifically with luminance calibration, the Gabor grating luminance having been scanned at 25 x 20 screen positions using our fiber optic system; and

8 shows the long-term stability of the luminance distribution of the projector, with parts (a) and (b) of the figure showing the uncalibrated luminance distribution on the matt screen a year and a half ago or today. There was only a return

Maintenance reasons to the manufacturer and replacement of a bulb (which entailed removing the bulb carrier cage) in between, (c) the figure shows the pixel-wise difference of the two distributions after normalization to the same mean luminance.

The present invention is based on standard projection techniques using liquid crystal (LCD) or thin film transistor (TFT) technology, which show drastic distortions in the luminance and contrast characteristics across the screen. Conventional luminance measurement and calibration techniques cannot be used in the vicinity of MRI scanners. With the help of fiber optics, screen luminances for the full space of screen positions and image gray values were measured and on this basis a compensation technique was developed which uses both luminance homogenization and position-dependent gamma correction. Using the technique described, images can be specified with high precision by a test person in a functional MRI using a matrix of target luminance values instead of local gray values with high precision. The experiments carried out are described below, which at the same time represent a preferred exemplary embodiment of the present invention.

1. Fiber optics for measuring the screen luminance

Standard luminance meters cannot be used to measure screen luminance simply because they do not work near the scanner. Successful calibration requires knowing the precise luminance distribution visible to the subject, which is why it is essential to measure the luminance from within the scamier, i.e. with the stimulus structure identical to that during a functional MRI recording. We have therefore established a fiber optic connection that allows us to transmit the local luminance on the screen to a standard digital luminance meter located in the adjacent scanner control room. A single, plastic-shielded fiber of 10 m length and 1 mm diameter was used (Hirschmann electronics type OKD 1000-B, core: 0.98 mm, overall diameter 2.2 mm, attenuation at 660 nm: 220 dB / km), which at the 0.25 mm grid of our screen covers about 13 pixels. We attached a 1.5 x 8.0 cm plastic plate with epoxy at right angles to the ends of the fiber cable so that one end of the cable could be firmly attached to the luminance meter and the other end could be positioned at a defined position on the projection screen could. In order to determine the attenuation or attenuation of the fiber optics with regard to light energy, 10 luminance measurements were carried out on 25 of the possible 256 gray values, both by means of the fiber optics and directly at the corresponding screen position. The results are shown in Fig. 2. The weakening or damping is well described by a linear function. If the regression is forced through the zero point and the attenuated signal is treated as a prediction variable, the (inverse) transfer function is the same

L = 73.3 'at (1) ie the attenuation factor is 73.3 or 18.65 dB (dB defined as 10 dB per unit logarithm).

2. Luminance distribution on the matt screen

The projector in our system (Sharp, model XG-SV 1E) has 3 LCD panels or surface elements in TFT active matrix driver technology, each with 832 x 624 pixels, illuminated by a 370 W metal halide bulb. The image is projected onto a matt screen (600 x 300 mm, 5 mm thick, on one side a matt surface element made of polymethane acrylate, ie plexiglass), which was attached to the entrance of the tube of the scanner and by the test person through a mirror that is placed near the subject's eyes.

Using the fiber optics and the digital luminance measuring device (type Mavo Monitor, manufactured by Gossen-Metrawatt, range 199.9 cd / m ² , resolution 0.1 cd / m ² , error 2.5% of the value), we have the screen luminance in measured on a grid of 40 x 30 screen positions, with five measurements being made for each of the 1200 positions. An experimenter lay in the scanner tube and held the fiber optic adapter on the screen in the required position, which was indicated to him by a computer-generated cross hair. The other took the measurements on the luminance meter in the neighboring scanner control room. For each measurement, the fiber optics were set to the required position again in order to achieve independent measurements. The measurements were carried out over two days; at the beginning of a session, the projector was allowed to warm up for 10 minutes. 3 shows the resulting luminance distribution on the screen after correction by Equation 1.

The luminance on the bright parts of the screen is greater in the middle left area (x = 140, y = 240) by a factor of three than on the darkest part in the upper right area (x = 780, y = 540).

3. The gamma curve at characteristic screen positions A complete description of the (static) transmission characteristics of the display system requires the specification of the luminance as a function of gray value and pixel position, of which the luminance distribution shown in FIG. 3 represents a cross section at a constant mean gray value. In order to describe the dependence of the luminance on the gray value, five representative positions on the screen (x ,, yϊ) from FIG. 3 were selected - with the maximum and minimum luminance and with three roughly equally spaced luminance values in between - and the complete gamma curve was obtained measured at these positions.

At each position, 10 measurements were taken at 25 equally spaced gray values, i.e. a total of 5 x 10 x 25 measurements. The results are shown in FIG. 4. A second order polynomial was fitted to each of the five data sets by regression

L - a ₍ χ, _y) G + b _(X) y) G + C _(X; y ₎ (2)

Where G: gray value; (x, y): screen position and a, b and c: position-dependent regression coefficients. The regression coefficients are summarized in Table 1. As can be seen from the determination coefficients R (also given in the table), less than 1% of the variance is not taken into account.

Table 1. Regression coefficients of the gamma curve at the five screen positions. (x, y: pixel coordinate; L ¹²⁸ : luminance for a gray value of 128 in cd / m ² ; a, b, c, regression coefficients in gig. 2; R: determination coefficient).

It should be noted that unlike LCD devices where a second order parabola gives excellent fits, the transfer function of a cathode ray tube (CRT) is closer to an exponential function with a non-integer exponent between 2 and 3, which is referred to as gamma , follows; the abbreviation "gamma curve" is derived, which is not related to the mathematical term gamma function.

4. The general gamma function

The task was to derive an approximation of the general function F ((x, y), G) from the five gamma curves (which is considered representative of the population for curves). Examination of the set of regression coefficients in Table 1 shows that the coefficient b of the linear term is quite similar between the five polynomials (close to approximately -1) and can be assumed to be constant with a small loss of accuracy. The coefficient of the absolute term, c, is quite variable (between 9 and 19 cd / m ² ) and is negatively correlated with the second order coefficient a. However, compared to the total range of predicted luminance values, 0 to 2500 cd / m ² , the absolute variation of c is small and can be expected to have a small impact on the quality of the overall fit. In order to reduce the number of free parameters in the general model, the fits shown in FIG. 3 were repeated, both b and c being forced in the regression to their respective arithmetic mean values in table 1 (last column):

L ⁼ ₍ χ, y ₎ G + b G + c, (3)

where L in cd / m ² , G the gray value (between 0 and 255) and a ( _{X) y} ) is freely variable. The resulting second order coefficients and determination coefficients are in Table 2 shown. As can be seen from the R ² values, the fits are about as good as before, so there is little loss of accuracy if b and c are assumed to be constant.

Table 2. Second order coefficients and determination coefficients

We now need to establish a relationship between the second order term and the characteristics of a screen position. The screen position itself is not relevant, and in order to describe the characteristic of a position, we have identified the (empirical) luminance at that position with a medium gray value, 128, by

12 ¹²⁸ ( ^χ , y) L 8 selected. The variation of the coefficient a was then described by regression to L ¹²⁸ .

A _{(x, y)} = pL 1 ¹ 28 ^Ö + q (4)

where p and q are free parameters. The coefficient was reasonably well predicted by Equation 4 (Fig. 5); the resulting coefficients were p = 5.168 10 ^"5 and q = 0.0215.

The complete model of the luminance of the transfer function is thus given by equations 3 and 4 and is repeated here with the parameters used as equations 5 and 6:

L = a _{(X; y)} G ² - 2.433G + 37.46, (5)

a _(x, y) = 5.168 10 ^{^"■} 5 ³ 1 ¹ 2 ^Z TL _8. ⁸ + 0.0215 (6) Together with the matrix of luminance values at medium gray level, L ( _X , _y ), these equations predict the luminance at any gray value and screen position.

5. The corrected picture

In order to obtain the desired gray values for a specific luminance L, equation (5) can be inverted to obtain G (L)

G = 0.58 / a + sqrt [(0.58 / a) ² + (L- 14.3) / a], (V)

where a = a _(x , _{y) is} the second order coefficient calculated from equation (6).

The calculation of the corrected images is then clear:

(1) Calculate from the equation (6) the coefficient matrix a _(X j _{; V} i ₎ at the 40 x 30 sample positions at which L ^{128 was} measured;

(2) Compute the full 832 x 624 matrix a ( _X , _y ) at all pixel positions by linear interpolation;

(3) Obtain the required gray value G from equation (7) for each pixel for the desired luminance L = L _(x , _y ).

Fig. 6 shows a comparison of the luminance distributions, specified as having a constant luminance, before and after the correction. Note that the "wrinkled towel" at L (top surface) turns into a fairly flat surface.

6. Dither the corrected image

Changing any image along sharply defined raster lines leads to highly visible contours in otherwise homogeneous image areas. Spatial smoothing is therefore used in the luminescence correction - commonly referred to as "dithering"

- the corrected image required. In a simple variant of it was on add a random number from the interval [-1, +1] to each pixel before rounding to the next integer or whole number. This represents a step (4), namely the last step in determining the desired gray values.

To illustrate the efficiency of the method, FIG. 7 shows the (actual) luminance profile of a Gabor grating, specifically scanned at 25 x 20 screen positions with our fiber optics.

7. Practical considerations

Calibrating a monitor using the described technique is quite complex: the 1800 measurements (1200 + 600) take about 6 hours. Will the work be worthless because the bulb needs to be changed? The luminance distributions shown in FIG. 3 actually come from a second current set of measurements. A year and a half earlier we had made the same measurements (1200 measurements, about 4 hours); in the meantime, the monitor was once with the manufacturer for maintenance and there was a replacement of the bulb, which required removal of the bulb carrier and what could be expected to change the position of the bulb somewhat in the optical arrangement. FIG. 8 shows the luminance distributions, ie the L ¹²⁸ matrix at the two points in time. Interestingly, the basic pattern does not change, with a bright area on the left that extends into the right. The change appears to be primarily a certain spatial distortion, which may result from a slightly changed position of the bulb in the diffusion chamber. However, the changes are large enough that they make recalibration meaningful.

Although the spatial L ¹²⁸ luminance distribution and the mean image luminance were changed due to aging and bulb changes, the shape of the gamma curve has changed only slightly, in both cases the second-order parabola is dominated by the second-order term, the latter being linear with the luminance medium gray value. Equations (5) and (6) for the old setup were L = a _(x , _y) G ² - 1.16 G + 14.3 (5a)

a _(x , _y) = 5.59 10 ^"5 L ¹²⁸ + 0.00929 (6a)

we can therefore assume that the transparency characteristics of the LCD (responsible for the ga curves) did not change much over time, and that the changes that occurred were due to a slightly changed optical arrangement and efficiency of the bulb. Such an assumption should facilitate repeated calibration.

If the compensation technology is used, its practical application is easy to understand for a specialist. Images can be standard pixel graphics files (pcx, gif, etc.), from a graphics program or generated by graphics algorithms (e.g. in MatLab). The MatLab script calculates a corrected image file, which is then used instead of the original in the stimulus presentation routine. Black and white videos can be processed by using the MatLab script.

How important is the use of a well calibrated visual stimulation or stimulus system in connection with fMRI research? The human visual system has a remarkable ability to compensate for wide variations in luminance through adaptation, and thus the field inhomogeneities appear to be less important than the numbers suggest. However, since the total field size is relatively small compared to the full visual field, these variations will occur in close proximity and local adaptation will play a minor role. The fact that the inhomogeneities are clearly visible implies that they are processed in visual areas of the cortex. In addition, since the gamma function also varies between screen positions, the stimulus pattern contrast will vary across the field and it is known that the COARSE response amplitudes depend on the pattern contrast. Skepticism is particularly appropriate when subtle flvIRI activation differences between the hemispheres in primary visual areas are interpreted with standard simulation systems. Certainly be in neuroophthalmic Diagnostics observed much higher standards, with standard background luminance being kept constant within narrow limits. Using the relatively clear compensation technique described here, the luminance visible to a test subject can be specified with good accuracy, in the complete (monochrome) space / luminance range, which is spanned by a standard 8-bit graphics system.

Claims

claims

1. A method for luminance-correct display of an image on a display device with a display surface, the display device depending on a position of the display surface in each case by a curve which has a functional relationship between a gray value to be displayed of the image and a displayed luminance at the position is characterized, characterized by the following steps:

(a) measuring a first plurality of discrete values of curves at a second plurality of positions of the display device, this for a plurality of gray values;

(b) measuring a third plurality of luminance values at a predetermined gray value at a plurality of positions of the display device;

(c) determining from the measurements of steps (a) and (b) an approximate functional relationship, specific to the display device, between a gray value of an image to be displayed and a displayed luminance for each position of the display device, the functional relationship having position-dependent parameters ;

(d) Calculating a modified, luminance-calibrated image from the image to be displayed by using one of the curves measured in step (a) from the gray values of the

Luminance values corresponding to the image are determined, the modified image with modified gray values being calculated from the luminance values by means of the functional relationship determined in step (c); and

(e) displaying the modified image on the display device.

2. The method according to claim 1, characterized in that in step (d) a simple, approximate functional relationship for the description of the position-dependent parameters is determined.

3. The method according to claim 1 or 2, characterized in that in a step (a ') before step (a) a calibration image with a constant gray value, the associated luminance value of which is preferably determined in a central region of the value range from Curves lies on the display device, the values of the gray curves measured in step (a) being determined essentially at positions of the second plurality of positions of the display device which have the essentially highest value, the substantially lowest value and various, in particular correspond to approximately equidistant, intermediate values of the luminance of the calibration image.

4. The method according to any one of claims 1 to 3, characterized in that the measurement of step (b) at a plurality of approximately equidistantly spaced positions is carried out substantially distributed over the entire display area.

5. The method according to claim 4, characterized in that in step (c) for the unmeasured positions of the display surface, an interpolation of the measured values obtained in step (b) is carried out.

6. The method according to claim 5, characterized in that the interpolation is a linear interpolation.

7. The method according to any one of the preceding claims 1 to 6, characterized in that the predetermined gray value of the measurement in step (b) is preferably in a central region of the value range of the associated gamma curves.

8. The method according to any one of the preceding claims 1 to 7, characterized in that in step (c) from the measurements carried out in step (a) a parameterized curve which represents the luminance as a function of the gray value is fitted.

9. The method according to any one of claims 1 to 8, characterized in that the measurements of steps (a '), (a) and / or (b) are carried out by means of fiber optics on the display panel of the display device.

10. The method according to any one of claims 1 to 9, characterized in that in step (d) spatial smoothing is used in the calculation of the modified gray values.

11. The method according to claim 10, characterized in that for spatially smoothing the modified gray values, the addition of a random number from the value range of [-1; +1] is carried out before rounding to the next integer in order to obtain the modified gray value.

12. The method according to any one of the preceding claims, characterized in that the display device is an LCD display device.

13. The method according to claim 12, characterized in that the display device is provided for use in an MRI scanner.

14. The method according to claim 8 and 12, characterized in that the parameterized curve is a second order polynomial.

15. The method according to claim 14, characterized in that the coefficient of the quadratic term of the second order polynomial is obtained from the measurement of step (b) and is a linear function of the position-dependent luminance of the predetermined gray value.

16. The method according to any one of the preceding claims, characterized in that the method is carried out several times with a plurality of images, in particular images successive in time.

17. The method according to claim 16, characterized in that the steps (a) to (c) are carried out only once for the display of a plurality of images for the display device used and the calculation of the modified, luminance-calibrated images in step (d) is essentially the same for the multitude of images.

18. The method according to any one of claims 1 to 17, characterized in that the measurement in step (b) is carried out independently for each color channel of the display device.

19. The method according to claim 18, characterized in that the steps (c) and (d) are carried out independently for each color channel of the display device, a modified, luminance-calibrated image being calculated for each of the color channels.

20. The method according to claim 19, characterized in that the image displayed in step (e) from the display device obtained in the individual color channels

Pictures.

21. Display device for performing the method according to one of the preceding claims.