EP0868817A2

EP0868817A2 - Sharpness improvement

Info

Publication number: EP0868817A2
Application number: EP97940279A
Authority: EP
Inventors: Michel Wouter Nieuwenhuizen; Leendert Albertus Dick Van Den Broeke; Maarten Wilhelmus Henricus Marie Van Dommelen
Original assignee: Koninklijke Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 1996-10-21
Filing date: 1997-09-26
Publication date: 1998-10-07
Also published as: JP2000502548A; WO1998018263A3; WO1998018263A2; KR19990076726A; MY130988A

Abstract

In a sharpness-improving method, the presence of predetermined colors (R, M) in an image signal (y-in, u-in, v-in) is determined (7) to obtain a control signal (CTRL), and a sharpness of the image signal (y-in, u-in, v-in) is enhanced (1-5) in dependence upon the control signal (CTRL).

Description

Sharpness improvement.

The invention relates to a method of and a circuit for sharpness improvement, and to an image display apparatus and a multi-media apparatus comprising such a sharpness improvement circuit.

Several known circuits of this type provide a sharpness improvement by emphasizing existing luminance transients when such a luminance transient is detected. This is logical because the sharpness perception of a picture is predominantly determined by the luminance signal.

It is, inter alia, an object of the invention to provide a better sharpness improvement. To this end, a first aspect of the invention provides a method as defined in claim 1. A second aspect of the invention provides a sharpness improving circuit as defined in claim 9. Third and fourth aspects of the invention provide an image display apparatus and a multi-media apparatus comprising such a sharpness improvement circuit. Advantageous embodiments are defined in the dependent claims.

In a sharpness-improving method in accordance with a primary aspect of the invention, the presence of predetermined colors in an image signal is determined to obtain a control signal, and a sharpness of the image signal is enhanced in dependence upon the control signal. In a preferred embodiment, the picture quality of TV signals is improved by applying peaking to the luminance channel in dependence upon the color information to enhance the sharpness perception of red and magenta parts in the image.

These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

In the drawings:

Fig. 1 shows a basic block diagram of a first embodiment of the invention;

Fig. 2 illustrates a first possible sharpness detection area;

Fig. 3 illustrates a preferred sharpness detection area;

Fig. 4 shows a possible realization of a peaking filter; Fig. 5 shows a preferred embodiment of an image display apparatus in accordance with the present invention;

Fig. 6 shows a multi-media apparatus in accordance with the present invention; and

Fig. 7 shows a preferred embodiment of a color-dependent sharpness improvement circuit in accordance with the present invention.

Saturated colors (mainly red, magenta and blue) often look blurred in current TV systems (PAL, NTSC and SECAM). This is partly caused by the fact that the color theory used for the conception of these TV systems overlooked the gamma correction in the camera.

A primary aspect of the invention aims at increasing the perceived sharpness of saturated red, magenta and blue. The circuit increases the luminance sharpness in saturated red and magenta parts of the screen. The visible results are positive (blue is not enhanced, because there is very little detail in blue in natural scenes, so only noise is enhanced). The circuit can easily be implemented in an IC.

Television signals are non-linear because of 1/γ pre-correction in the camera and the 7 characteristic of the CRT. This basic property is often overlooked, because one tends to test TV sets with test pictures like a color bar and a grey-scale. These pictures are generated without 7 pre-correction.

In all normal TV systems, the camera sensor decomposes the picture into three primary colors: R, G and B. The 1/7 pre-correction is added immediately after the sensor and before the matrixing of the signals into three signals which are suited for efficient transfer: Y, U and V. In a TV receiver, these signals are de-matrixed into the original R, G and B and displayed on the inherently non-linear CRT.

One of the consequences of this non-linearity is the failure of the constant luminance principle. In accordance with this principle, it is possible to choose orthogonal axes for the representation of a TV signal in such a way that all luminance information is in the Y signal. The other two axes then comprise only color information. Normally, R-Y (or V) and B-Y (or U) are chosen for these color axes.

Due to the above-mentioned 7, there is no such orthogonality. This means that there is always cross-talk between the luminance and chroma information. A simple calculation shows that, for some colors (mainly blue, magenta and red), most of the luminance information is coded in the color signals. Other colors are less influenced (yellow, cyan, green).

Since decoding of the Y, U and V signals back into the color signals R, G and B is the inverse process of coding, there is no overall error in an system which is not band-limited. In that ideal case, it is possible to retrieve the luminance information with the original bandwidth. The invention is based on the recognition that in a real-world TV system, both U and V have a strongly reduced bandwidth (approximately 700 kHz compared with 4 MHz for Y) which hampers the reconstruction of the original luminance information in the TV receiver. As a result, the effective luminance bandwidth as displayed on the screen will be limited, mainly for saturated colors with a low luminance content like red, magenta and blue. This problem is inherent in the current TV systems and is known as the constant luminance defect.

This constant luminance defect is important because the perceived sharpness impression of a picture is predominantly determined by the luminance information. So, if part of the luminance information is transmitted via the chroma channels U and V, the luminance bandwidth will be less than is ideally expected. This results in a blurred image in saturated blue, magenta and red parts.

The invention is further based on the recognition that it is possible to improve the picture by means of the following algorithm: detect where there is red, blue or magenta in the picture and improve the sharpness of the color signals locally. Since the chrominance signals are band-limited to approximately 700 kHz, it is no use increasing the high frequencies of the chrominance signals (they are simply not there). Instead, the luminance sharpness must be increased if the enhancement is effected in the YUV domain. While it is conceivable to enhance the sharpness of the color signals in the RGB domain, sharpness enhancement of the luminance signal is preferred for cost considerations.

Contrary to normal peaking or sharpness algorithms, the extra gain is preferably applied for relatively low frequencies (i.e. 700 kHz to 2 MHz, while prior art peaking commonly affects the frequency range between 2 and 4 MHz). This is desired because the information that is lacking starts at approximately 700 kHz. Above 2 MHz, the influence of this feature turns out to be less important. Near the color subcarrier, the effect should be negligible to avoid amplification of cross-luminance effects.

The sharpness enhancement of the present invention is preferably used in addition to existing sharpness enhancement algorithms. While the existing algorithms are generally satisfactory for enhancing the luminance sharpness, they are insufficiently effective in red, magenta and blue. As for these colors, the existing algorithms mainly boost noise because they peak at a too high frequency. If this peaking were optimized for (saturated) red, magenta and blue, the corresponding parts of the image would be improved, while the other parts (grey, green and yellow) would become very ugly. Consequently, use of a conventional sharpness enhancement for all colors is preferred and, in addition thereto, the enhancement proposed by the present invention for at least the colors red and magenta.

Fig. 1 shows a basic block diagram of a first embodiment of a sharpness improvement circuit SI in accordance with the present invention. A luminance component y- in of an input image signal is applied to a peaking circuit 1 which may comprise a second derivative-obtaining circuit. A peaking signal output of the peaking circuit 1 is multiplied by a control signal CTRL by means of a multiplier 3, and subsequently added to the input luminance component y-in by an adder 5 to produce an output luminance component y-out. The control signal CTRL is derived from input chrominance components u-in and v-in of the image signal by a color detector 7 which detects at least the colors red and magenta, and possibly also the color blue. The chrominance components u-in and v-in are supplied unchanged at outputs u-out and v-out, respectively.

Fig. 2 illustrates a first possible sharpness detection area in a color diagram determined by the chrominance components B-Y on the horizontal axis and R-Y on the vertical axis. The colors red (R), magenta (M), blue (B), cyan (C), green (G), and yellow (Y) are indicated at the corners of the color diagram. Fig. 2 shows how the U and V signals span the color plane. If one wants to detect red, blue and magenta, it is clear that a detection area could look like the shaded area. This area can be described as: (B-Y)/2.03 + (R-Y)/l .14 >_. 0.

To avoid switching effects (one color is detected, whereas another, similar color, is not), the cross-over between detected and non-detected colors is preferably gradual. It is also important that colorless areas are not detected because this would lead to a sharpening of grey-tones, which is not wanted with the color-specific algorithm of the present invention.

Tests revealed that it might be better not to enhance the blue parts because there are few natural objects which are both blue and have a lot of detail. Another reason for not enhancing blue is the limited signal to noise ratio of cameras for blue. It appeared to be better to limit the enhancement to red and magenta. This leads to an adapted formula: α*(B-Y) + 0*(R-Y) - C >. 0. (1)

Fig. 3 illustrates a preferred sharpness detection area, which no longer contains the color blue.

Studies showed that formulae other than formula (1) are possible as well. The following formula (2) proved to be very effective:

CTRL = - C + max ( min [ (α*V + β*\J), (y*V + δ*U) ] , C ) (2) with V = (R-Y)/ 1.14

U = (B-Y)/2.03 α, β, 7, δ all non-negative and preferably about 1, and C non-negative and preferably between 0 and 0.1

Fig. 4 shows a possible realization of the peaking filter 1 of Fig. 1. The input luminance component y-in is applied to a cascade connection of two delay cells 11, 13. An input of the delay cell 11 and outputs of the delay cells 11 and 13 are connected to respective inputs of an adder 21 thru a multiplier 15 multiplying by -0.5, a multiplier 17 multiplying by 1, and a multiplier 19 multiplying by -0.5, respectively. The multiplier 17 may be a direct connection. An output of the adder 21 furnishes the peaking signal output. The peaking circuit of Fig. 4 is a simple -1/2, 1, -1/2 FIR filter. The length of the delay line determines the peaking frequency. This should be chosen to complement the chroma bandwidth of 700 kHz. Taking the low Q factor of the FIR filter into account, the optimal peaking frequency is around 1.5 MHz. A delay time per cell of approximately 250 ns was used for a prototype. Probably a slightly longer delay might be chosen for an optimum result. Those skilled in the art may wish to replace the peaking filter of Fig. 4 by other peaking filters.

The amount of peaking can be controlled by multiplying the peaking signal with the control signal CTRL. If the control signal CTRL is zero, there is no peaking. If the control signal CTRL is 1, the peaking is maximum.

Fig. 5 shows a preferred embodiment of an image display apparatus in accordance with the present invention. Before they are processed, the input components y-in, u-in and v-in are clamped by clamping circuits 23, 25 and 27, respectively. The clamped luminance component y-in is applied to the peaking filter of Fig. 4. The clamped chrominance components u-in and v-in are applied to multipliers 41 and 43, respectively, after having been subjected to compensating delays in delay cells 29 and 31, respectively. The outputs of the multipliers 41 and 43 are summed by an adder 45. The sum is clamped by a clamp 47 and compared with a reference voltage C=Vref by a (soft) comparator 49 to obtain the control signal CTRL. The compensating delay cells 29 and 31 furnish the output chrominance components u-out and v-out, respectively. The embodiment of Fig. 5 corresponds to formula (1); in an embodiment which corresponds to formula (2), the elements 41-49 have to be replaced by another processor which calculates CTRL on the basis of U and V.

The output components y-out, u-out and v-out are applied to a matrix circuit MX which furnishes red, green and blue color signals. The R, G and B color signals are processed by a red processor RP, a green processor GP and a blue processor BP, respectively. Output signals of the color processors RP, GP and BP are applied to a display tube DT.

In Fig. 5, the actual color detector 7 of Fig. 1 has been drawn in some more detail. The sum of U and V is compared with Vref in the soft comparator 49. In practice, this comparator is made with a long-tailed pair, which yields a cross-over region of approximately 250 mV. This is sufficient for a gradual cross-over between detected and non-detected colors.

The additional clamp 47 between the U and V adder 45 and the soft comparator 49 is desired for correct DC biasing. With the choice of V_ref, it is possible to limit the detection area in such a way that colorless areas are not affected by the process.

Fig. 6 shows a multi-media apparatus in accordance with the present invention. The multi-media apparatus has an input 61 for receiving image signals to be processed by an image processor (IP) 63 comprising a sharpness improvement circuit SI of the type shown in Fig. 1. A CD-ROM player 65 is present for providing image and/or data signals II and sound signals S A sound-processing unit (part SND1 of a sound card) 69 receives sound signals from an input 67, and furnishes data signals Dl and sound signals S2. Another processor 73 operates on text signals (TXT) received from a keyboard connector 71. Image and/or data outputs of the devices 63, 65, 69 and 73 are applied to a video unit (graphics card) 75 which furnishes an output video signal to an output 77 to which a monitor (not shown) can be connected. Sound outputs SI and S2 of the CD-ROM player 65 and the sound-processing unit 69 are connected to an output part SND2 of the sound card, whose output is connected to a sound output of the multi-media apparatus.

Fig. 7 shows a preferred embodiment of a color-dependent sharpness improvement circuit in accordance with the present invention. The input luminance signal y-in = Y0 is applied to a delay line having 8 delay cells 81-88 for obtaining delayed luminance signals Y1-Y8. The signals Y0, Y4 and Y8 are applied to the low-frequency peaking filter 1. A coring circuit 91 and a switch 93 are placed between the peaking filter 1 and the fader 3 which is controlled by the color-dependent sharpness control signal CTRL. The switch 93 is controlled by a control signal CDS indicating whether the color-dependent sharpness control in accordance with the present invention is desired.

A second fader 107 is placed between the fader 3 and the adder 5. The signals Yl, Y2, Y4, Y6 and Y7 are applied to a high-frequency peaking filter 94 which is followed by a coring circuit 95 and a fader 97. By means of a control signal CFS, applied to the high-frequency peaking filter 94, a selection is made between a 2.38 MHz peaking frequency and a 3.58 MHz peaking frequency.

All luminance signals Y0-Y8 are applied to a minimum and maximum detector 99 followed by an amplitude detection circuit 101 which determines AMP = MAX - MIN. The detected amplitude AMP is used to control the faders 97 and 107 via comparators 103 and 105, respectively. The purpose of the amplitude-dependent control of the peaking is to prevent spot blooming, resulting in reduced sharpness, from occurring as a result of a too large sharpness enhancement signal. The faders 97 and 107 are controlled by the detected amplitude AMP in such a manner that the output signal will never exceed 110% of the maximum input signal amplitude. Offset voltage sources 107-111 provide three different offset voltages between which a selection is made by a switch 113 in dependence upon the control signals CDS (indicating whether color-dependent sharpness control is desired) and CFS (indicating the peaking frequency of the high-frequency peaking filter 94). The offset voltage selected by the switch 113 is applied to the comparator 103, while a fourth offset voltage is applied to the comparator 105 by offset voltage source 115. These different offsets are desired because the detected amplitude AMP is slightly frequency-dependent, so that AMP should be adjusted in dependence upon CFS. Obviously, if the color-dependent sharpness improvement is switched on, the correction in dependence upon AMP should be adjusted, which explains the CDS-dependent control of switch 113.

A further modification between the embodiment of Fig. 7 and that of Fig. 5 is that minimum detection arrangements 117-119 and 121-123 are inserted before the multipliers 41 and 43, respectively. Each minimum detection arrangements comprises a series connection of two delay cells 117, 118 and 121, 122, and a minimum detection circuit 119, 123 which selects the minimum of the received chrominance signal u-in, v-in and the once and twice delayed chrominance signals supplied by the delay cells. These minimum detection arrangements serve to ensure that the color-dependent sharpness control is only effective for sharpening the high-frequency components within a colored object, while undesired overshoots are prevented from occurring at the boundaries of colored objects.

The sharpness improvement algorithm was tested on a test signal having several highly saturated red, blue and magenta areas. The first conclusions are: the circuit works well and gives the reproduction of saturated red and magenta a more pleasing effect with both PAL and NTSC signals. The sharpness of saturated parts is increased. There seem to be few cases where there is saturated blue with a lot of details. Most of the blue is water or sky, so that a possible increase of noise is more noticeable in blue than in magenta and red. The detection axis has therefore been changed to some extent, so that less blue is detected. This improved the general picture quality.

With more noisy signals, like off-air signals, it will be clear that the enhancement should be limited. If there is too much color sharpness, the extra noise could become objectionable. The coring circuits 91 , 95 of Fig. 7 serve to limit the enhancement.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended claims. For example, a peaking circuit having a controllable peaking frequency is conceivable, in which the peaking frequency for red and magenta is lower than the peaking frequency for the other colors. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The invention can be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer.

Claims

Claims:

1. A sharpness-improving method, comprising the steps of: determining (7, 41-49) the presence of predetermined colors (R, M) in an image signal (y-in, u-in, v-in) to obtain a control signal (CTRL); and enhancing (1-5, 11-21) sharpness of said image signal (y-in, u-in, v-in) in dependence upon said control signal (CTRL).

2. A method as claimed in claim 1, wherein said predetermined colors include red (R) and magenta (M).

3. A method as claimed in claim 1 or 2, wherein said sharpness-enhancing step includes the step of enhancing sharpness of a luminance component (y-in) of said image signal (y-in, u-in, v-in) in dependence upon said control signal (CTRL) indicating the presence of said predetermined colors (R, M).

4. A method as claimed in claim 1 , wherein said determining step (7, 41-49) includes the step of making (41-45) a predetermined linear combination of chrominance signals (u-in, v-in).

5. A method as claimed in claim 4, wherein said control signal (CTRL) is obtained by thresholding (49) said predetermined linear combination.

6. A method as claimed in claim 1 , wherein said sharpness-enhancing step

(1-5, 11-21) includes the step of enhancing frequencies between about 700 kHz and about 2

MHz.

7. A method as claimed in claim 6, wherein said sharpness enhancing step

(1-5, 11-21) includes the steps of: obtaining (11-21) a differential signal by means of delay cells (11, 13) providing a delay suitable to select said frequencies to be enhanced, and a subtracting unit

(15-21); multiplying (3) said differential signal by said control signal (CTRL) to obtain a sharpness adjustment signal; and adding (5) said sharpness adjustment signal to said image signal (y-in, u- in, v-in).

8. A method as claimed in claim 7, wherein a luminance component (y-in) of said image signal (y-in, u-in, v-in) is applied to said delay cells (11, 13), and said sharpness adjustment signal is added (5) to said luminance component (y-in).

9. A sharpness-improving circuit, comprising: means (7, 41-49) for determining the presence of predetermined colors (R, M) in an image signal (y-in, u-in, v-in) to obtain a control signal (CTRL); and means (1-5, 11-21) for enhancing sharpness of said image signal (y-in, u- in, v-in) in dependence upon said control signal (CTRL) to obtain an enhanced image signal (y-out, u-out, v-out).

10. A sharpness-improving circuit as claimed in claim 9, wherein said predetermined colors include red (R) and magenta (M).

11. A sharpness-improving circuit as claimed in claim 9 or 10, wherein said sharpness-enhancing means (1-5, 1 1-21) include means for enhancing sharpness of a luminance component (y-in) of said image signal (y-in, u-in, v-in) in dependence upon said control signal (CTRL) indicating the presence of said predetermined colors (R, M).

12. A sharpness-improving circuit as claimed in claim 9, wherein said determining means (7) comprise minimum detection arrangements (117-119, 121-123) indicating a minimum of input chrominance signals (u-in, v-in) and delayed chrominance signals, wherein said control signal (CTRL) is only active within colored objects.

13. An image display apparatus comprising: a sharpness-improving circuit (1-7) as claimed in claim 9; and means (DT) for displaying said enhanced image signal (y-out, u-out, v-out).

14. A multi-media apparatus for processing image signals and at least one other item from a group including data, text and/or sound, the apparatus comprising: an image processor (IP) including a sharpness-improving circuit (SI) as claimed in claim 9; and means (75) for furnishing said enhanced image signal (y-out, u-out, v-out) and/or said at least one other item.