WO2008001239A2

WO2008001239A2 - System and method for correcting colour images

Info

Publication number: WO2008001239A2
Application number: PCT/IB2007/051983
Authority: WO
Inventors: Volker Blume
Original assignee: Nxp B.V.
Priority date: 2006-06-28
Filing date: 2007-05-25
Publication date: 2008-01-03
Also published as: CN101513036A; WO2008001239A3; EP2039145A2

Abstract

In order to further develop a device (100) as well as a method for correcting at least one colour image, the method comprising the steps of - detecting the luminance (L), the saturation (S) and the hue (H) of the colour image, modulating the white -point (W), the saturation (s) and the hue (H) of the colour image, generating at least one colour gamut matrix for a colour conversion or colour mapping of the colour image, and hard clipping the converted or mapped colour image, in such way that the amount of calculation cycles is reduced by maintaining most of the clipping preserving effect of the adaptive colour gamut mapping (ACGM), it is proposed - to average the white-point (W), the saturation (S) and the hue (H) of the colour image, and to mix the averaged white-point (W), the averaged saturation (S) and the averaged hue (H) of the colour image in advance of the step of generating the colour gamut matrix.

Description

DESCRIPTION

SYSTEM AND METHOD FOR CORRECTING COLOUR IMAGES

The present invention relates to a system for correcting at least one colour image according to the preamble of claim 1.

The present invention further relates to a method for correcting at least one colour image according to the preamble of claim 7.

Mobile liquid crystal displays (LCDs) show a specific colour characteristic due to their used backlight illumination based on light emitting diode (LED) technology. The primary colours red, green and blue strongly deviate from the coordinates which are standardized by the European Broadcast Union (EBU) or which are defined as sRGB (standard Red Green Blue).

In this context, it has to be taken into account that the primaries according to

C[onsultative]C[ommittee for]I[nternational]R[adio] Rec. 709 (= worldwide H[igh]D[efinition]T[ele]V[ision] colour standard) are incorporated into the sRGB specification.

However, the sRGB transfer function (Gamma 0.45) is different from the transfer function standardized for studio video (technical background: CCIR Rec. 709 takes the coordinates from EBU red, EBU blue and the average rounded to an accuracy of two digits from EBU green and S[ociety of]M[otion]P[icture]T[elevision]E[ngineers] green.

The typical characteristic of such a portable display shows on the one hand a low capability of displaying the primary colours red and green; on the other hand, the colour coordinate of the primary colour blue typically overlaps the sRGB colourspace but does not map to the sRGB standardized positio meaning that the display in total tends to a bluish colour reproduction, and even the white point deviates into the blue direction. This effect can be summarized as a "blue-stretch behaviour". Fig. 1 shows the gamut of a typical mobile display in comparison with the sRGB colour triangle. Obviously a colour conversion (mapping) methodology is required otherwise false colour impression disturbs the picture quality. Especially skin and green tones extremely suffer from wrong colour reproduction due to the capability of the human visual perception to recognize those colours.

A typical methodology for colour gamut mapping (CGM) can be accomplished by using a 3 x 3 matrix operation in the RGB domain. The CGM matrix operation

MTX _cgm = (MTX_dιsplay Y - MTX_{sRGB mput} describes the mathematical operation to be performed.

As input to this CGM matrix operation, the CGM input values, i. e. the measured display primaries as well as the input sRGB primaries are required:

measured display primaries: input sRGB primaries: x y x y

R 0.5241 0.3302 R 0.6400 0.3300

G 0.3049 0.5261 G 0.2900 0.6000

B 0.1521 0.1598 B 0.1500 0.0600 W 0.2886 0.3330 W 0.3127 0.3290

The block diagram of Fig. 2A describes the steps required for proper signal processing.

Unfortunately, a matrix operation with fixed coefficients does not take possible clipping artifacts into account; especially if the display gamut is smaller than the input gamut annoying clipping will occur. Soft clipping is a possible solution for this problem; in this context, soft clipping is realized by an adaptive colour gamut mapping (ACGM) in which the coefficients of the matrix are adaptively calculated for each pixel value. In Fig. 2B, additional measurement blocks are added to adaptively determine the matrix coefficients as realized in Philips' LifePix™ comprising algorithms significantly enhancing the image quality and performance of mobile displays, in particular of liquid crystal displays (LCDs) or of organic light emitting diode (OLED) displays.

The basis of this advanced colour gamut mapping algorithm includes four steps: the first step is to reduce the brightness of saturated colours in order to create room for the mapping algorithm; the second step is to reproduce all the colours inside the display gamut with the correct hue and saturation; the third step involves clipping the pixels outside the display gamut by adding a certain amount of white; in the final step, pixels which are too bright are reduced without altering the saturation or hue.

This concise colour mapping algorithm is designed to identify the optimal colour ranges of the primary colours (red, green, blue) and to reprocess the colour information in order to produce more realistic colours; by this, mobile LCDs or OLED displays are complemented without compromising performance or changing the panel. The result is a vividly enhanced colour reproduction on any mobile device with a colour display, without compromising on other areas of screen performance.

The visual perception of this pixel adaptive solution has been recommended as so well that for further investigation this way of video processing has been taken as a reference. Regarding software implementation a major disadvantage of this adaptive colour gamut mapping (ACGM) is the large amount of required calculation cycles.

Regarding the technological background of the present invention, further reference can be made to - prior art document WO 2005/043887 Al relating to a smart clipper for mobile displays; however, said known smart clipper is not frame-based but pixel-based; prior art document US 2003/0189576 Al relating to a method and apparatus for displaying higher colour resolution on a hand-held LCD device; however, neither said known method nor said known apparatus is based on a colour gamut algorithm adapting the colour area to the specific properties of the display; - prior art document US 2006/0018536 Al relating to a embedded colour gamut mapping algorithm for a printer; however, said known algorithm is not frame- updated; prior art article "The Mathematics of Color Calibration" by Michael J. Vrhel and H. Joel Trussell, Image Processing, October 4 to 7, 1998, ICIP (1) 1998, pages 181 to 185; however, said known colour calibration is pixel-based; and prior art article "Implementation of Real Time Color Gamut Mapping Using Neural Network" by Hak-Sung Lee and Dongil Han, Soft Computing in Industrial Applications, June 28 to 30, 2005, SMCia/05, pages 138 to 141; however, said known implementation is pixel-based.

Starting from the disadvantages and shortcomings as described above and taking the prior art as discussed into account, an object of the present invention is to further develop a system of the kind as described in the technical field as well as a method of the kind as described in the technical field in such way that the amount of calculation cycles is reduced by maintaining most of the clipping preserving effect of the adaptive colour gamut mapping (ACGM).

The object of the present invention is achieved by a system comprising the features of claim 1 as well as by a method comprising the features of claim 7. Advantageous embodiments and expedient improvements of the present invention are disclosed in the respective dependent claims.

The present invention is principally based on the idea of colour gamut mapping without feedback loop on frame-based matrix update. In this context, the motivation for developing such frame-updated colour gamut mapping (FUCGM) is to reduce the amount of calculation cycles by maintaining most of the clipping preserving effect of the adaptive colour gamut mapping (ACGM), preferably by using only matrix per frame and/or per field.

For this reason, optionally the ACGM of LifePix™ can be used as a basis in order to transfer the advanced colour gamut mapping algorithm from a pixel-based version into a frame -based version.

The person skilled in the art will appreciate that according to the present invention an adaptation to the colour space of the display can be performed in dependence on the data of the colour image; said adaptation is performed by means of only one constant set of coefficients of the colour space per colour image, thus avoiding unwanted clipping artefacts.

By adapting the colour space by means of a fixed set of coefficients, the calculation effort can be significantly reduced especially for real-time applications supported by software. The frame -updated colour gamut mapping (FUCGM) algorithm according to the present invention is designed for mobile use but can also be used with all other applications where the colour space is to be converted.

According to the present invention, the system as well as the method to perform the adaptive colour gamut mapping on frame wise updated matrix coefficients are used to overcome blue- stretched behaviour typically existing in liquid crystal displays (LCDs) or in organic light emitting diode (OLED) displays, for instance of mobile phones.

The present invention can be applied in a video system, for example for a mobile display on the basis of at least one low-power single chip T[hin]F[ilm]T[ransistor] LCD driver with integrated source driver, gate driver, display R[andom]A[ccess]M[emory], timing control and D[irect]C[urrent]-to-D[irect]C[urrrent] converter. The expected quality level is in between a static colour conversion and Philips' pixel based LifePix™ methodology. As already discussed above, there are several options to embody as well as to improve the teaching of the present invention in an advantageous manner. To this aim, reference is made to the claims respectively dependent on claim 1 and on claim 7; further improvements, features and advantages of the present invention are explained below in more detail with reference to a preferred embodiment by way of example and to the accompanying drawings where

Fig. 1 schematically shows a diagram of mobile display gamut compared to sRGB gamut; Fig. 2A schematically shows a block diagram of a standard colour gamut mapping

(CGM) operation according to the prior art; Fig. 2B schematically shows a block diagram of an adaptive colour gamut mapping

(ACGM) operation according to the prior art;

Fig. 3 schematically shows a block diagram of a measurement system of a frame - updated colour gamut mapping (FUCGM) operation according to the present invention;

Fig. 4 schematically shows a block diagram of a luminance, saturation and hue (LSH) detector according to the present invention, said LSH detector being part of the FUCGM measurement system of Fig. 3; Fig. 5 schematically shows a block diagram of a white-point, saturation and hue

(WSH) modulator according to the present invention, said WSH modulator being part of the FUCGM measurement system of Fig. 3; Fig. 6 schematically shows a block diagram of an average unit / critical colour correction (CCC) unit according to the present invention, said average unit / CCC unit being part of the FUCGM measurement system of Fig. 3;

Fig. 7 schematically shows a block diagram of a colour garmut matrix generator, said colour garmut matrix generator being part of the FUCGM measurement system of Fig. 3; and

Fig. 8 schematically shows a block diagram of a video matrix, said video matrix being part of the FUCGM measurement system of Fig. 3. The same reference numerals are used for corresponding parts in Fig. 1 to Fig. 8.

In order to better understand the proposed solution according to the present invention, Fig. 3 shows a block diagram describing the measurement system 100 of a frame - updated colour gamut mapping (FUCGM) algorithm according to the present invention.

The FUCGM algorithm operates in intra frame mode; consequently, only storage of an actual frame or field is required.

An adjustable subsampling is performed in horizontal and vertical direction by means of a subsampling unit 10 being provided with the input signal IS. The subsampling factor can depend on the incoming resolution of the input signal IS and can vary between two (= 2) for quarter common intermediate format (QCIF) and eight (= 8) for Video Graphics Array (VGA) equal in both directions; a higher subsampling factor than eight is not recommended and results in a worse performance.

After this adjustable subsampling in horizontal and vertical direction, a luminance, saturation and hue (LSH) detector 20 derives a luma signal (with reference numeral lum in Fig. 3) and a saturation signal (with reference numeral sat in Fig. 3) from the RGB input.

Both signals, i. e. the luma signal lum as well as the saturation signal sat are fed to a white-point, saturation and hue (WSH) modulator 30 being also provided with the three parameters WLM, SSm, HSm.

In this context, white-point (with reference numeral W in Fig. 3), saturation (with reference numeral S in Fig. 3) and hue (with reference numeral H in Fig. 3) represent the three degrees of freedom for colour conversion which calculate the attenuation parameters of the colour matrix from the slope values derived from the display colour coordinates; the slope values are calculated upfront and depend on the display colour coordinates as well as on the colour coordinates of the input signal IS. The attenuation parameters are collected separately for average calculation (with reference numeral 6Ow for average white-point, reference numeral 60s for average saturation, and reference numeral 6Oh for average hue in Fig. 3). In this context, the average is calculated over the number of pixels taken into measurement.

As can be further taken from Fig. 3, a critical colour correction (CCC) area detector 40 is connected behind the subsampler 10 and in parallel to the luminance, saturation and hue (LSH) detector 20.

If this CCC area detector 40 has indicated the pixel value as a "critical colour" (with such critical colour being set by parameters and describing a colour area of highly saturated greenish-yellow to redish-yellow) the value triplet will be additionally stored in a respective separated average routine 62w for white-point W, 62s for saturation S, and 62h for hue H. In this context, the critical colour average calculates over the number of critical colour pixels, which are stored separately.

When the frame is completed depending on the critical colour correction (CCC) pixel count being performed by a critical colour correction (CCC) pixel counting unit 50 being connected behind the CCC area detector 40, a handover of the calculated attenuation parameters to a colour gamut matrix generator 80 is performed by a mix routine 70 associated to a further mixer 72 being connected behind the critical colour correction (CCC) pixel counter 50. As can be taken from Fig. 3, colour gamut matrix generator 80 is provided with three parameters WRGB, SRGB, HRGB-

The horizontal/vertical subsampling unit 10, the luminance, saturation and hue (LSH) detecting unit 20, the white-point, saturation and hue (WSH) modulating unit 30, the average calculation routine 6Ow for white -point W, the average calculation routine 60s for saturation S, the average calculation routine 6Oh for hue H, the mix routine 70, and the colour gamut matrix generating unit 80 are assigned to the adaptive processing. Now the preloaded conversion matrix 90 will be attenuated. In the video processing, colour-mapping matrix coefficients S₁, gi, bi, r_l5 S₂, b₂, r₂, g₂, S₃ derived from that calculation are loaded once per frame/per field into the matrix 90, which enables the colour mapping operation to be performed. A simple hard clipper 92 in the post processing, in particular in the post video processing, ensures the predefined value accuracy.

A more detailed disclosure of the luminance, saturation and hue (LSH) detector 20 is given by way of Fig. 4. This LSH detector 20 is designed to derive the luma value lum as well as the saturation value sat from the RGB signal Ri, Gi, Bi.

Taking into account that the square routine is a simple but less accurate method to simulate the influence of gamma, the RGB square routine 22

passes the data Rs, Gs, Bs, i. e. the squared input values Ri², Gi², Bi² to a RGB shuffle operation 24

Max = MAX(R_s , G_s , B_s ) Med = MED(R₃ , G₃, B₃ ) Min = MIN(R_S , G_S , B_S ) sorting the values Rs, Gs, Bs by the maximum Max, by the median Med, and by the minimum Min.

By way of the maximum value Max = MAX(Rs, Gs, Bs), of the median value Med =

MED(Rs, Gs, Bs), and of the minimum value Min = MIN(Rs, Gs, Bs), the luminance determination Ld (= luma signal lum), the saturation determination Si, and the hue determination Cs is performed according to the equations

„ Max - Min _ Med - Min

L_d = Max, S₁ = , C₅ =

Max + Min Max + Min in a luminance, saturation and hue (LSH) determination unit 26 being connected behind the shuffler 24. Behind the LSH determination unit 26, the uncorrected saturation value Si is corrected also by way of the determined hue value Cs according to the equation

in a saturation corrector 28 resulting in the corrected saturation value Sd, i. e. in the saturation signal sat.

A more detailed disclosure of the white point, saturation and hue (WSH) modulator 30 is given by way of Fig. 5. This WSH modulator 30 is connected behind the luminance, saturation and hue (LSH) detector 20 and is provided with the luma signal lum

(= determined luminance value Ld) as well as with the saturation signal sat (= corrected saturation value Sd).

More particularly, the WSH modulator 30 uses the luma signal Ld and the saturation signal Sd in order to calculate the attenuators of the following matrix processing according to the following equations for the determination of the correction value or reduction value Wc for white-point W, of the correction value or reduction value Sc for saturation S, and of the correction value or reduction value Hc for hue H:

W_c = (l-L_d )+ (WL_m -WL_m - L_d ) S_e = {l-S_d )+{SS_m - SS_m - S_d )

H_c = (l- S_d )+ (HS_m - HS_m - S_d )

Therefore display dependent modulator slopes WLm for white -point W, SSm for saturation S, and HSm for hue H are provided for calculation.

In detail, a white-point modulator 32 being part of the WSH modulator 30 is provided with the luma signal lum (= determined luminance value Ld) as well as with the first display parameter WLm for white-point W so as to calculate the first correction value Wc, namely the attenuation value for white-point W. Further, a saturation modulator 34 being also part of the WSH modulator 30 is provided with the saturation signal sat (= determined saturation value Sd) as well as with the second display parameter SSm for saturation S so as to calculate the second correction value Sc, namely the attenuation value for saturation S.

Finally, a hue modulator 36 being also part of the WSH modulator 30 is provided with the saturation signal sat (= determined saturation value Sd) as well as with the third display parameter HSm for hue H so as to calculate the third correction value Hc, namely the attenuation value for hue H.

A more detailed disclosure of the average (AVG) unit / critical colour correction (CCC) unit (with reference numeral AVG/CCC) is given by way of Fig. 6. This AVG/CCC unit is designed to determine the attenuation of the matrix, which performs the colour mapping for the whole frame.

The average unit 60, 62 comprises in principal six bins 6Ow, 62w, 60s, 62s, 6Oh, 62h, namely three bins 6Ow, 60s, 6Oh for the standard average (AVG) determination and three additional bins 62w, 62s, 62h for the "critical colour correction (CCC)" average (AVG) calculation..

For each white-point attenuation, saturation attenuation, and hue attenuation, two respective average bins 6Ow, 62w (for white-point attenuation), 60s, 62s (for saturation attenuation), and 6Oh, 62h (for hue attenuation) are available.

The mixing routine 70 following the average bins 60, 62 is implemented by respective mixing units 7Ow (for white-point W), 70s (for saturation S), and 7Oh (for hue H); each of these mixing units 7Ow, 70s, 7Oh selects the amount of influence from which average bin 6Ow, 62w, 60s, 62s, 6Oh, 62h the value will be taken. The mixer 70 is controlled by the amount of pixels, which are detected as a critical colour. The average (AVG) unit / critical colour correction (CCC) unit AVG/CCC works as follows:

The attenuation values Wc, Sc, Hc previously derived in the white-point, saturation and hue (WSH) modulator 30 (cf. Fig. 5) are collected in the standard average unit 6Ow, 60s, 6Oh. In case of measuring a critical colour these attenuation values Wc, Sc, Hc are additionally collected in the critical colour correction (CCC) average (AVG) bin 62w, 62s, 62h. The critical colour correction (CCC) area detector 40 makes the decision whether the value is stored in the CCC AVG bin 62w, 62s, 62h.

For mobile displays the critical colour on which the mapping operation shall take special attention is a greenish yellow. There are two reasons to take special care of this colour: - the high luma fraction of greenish yellow colour, which makes any disturbance in the colour reproduction especially visible; and the incapability of mobile displays to reproduce this greenish yellow colour area in its full saturation.

A static colour mapping simply clips this greenish yellow colour at the gamut border.

Annoying flat yellow colour areas without any visible structure are the result of this clipping operation. Therefore the influence of the matrix is attenuated according to the defined critical colour.

The critical colour correction (CCC) area detector 40 is a simple comparison block and is designed to determine the colour area to which a critical colour is assigned. The CCC area detector 40 takes the following three criteria (i), (ii), (iii) for critical colour decision into account. In this context, it is to be taken into account that the following exemplary values are dependent on quantization as well as on implementation; consequently, for cases of different application other criteria for determining the critical colour value can be employed. (i) absolute difference of the primaries red and green should be below a certain threshold and defines the colour area width (for instance, for RGB in exemplary quantization of eight bit an exemplary threshold for the colour area width is 35);

(ii) absolute difference of the primaries red and blue should exceed a certain threshold, which defines the red-blue border of the colour area (for instance, for RGB 888 an exemplary threshold for the red-blue border is 120);

(iii) minimum offset of the primary green which limits the area in the green direction

(for instance, for RGB 888 an exemplary green offset is 100); concurrently with the green offset a minimum luma value can be defined for the critical colour area.

If all these conditions (i), (ii), (iii) are true the respective attenuation value Wc, Sc, Hc will be additionally stored in the three bins 62w, 62s, 62h for the critical colour correction (CCC) average (AVG) calculation, and the CCC pixel count will be increased by means of the critical colour correction (CCC) pixel counter 50.

Depending on the CCC pixel count, the average will be calculated on every CCC AVG bin 62w, 62s, 62h, and the mixer 70, 72 will be controlled as well.

In case of a positive detected critical colour the reliability of the the critical colour correction (CCC) average (AVG) routine has to be improved which is done by increasing the number of measurement probes. Therefore the horizontally adjacent pixels are taken into account as well.

In this context, it has to be taken into consideration that small yellow areas are especially visible and can accidentally slip through the measurement subsampling grid. In order to strengthen the influence of these areas the local probe extension can be implemented. Concurrently, thin vertical structures, i. e. thin lines are not overemphasized.

The number of pixels detected as critical pixels control the mixing routine 70, 72. The mixer 72 has a slope start value and a slope stop value, for instance, for Quarter Quarter Video Graphics Array (= QQVGA with a resolution of 160 x 120 pixels) the slope start value is 2 and the slope stop value is 16.

In case the CCC pixel count is lower than or equal to the slope start value the standard average value is used for matrix attenuation; in case the CCC pixel count is greater than the slope stop value the critical colour correction average value is directed to the conversion matrix 90.

As can be taken by the graphic illustration of the mixer 72 in Figs 3 and 6, a linear attenuation mix is calculated from the standard AVG and from the CCC AVG between the slope start value and the slope stop value.

A more detailed disclosure of the matrix generating unit 80 is given by way of Fig.7.

The matrix generator 80 is connected behind the mixing routine 70, 72 and performs the determination of the coefficients S₁, gi, bi, r_l5 S₂, b₂, r₂, g₂, S₃ of the conversion matrix

90; therefor, each respective coefficient W_RGB (= Wr, Wg, Wb) for white -point W, S_RGB (= Sr, Sg, Sb) for saturation S, and H_RGB (= Hr, Hg, Hb) for hue H is attenuated by the respective correction value or reduction value Wc for white-point W, Sc for saturation S, and Hc for hue H calculated from the former average routine 60, 62 (cf. Fig.7): K₁₀=S_r S_c+H_r H_c

^K _2o =S_r S_c-H_r H_c

K₂₁=S_g S_c+H_g H_c

^01 ^{= S}g ^Sc ^{~ H}g ^Hc

K₀₂=S_b-S_c+H_b-H_c κn = S_b S_c-H_b H_c K₁₁=W₁-W₀-S₁-W₀-[I-S₀) κ₂₂=w_b-w_c-s_b-w_c-(ι-s_c)

The generated coefficients K_Oo, Ko₁, K₀₂, Ki₀, Kn, Ki₂, K₂₀, K₂₁, K₂₂ (corresponding to S₁, gi, bi, ri, s₂, b₂, r₂, g₂, S3) are loaded once per frame into the conversion matrix 90, and the whole frame is processed with these settings.

A more detailed disclosure of the conversion matrix or video matrix 90 is given by way of Fig.8.

By further using the red input signal Ri (= red part of RGB input signal), the green input signal Gi (= green part of RGB input signal), and the blue input signal Bi (= blue part of RGB input signal), the conversion matrix or video matrix 90 performs the colour mapping which is loaded once per frame with the coefficients K_Oo, Ko₁, K₀₂, Ki₀, Kn, Ki₂, K₂₀, K₂₁, K₂₂

(corresponding to S₁, gi, bi, r_l5 S₂, b₂, r₂, g₂, S3) previously determined by the matrix generator 80; the actual matrix operation for the colour gamut mapping is performed in the conversion matrix or video matrix 90 on the basis of the following equations:

R₀ =R, + K₀₀ R₁₊K₀₁ (G₁ -R)₊K₀₂ [B₁ -R₁) G₀=G₁₊K_n G₁₊K_n [B₁-G)₊K₁₀ [R₁-G₁) B₀ =B_t +K₂₂ B₁ +K₂₀.[R₁-B)₊K₂₁ [G₁-B)

In this context, it should be considered that the video matrix 90 performs in the nonlinear light domain. A resulting small error in the red output signal Ro (= red part of RGB output signal), the green output signal Go (= green part of RGB output signal), and - the blue output signal Bo (= blue part of RGB output signal) is taken into account; due to the frame-based processing, the errors from non-linear light might be masked. -16-

LIST OF REFERENCE NUMERALS

100 system, in particular measurement system

10 subsampling unit, in particular horizontal and/or vertical subsampling unit 20 luminance, saturation and hue (LSH) detector

22 square routine, in particular RGB square routine, of luminance, saturation and hue (LSH) detector 20 24 shuffler, in particular RGB shuffle operation, of luminance, saturation and hue (LSH) detector 20 26 luminance, saturation and hue (LSH) determination module, of luminance, saturation and hue (LSH) detector 20 28 correction unit, in particular saturation correction module, of luminance, saturation and hue (LSH) detector 20

30 white -point, saturation and hue (WSH) modulator 32 white -point modulating module of white -point, saturation and hue (WSH) modulator 30 34 saturation modulating module of white-point, saturation and hue (WSH) modulator 30

36 hue modulating module of white-point, saturation and hue (WSH) modulator 30

40 critical colour correction (CCC) area detector

50 critical colour correction (CCC) pixel counter

60 first part of average calculating unit 60, 62

6Oh average (AVG) calculation routine for hue H or average (AVG) calculation unit for hue H

60s average (AVG) calculation routine for saturation S or average (AVG) calculation unit for saturation S 6Ow average (AVG) calculation routine for white -point W or average (AVG) calculation unit for white-point W 62 second part of average calculating unit 60, 62 62h critical colour correction (CCC) average (AVG) calculation routine for hue H or critical colour correction (CCC) average (AVG) calculation unit for hue H

62s critical colour correction (CCC) average (AVG) calculation routine for saturation S or critical colour correction (CCC) average (AVG) calculation unit for saturation S 62w critical colour correction (CCC) average (AVG) calculation routine for white -point W or critical colour correction (CCC) average (AVG) calculation unit for white-point W 70 mix routine

7Oh mixing unit for hue H

70s mixing unit for saturation S

7Ow mixing unit for white-point W

72 second mixer or further mixing unit 80 matrix generating unit, in particular colour gamut matrix generator

90 matrix, in particular conversion matrix or video matrix

92 hard clipper

AVG/CCC average (AVG) unit / critical colour correction (CCC) unit B third primary colour blue Bi blue input signal, in particular blue part of RGB input signal IS

Bo blue output signal, in particular blue part of RGB output signal OS

Bs squared blue input signal, in particular squared blue part of RGB input signal bi first colour- mapping matrix coefficient of third primary colour blue B second colour-mapping matrix coefficient of third primary colour blue B

CCC critical colour correction

Cs determined hue signal or determined hue value

D display

DG de-gamma GA gamma

G second primary colour green Gi green input signal, in particular green part of RGB input signal IS

Go green output signal, in particular green part of RGB output signal OS

Gs squared green input signal, in particular squared green part of RGB input signal gi first colour-mapping matrix coefficient of second primary colour green G g₂ second colour-mapping matrix coefficient of second primary colour green G

H hue, in particular hue signal or hue value, representing third degree of freedom for colour conversion Hc attenuation value or correction value or reduction value for hue H H_RGB third coefficient or third parameter, in particular hue coefficient or hue parameter, provided to colour gamut matrix generator 80, namely Hb hue coefficient or hue parameter for third primary colour blue B

Hg hue coefficient or hue parameter for second primary colour green G

Hr hue coefficient or hue parameter for first primary colour red R HSm third parameter provided to white-point, saturation and hue (WSH) modulator 30, in particular display dependent modulator slope for hue H IS input signal, in particular RGB input signal

K₀O first line, first column coefficient generated in matrix generating unit 80

K₀I first line, second column coefficient generated in matrix generating unit 80 Ko2 first line, third column coefficient generated in matrix generating unit 80

Kio second line, first column coefficient generated in matrix generating unit 80

Ki i second line, second column coefficient generated in matrix generating unit 80

K₁₂ second line, third column coefficient generated in matrix generating unit 80

K₂O third line, first column coefficient generated in matrix generating unit 80 K₂i third line, second column coefficient generated in matrix generating unit 80

K₂₂ third line, third column coefficient generated in matrix generating unit 80

L luminance

Ld = lum luma signal or luma value, in particular determined luminance signal or determined luminance value Max maximum of squared input signals Rs, Gs, Bs

Med median of squared input signals Rs, Gs, Bs Min minimum of squared input signals Rs, Gs, Bs

OS output signal, in particular RGB output signal

R first primary colour red

Ri red input signal, in particular red part of RGB input signal IS Ro red output signal, in particular red part of RGB output signal IS

Rs squared red input signal, in particular squared red part of RGB input signal ri first colour-mapping matrix coefficient of first primary colour red R r₂ second colour-mapping matrix coefficient of first primary colour red R

S saturation, in particular saturation signal or saturation value, representing second degree of freedom for colour conversion

Sc attenuation value or correction value or reduction value for saturation S

5₁ first colour- mapping matrix coefficient

5₂ second colour-mapping matrix coefficient

5₃ third colour-mapping matrix coefficient sat = Sd saturation signal or saturation value, in particular determined saturation signal or determined saturation value

Si uncorrected saturation signal or uncorrected saturation value

S_RGB second coefficient or second parameter, in particular saturation coefficient or saturation parameter, provided to colour gamut matrix generator 80, namely Sb saturation coefficient or saturation parameter for third primary colour blue B

Sg saturation coefficient or saturation parameter for second primary colour green G

Sr saturation coefficient or saturation parameter for first primary colour red R

SSm second parameter provided to white-point, saturation and hue (WSH) modulator 30, in particular display dependent modulator slope for saturation S W white or white -point, in particular white -point signal or white -point value, representing first degree of freedom for colour conversion

Wc attenuation value or correction value or reduction value for white-point W WLm first parameter provided to white-point, saturation and hue (WSH) modulator

30, in particular display dependent modulator slope for white-point W W_RGB first coefficient or first parameter, in particular white-point coefficient or white-point parameter, provided to colour gamut matrix generator 80, namely Wb white -point coefficient or white-point parameter for third primary colour blue B Wg white -point coefficient or white-point parameter for second primary colour green G Wr white-point coefficient or white -point parameter for first primary colour red R

Claims

1. A system (100) for correcting at least one colour image, the system (100) comprising - at least one luminance, saturation and hue (LSH) detecting unit (20) for detecting the luminance (L), the saturation (S) and the hue (H) of the colour image, at least one white -point, saturation and hue (WSH) modulating unit (30) for modulating the white-point (W), the saturation (s) and the hue (H) of the colour image, at least one matrix generating unit (80), in particular at least one colour gamut matrix generator, for generating at least one colour gamut matrix for a colour conversion or colour mapping (90) of the colour image, and at least one hard clipping unit (92) for hard clipping the converted or mapped colour image, characterized by at least one average calculating unit (60: 6Ow, 60s, 6Oh) for averaging the white -point (W), the saturation (S) and the hue (H) of the colour image, and by at least one mixing unit (70: 7Ow, 70s, 7Oh; 72) - being connected before the matrix generating unit (80), and being designed for mixing the averaged white-point (W), the averaged saturation (S) and the averaged hue (H) of the colour image.

2. The system according to claim 1, characterized by at least one critical colour correction average calculating unit (62: 62w, 62s, 62h) being connected in parallel to the average calculating unit (60: 6Ow, 60s, 6Oh), and being designed for critical colour correction averaging the white-point (W), the saturation (S) and the hue (H) in case the pixel value of the colour image indicates a critical colour being set by at least one parameter.

3. The system according to claim 1 or 2, characterized by at least one subsampling unit (10), in particular by at least one horizontal and/or vertical subsampling unit, - being connected before the luminance, saturation and hue (LSH) detecting unit

(20), and being designed for horizontally and/or vertically subsampling the input signal (IS), in particular the R[ed]G[reen]B[lue] input signal.

4. The system according to at least one of claims 1 to 3, characterized by at least one critical colour correction area detector (40) being connected behind the subsampling unit (10) and in parallel to the luminance, saturation and hue detecting unit (20), and indicating the pixel value of the colour image as the critical colour; and/or - by at least one critical colour correction pixel counting unit (50) being connected behind the critical colour correction area detector (40) and before the mixing unit (70: 7Ow, 70s, 7Oh; 72), and being designed for critical colour correction pixel counting.

5. The system according to at least one of claims 1 to 4, characterized in that the luminance, saturation and hue detecting unit (20) conmprises at least one square routine (22), in particular at least one R[ed]G[reen]B[lue] square routine, at least one shuffling unit (24), in particular at least one R[ed]G[reen]B[lue] shuffle operation, connected behind the square routine (22), at least one luminance, saturation and hue determination module (26) connected behind the shuffling unit (24), and at least one correction unit (28), in particular at least one saturation correction module, connected behind the luminance, saturation and hue determination module (26).

6. The system according to at least one of claims 1 to 5, characterized in that the white -point, saturation and hue modulating unit (30) conmprises at least one white -point modulating module (32), - at least one saturation modulating module (34) connected in parallel to the white -point modulating module (32), and at least one hue modulating module (36), connected in parallel to the white- point modulating module (32) as well as to the saturation modulating module (34).

7. A method for correcting at least one colour image, the method comprising the steps of detecting the luminance (L), the saturation (S) and the hue (H) of the colour image, modulating the white -point (W), the saturation (s) and the hue (H) of the colour image, generating at least one colour gamut matrix for a colour conversion or colour mapping of the colour image, and hard clipping the converted or mapped colour image, characterized - by averaging the white-point (W), the saturation (S) and the hue (H) of the colour image, and by mixing the averaged white-point (W), the averaged saturation (S) and the averaged hue (H) of the colour image in advance of the step of generating the colour gamut matrix.

8. The method according to claim 7, characterized by at least one step of critical colour correction averaging of the white -point (W), the saturation (S) and the hue (H), said critical colour correction averaging being performed in parallel to the step of averaging in case the pixel value of the colour image indicates a critical colour being set by at least one parameter.

9. The method according to claim 7 or 8, characterized by at least one step of horizontally and/or vertically subsampling the input signal (IS), in particular the R[ed]G[reen]B[lue] input signal, in advance of the step of detecting the luminance (L), the saturation (S) and the hue (H) of the colour image.

10. Use of at least one system (100) according to at least one of claims 1 to 6 and/or of a method according to at least one of claims 7 to 9 for colour gamut mapping without feedback loop on frame-based matrix update, in particular in order to overcome blue- stretched behaviour existing in liquid crystal displays (LCDs) or in organic light emitting diode (OLED) displays, for instance of mobile phones.