DE4443284A1 - Reducing higher frequency noise in video signal during reproduction on screen of CRT - Google Patents

Abstract

A detail detection of the video signal is carried out in areas of a partial image or a full image without revealed details. A superimposing circuit (7) is controlled in dependence on the generated signal of the non-appearing details. The signal in the areas without revealed details is superimposed by signal from a noise reduction signal processor, e.g. a low pass filter. The input signal is applied to areas with detected detail in dependence on the derived value of the detail detector. The circuit includes a low pass filter (2) to suppress HF noise and a detail detector (4) to derive HF components in an image. If such HF components are not registered, a superimposing circuit (3) superimposes the signal from the low pass filter (2). If the detail detector (4) registers HF components with an amplitude greater than the noise amplitude, the original signal or a processed signal is superimposed.

Description

The present invention relates to a Video signal processing circuit for noise reduction of a Video signal, the video signal from a series in time consecutively transmitted video images, where a Video image from a number transmitted one after the other in time Video lines consist of one video line consisting of a number there are temporally successively transmitted pixels, whereby the video signal is a target signal and a for each pixel Has noise signal.

Noise reduction circuits are different Expression known. One method is e.g. B. as a coring technique known, the high-frequency components (in the line direction) can be evaluated with a non-linear characteristic in order to Low amplitude signal components (preferably noise) to suppress. Due to the non-linear evaluation, depending on the noise amplitude, however, noise disturbances in the image again break through or suppress fine details.

There are also recursive filters with image delays known, which are controlled by a motion detector. In addition to the high effort, there is the disadvantage that in Noise reduction is carried out in moving areas and Movement artifacts can occur.

Also have proven to be suitable means for noise reduction Multi-dimensional median filters are proven or highly complex signal-dependent algorithms.

Many of the known noise reduction methods require a high amount of hardware and also generate often additional interference components in the original signal are not included.  

From DE 24 13 799-A1 is one Video signal processing circuit known. There it will Video signal an adder and a changeable Combination circuit supplied, the changeable Combination circuit additionally one in time delayed video signal is supplied. Depending on the difference between undelayed and delayed video signal the changeable combination circuit directly that instantaneous video signal or a combination of delayed and undelayed video signal to the adder. The delay is preferably one image period, can also be a line duration or shorter. With this Circuit, a one-dimensional noise reduction is possible.

Circuits with a similar effect are e.g. B. from US 4 058 836-A or the article "Digital Techniques for Reducing Television Noise "by John P. Rossi, published in SMPTE-Journal, March 1978, vol. 87.

The object of the present invention is a Method and video signal processing circuit specify by means of a more efficient noise reduction is achievable.

The object is achieved by the method specified in claim 1 and by those specified in claims 8, 9 and 12 Circuits solved.

Advantageous process steps are in claims 2 to 7 specified. Advantageous further developments of Circuit designs in the remaining subclaims.

In a preferred embodiment of the circuit, this is Video signal at an input as an input signal in the Video signal processing circuit feedable. The Video signal becomes a basic detail detector and one Basic crossfade circuit supplied, the basic detail detector delivers a basic control signal via a basic control line the basic crossfade circuit, the basic crossfade circuit  provides an intermediate signal to an additional detail detector and an additional crossfade circuit is supplied which Additional detail detector delivers via an additional control line an additional control signal to the additional crossfade circuit Auxiliary crossfader provides an output signal that is processed video signal at an output of the Video signal processing circuit is tapped, the Intermediate signal is related to the input signal Reduced noise when the basic detail detector is on Input signal detected, its temporal fluctuation is low in terms of a basic time constant that The output signal is related to the intermediate signal Reduced noise when the additional detail detector is on Intermediate signal detected, its temporal fluctuation is small with respect to an additional time constant that Basic time constant is different from that Additional time constants.

This means that noise reduction is not just one thing, but one available in two dimensions. If necessary, can by expanding the video signal processing circuit by a third dimension also a three-dimensional one Noise reduction can be achieved.

Even before the circuit according to the invention Input signal can also be enhanced in detail, which also contributes to an increase in quality.

If the intermediate signal matches the input signal in essentially matches or regarding the Input signal is enhanced when the Basic detail detector detects an input signal whose temporal fluctuation with respect to the basic time constant large is, and the output signal with the intermediate signal in essentially matches or regarding the Intermediate signal is enhanced when the Additional detail detector detects an intermediate signal, the temporal fluctuation with regard to the additional time constants large  the circuit responds quickly Signal changes.

If the change from the noise-reduced intermediate or Output signal to that with the input or intermediate signal essentially identical intermediate or Output signal or to the input or Intermediate signal detailed intermediate or Output signal is stretched in time and / or stretched in relation to the signal level, avoided switching over again interference induced.

If the time constants of the detail detectors are determined in this way are that the pixels whose signals from the Detail detectors are processed for each detail detector Arranged linearly, the circuits are special simply set up. There are different ones Circuit combinations possible. For example it is possible that the pixels whose signals from Basic detail detector are processed in a video image horizontally next to each other, vertically one below the other or are arranged diagonally. It is also possible that the Pixels, the signals from the additional detail detector are processed, vertically one below the other in a video image or diagonally or in successive video images are arranged one behind the other.

If the detail detectors have noise level detectors and the input or the intermediate signal fluctuating slightly be classified if the temporal fluctuation of the Input or intermediate signal less than that is detected, there is an automatic Adaptation of the video signal processing circuit to the current noise level of the video signal.  

If the basic detail detector and the basic crossfade circuit or the additional detail detector and Additional crossfade circuit a common circuit part have, the effort to implement the Video signal processing circuit reduced. The circuit is therefore easier and cheaper to manufacture.

The circuit according to the invention combines a low Hardware expenditure with a good noise reduction without significant additional disturbances. In particular, in the case the planar processing every movement avoided.

Further advantages and details emerge from the following description of an embodiment and in connection with the drawings.

Show

Fig. 1 is a video signal processing circuit according to the invention,

Fig. 2 shows the effect of the circuit of Fig. 1 in the frequency domain,

Fig. 3 shows the operation of a further circuit of the prior art in the frequency domain,

Fig. 4 is an advanced video signal processing circuit according to the invention,

Fig. 5 shows the effect of the circuit of Fig. 4 in the frequency plane,

FIGS. 6 and 7 basic or auxiliary circuits of FIGS. 4 and

Fig. 8 is a block diagram of a noise-controlled detail detector.

A video signal usually consists of a series video images transmitted sequentially in time. Each Video image consists of a number in succession transmitted, arranged vertically one below the other Video lines. Each video line consists of a number successively transmitted, horizontally next to each other arranged pixels. Ideally, the video signal is there for each pixel only from the target signal. In practice assigns the video signal additionally for each pixel a noise signal. The elimination of this noise signal is the object of the present invention.

In the circuit according to FIG. 1, the video signal fed in at the feed point 1 is low-pass filtered in the low-pass filter 2 and the low-pass filtered video signal and the unfiltered video signal are supplied to the cross-fade circuit 3 . At the same time, it is checked in the detail detector 4 whether the video signal is rich or poor in details. Depending on the detection result, the cross-fade circuit 3 is then controlled so that the output signal of the cross-fade circuit 3 , which is present at the take-off point 5 , corresponds to the low-pass filtered video signal, the unfiltered video signal or a mixture of both signals. In front of the input signal or between the input signal and the crossfade circuit, there may additionally be a circuit for enhancing the details.

The input signal is therefore filtered with a low-pass filter in the circuit according to FIG. 1 in order to suppress higher-frequency noise. For this purpose, averaging over n pixels are preferably suitable, but also recursive low-pass filters, logical filters or median filters. A detail detector determines whether there are higher-frequency components in the image that have a larger amplitude and then generally exceed the noise amplitude. This is the case, for example, on high-contrast edges or in finely structured areas. At these detected points, the original signal is faded over to preserve the edge sharpness or fine structures. A soft crossfade has advantages over a hard switchover, since a hard switchover causes additional disturbances at the switchover points. Instead of fading to the original signal, fading to a processed signal is also conceivable, e.g. B. edge sharpening could be performed.

In the simplest case, the low-pass filter 2 and the detail detector 4 can be, for example, horizontally acting circuits that only require pixel enlargements. In this case there would be a horizontal noise reduction, which is shown in FIG. 2 in the frequency plane. Because of the non-linear processing, an exact analysis of the circuit in the frequency level is not possible, but the properties of the circuit can be estimated on the basis of sinusoidal input signals in the multi-dimensional frequency level. For this purpose, Fig. 2 shows two-dimensional cut-outs in the f _x, f _y, f _t -Frequenzebene, wherein the axis is not shown in each case was set to zero. The illustration relates to an interlaced scan with 625 lines and 50 Hz field frequency. Noise interference is reduced in the shaded areas. This is the case in the planar f _x , f _y frequency plane for low horizontal spatial frequencies, regardless of vertical spatial frequencies. In the temporal direction, the noise reduction is independent of vertical spatial frequencies, but is limited to low horizontal spatial frequencies.

With the one-dimensional structure for noise reduction to get good results already, but they will Image correlations in the other dimensions are not yet available Noise reduction used.

In contrast, FIG. 3 shows the effectiveness of a combination of detail detector 4 , low-pass filter 2 and cross-fade circuit 3 when low-pass filtering and detailed detection in a planar area, that is to say e.g. B. an area of 5 × 5 pixels. 3 as is immediately evident from the comparison of Fig. 2 with Fig., 3 is obtained in Fig. In a significantly lower frequency range of noise reduction as in Fig. 2, in spite of increased circuit scale.

In contrast, the circuit according to the invention shown in FIG. 4 consists of the series connection of two one-dimensional circuits. The video signal fed in at the input point 1 as an input signal is supplied on the one hand via the basic low-pass filter 6 or a comb filter and on the other hand directly to the basic crossfade circuit 7 . Furthermore, the input signal is fed to the basic detail detector 8 , which is described in more detail below, and whose output signal (= basic control signal) is supplied to the basic crossfade circuit 7 via the basic control line 9 . Depending on the signal that is transmitted via the basic control line 9 , the cross-fade circuit 7 supplies the unfiltered or the low-pass filtered input signal or a combination thereof as an intermediate signal. As a result, the intermediate signal is thus reduced in noise with respect to the input signal when the basic detail detector 8 detects an input signal whose temporal fluctuation with respect to a basic time constant τ 1 to be defined is small. Conversely, the intermediate signal coincides essentially with the input signal if the input signal detected by the basic detail detector 8 has a large temporal fluctuation with respect to the basic time constant τ₁.

As an alternative to providing the unfiltered input signal, the input signal could also be processed otherwise in the event of large temporal fluctuations, e.g. B. it could be detail-enhanced or a detail enhancement could take place between the input signal 1 and the cross-fade circuit 7 .

The intermediate signal is now supplied on the one hand directly and on the other hand via the additional low pass 10 or a comb filter of the additional crossfade circuit 11 . Furthermore, the intermediate signal is fed to the additional detail detector 12 , which checks the intermediate signal for whether it is rich or poor in details. Depending on the test result, the additional detail detector 12 supplies a corresponding control signal (= additional control signal) via the additional control line 13 to the additional crossfade circuit 11 . In accordance with this control signal, the additional crossfade circuit 11 supplies either the unfiltered or the low-pass filtered intermediate signal or a mixture of both signals as an output signal to the take-off point 5 . There the output signal can be tapped as a processed video signal.

The output signal is therefore related to the intermediate signal reduced noise when the temporal fluctuation of the Additional detailed detector τ₂ detected intermediate signal regarding an additional time constant to be defined τ₂ is low. Conversely, the output signal matches the Intermediate signal essentially coincides when the temporal Fluctuation of the intermediate signal with respect to the Additional time constant τ₂ is large. Alternatively to unprepared intermediate signal could also turn on other output signal are supplied, e.g. B. a detailed intermediate signal.

The time constants τ₁, τ₂ the detailed detectors 8, 12 are determined so that the pixels whose signals are processed by the detail detectors 8, 12 are arranged linearly to each minutiae detector 8, 12th The basic time constant τ₁ can, for. B. be chosen such that it corresponds to the time between two pixels arranged side by side. In this case, the pixels, the signals of which are processed by the basic detail detector 8 , lie horizontally next to one another in a video image. In the horizontal and / or vertical or temporal direction, it may be cheaper to take the distance between two or more pixels for τ₁. The spectral sensitivity can be changed via τ₁. If the basic time constant τ₁ corresponds to the time of a video line, the processed pixels in a video image are vertically below one another. If the basic time constant τ₁ corresponds to the time of a video line +/- the time of a pixel, the processed pixels are arranged diagonally. If the basic time constant τ₁ corresponds to the time required for the transmission of an entire video image, the processed pixels are arranged one after the other in successive video images.

What is said about the basic time constant τ₁ applies in the same way for the additional time constant τ₂. It is only necessary to ensure that the time constants τ₁, τ₂ are different from each other. The result is a circuit with only a small amount of hardware, which nevertheless leads to a good noise reduction without any significant additional disturbances. In particular, in the event that the basic time constant τ₁ corresponds to the time between pixels arranged next to one another and the additional time constant τ₂ corresponds to the time between two video lines arranged one below the other, any movement blurring is avoided. This case is described below in connection with FIG. 5 in the frequency plane.

Fig. 5 shows the effectiveness of the circuit of Fig. 4 in the frequency plane. The representation of FIG. 5 corresponds systematically to the representations in FIGS. 2 and 3. For the representation in FIG. 5, it was assumed that the basic time constant of the basic detail detector 6 for a horizontal detailed detection and the additional time constant of the additional detailed detector 12 for a vertical detail detection is designed. As can be seen from FIG. 5, noise is reduced in a considerably larger part of the frequency spectrum than in the circuits corresponding to FIGS. 2 and 3. However, the effort to implement the circuit according to FIG. 4 is only insignificantly greater than the effort to implement the circuit according to FIG. 1 and is comparable to a circuit that performs a two-dimensional noise reduction in the frequency range, as shown in FIG. 3. In addition, the circuit of FIG. 4 could also be supplemented by a third block acting in the temporal direction. In this case, at least one noise reduction would then take place in all areas of the three-dimensional frequency space.

In the double-hatched areas in FIG. 5, the noise reduction corresponds to that of planar filtering, since noise is uncorrelated during transmission in the horizontal and vertical directions and the noise reduction of the two adaptive filters is superimposed. In Fig. 5, there are also the hatched areas in which there is a one-dimensional reduction in noise interference.

The series connection is essentially one-dimensional Detectors and filters for noise reduction can continue for example, in a temporal or diagonal direction.

The detail detector, in contrast to the direct one nonlinear filtering (e.g. coring technique) a targeted Postprocessing of the detection signal with which the Properties of the overall circuit can be influenced positively can.

Fig. 6 shows an example of a possible circuit for the construction of the basic circuit. Referring to FIG. 6, the low-pass filter 6 on four delay elements 14. In each of the delay elements 14 , the input signal is delayed by one pixel. The output signals of the delay elements 14 and the undelayed input signal are fed to the summation element 15 . The output signal of the summation element 15 is fed to the cross-fade circuit 7 via the amplifier 16 .

The amplification of the amplifier 16 is selected such that it corresponds to the reciprocal of the number of signals fed to the summation element 15 . Since five input signals are fed to the summation element 15 in the present case, the amplification of the amplifier 16 is therefore 1/5 or 0.2.

In order to achieve the same average group delay times, the crossfade circuit 7 is also not supplied directly with the input signal fed in, but rather with the video signal delayed by two pixels, but otherwise unfiltered.

In the cross-fade circuit 7 , the low-pass filtered signal is supplied on the one hand directly to the summation element 17 and on the other hand via the amplifier 18 to the summation element 19 . The gain of amplifier 18 is -1. The amplifier thus acts as an inverter. The unfiltered video signal delayed by two pixels is fed to the summation element 19 as the second input signal. The output signal of the summation element 19 is fed back to the summation element 17 via the multiplier 20 . A second signal, the value of which lies between zero and one, is fed to the multiplier 20 via the control line 9 . If the value of the control signal is zero, the output signal of the crossfade circuit 7 , that is to say the intermediate signal according to FIG. 4, is equal to the low-pass filtered signal. If the value of the control signal is one, the summation element 17 supplies the pure video signal, which is only delayed by two pixels. If the value of the control signal is between zero and one, a mixture of both signals is output. The change of the control signal from zero to one and vice versa is therefore stretched in relation to the signal fluctuation.

The control signal is also changed from zero to One and vice versa preferably also stretched in time, because this causes interference at the switching points, which is a hard one Switching would be avoided.

The multiplier 20 is controlled via the detail detector 8 , to which the undelayed input signal and on the other hand the input signal delayed in this case by two pixels are supplied as input signals. Low pass filter 6 and detail detector 8 therefore share the first two delay elements 14 of the low pass 6 . One of the two signals fed to the detail detector 8 is inverted in the amplifier 21 before the signals are added in the summation element 22 . The output signal of the summation element 22 is fed to the absolute value generator 23 . The output signal of the magnitude generator 23 is fed directly to the summation element 25 and then to the amplifier 26 on the one hand and on the other hand via the delay elements 24 , which delay the signal by one pixel each. The amplification of the amplifier 26 again corresponds to the reciprocal of the number of signals supplied to the summation element 25 , in this case it is 1/3 or 0.33. The output signal of the amplifier 26 is evaluated in the evaluator 27 in accordance with a non-linear, symbolically represented characteristic curve. The output signal of the evaluator 27 then provides the control signal for the multiplier 20 . A subsequent temporal expansion of the control signal is still possible via z. B. a low pass.

An improvement in the signal-to-noise ratio of approximately 7 dB can be achieved by means of the circuit shown in FIG. 6. In contrast, only an improvement of almost 5 dB can be achieved with the circuit according to FIG. 7. However, the effort to implement the circuit according to FIG. 7 is considerably less. The same reference numerals in FIG. 7 mean the same elements as in FIG. 6.

The circuits shown in FIGS. 6 and 7 can be transferred directly to the additional circuit. It is only necessary to adjust the time constant. The circuit of FIG. 7 is particularly advantageous for vertical noise reduction, since only three line memories are required as delay elements for the entire circuit.

In addition, it should be noted that instead of averaging Other low-pass filters, e.g. B. recursive low-pass filters, logical filters or median filters, could be used.  

The evaluators 27 according to FIGS. 6 and 7 are static. A further improvement of the video signal processing circuit can therefore be achieved if the evaluators 27 are dynamized in a suitable manner. One possibility for dynamization is to supply the output signal of the magnitude generator 23 to a noise level detector 28 and to automatically determine the noise component of the video signal there, as can be seen from FIG. 8. Depending on the output signal of the noise detector 28 , either the characteristic curve of the evaluator 27 can then be varied directly or one of several evaluators 27 with different characteristic curves can be controlled via the switch 29 . The construction of noise detectors is known per se. By way of example only, reference is made to the unpublished German patent application P 43 41 760.4. As a result, the input or the intermediate signal is classified as slightly fluctuating if the temporal fluctuation of the input or the intermediate signal is less than the actually detected noise level. The noise reduction can thus be further optimized.

The noise component can be determined directly from the Video signal done, but at this point you can circuits of the detail detector are also advantageously used become. In this example, the noise component after the Determination of amounts; after low pass filtering that would be also possible. Depending on the noise voltage, there is a Reversal of the transfer characteristic, switchable here shown. In the top switch position (Transfer characteristic T₁ = 1) - if only a very small one Noise amplitude was determined -, the detail detector switched off and there is no noise reduction. Since that Noise is generally independent of direction and uncorrelated the determined noise amplitude can also be used for other Tax purposes are brought out.

This circuit takes into account the fact that an integral part of the detail adaptive Noise reduction is the detail detector that controls  between the original signal and the noise-reduced signal takes over. In pictures with a high signal-to-noise ratio of course no noise reduction required, too strong disturbed signals, however, could be the transfer characteristic be less sensitive to fine details. Instead of one A compromise in the choice of the transfer characteristic can be a noise-dependent control of the detail detector - in particular the transfer characteristic - take place.

Reference list

1 entry point
2 , 6 , 10 low passes
3 , 7 , 11 crossfades
4 , 8 , 12 detail detectors
5 delivery point
9 , 13 control lines
14 , 24 delay elements
15 , 17 , 19 , 22 , 25 summation terms
16 , 18 , 21 , 26 amplifiers
20 multipliers
23 amount calculator
27 evaluators
28 noise detector
29 switches
f _x , f _y , f _t frequencies
τ₁, τ₂ time constants.

Claims (22)

1. A method for reducing the higher-frequency noise in a video signal when playing on a screen of a video system, characterized in that a detailed detection of the video signal in areas of a field or frame is carried out without pronounced details and depending on the determined signal of the non-pronounced detail Fader circuit ( 7 ) is controlled in which the signal is blended in the areas without pronounced details of a signal which is used by a signal processing circuit for noise reduction, e.g. B. a low-pass filter of the signal is derived, and that in areas with detected details, the input signal is applied depending on the determined value of the detail detector. 2. The method according to claim 1, characterized characterized that in areas with detected Details instead of a signal on the input signal that contains the details. 3. The method according to claim 2, characterized characterized in that the input signal or the details contained signal is amplified.   4. The method according to claim 1, 2 or 3, characterized by the determination of the Details in the lines of an image with a Signal processing circuit. 5. The method according to any one of the preceding claims, characterized in that after the Blending the signals of the signal processing with the Input signals the received output signal a vertical Detail detector is supplied, which is a function of vertical jumps with another crossfade in areas with little detail on a vertical filter Noise reduction fades and in the case of detected details the input signal crossfades, which second Crossfade circuit depending on the through the Vertical detail detector controlled fading factor controlled becomes. 6. The method according to any one of the preceding claims, characterized in that the Crossfading between the detected points on the Original signal is soft. 7. The method according to any one of the preceding claims, characterized in that a multidimensional signal processing (planar and / or temporal) for the noise reduction in a filter circuit from at least two essentially one-dimensional adaptive filters. 8. Circuit for performing a method according to previous claims, thereby characterized in that a low-pass filter for Suppression of higher frequency noise is provided that a detail detector is provided, the higher frequency components determined in the picture that in the event that such higher frequencies Shares are not found, the crossfade the signals filtered by the low-pass filter, and that in the Case that the detail detector has higher frequency components  notes that have amplitudes greater than that Are noise amplitude, the original signal or a processed signal is blended. 9. Circuit for performing a method according to one of the Claims 1 to 7, characterized characterized in that a noise reduction multidimensional planar or temporal Signal processing circuit out essentially one-dimensional filter circuits. 10. Circuit for performing a method according to claim 8 or 9, characterized in that after the horizontal fade, a vertical one Cross-fade control takes place and that a vertical low pass is provided for noise reduction, and that the fade between the original signal and the noise-reduced signal depending on the output signal of the vertical detector is controlled. 11. Circuit according to claim 10, characterized characterized that a third Signal processing stage for diagonal noise reduction it is provided that a diagonal low-pass filter and a Diagonal detail detector are provided, and that the further Crossfade depending on the determined Noise signal a transition from the input signal to the noise-free signal or to the original signal. 12. Video signal processing circuit for performing the method according to claim 1 to 7, wherein

• the video signal consists of a series of chronologically transmitted video images,
• a video image consists of a number of video lines transmitted one after the other in time,
• a video line consists of a number of successively transmitted pixels,
• the video signal has a target signal and a noise signal for each pixel,

with the following features:

• the video signal can be fed as an input signal into the video signal processing circuit at an input ( 1 ),
• - The video signal is fed to a basic detail detector ( 8 ) and a basic cross-fade circuit ( 7 ),
• - the basic detail detector (8) provides a basic control line (9) a base control signal to the basic cross-fading circuit (7),
• - The basic crossfade circuit ( 7 ) supplies an intermediate signal which is fed to an additional detail detector ( 12 ) and an additional crossfade circuit ( 11 ),
• - The additional detail detector ( 12 ) delivers an additional control signal to the additional crossfade circuit ( 11 ) via an additional control line ( 13 ),
• - The additional crossfade circuit ( 11 ) provides an output signal that can be tapped as a processed video signal at an output ( 5 ) of the video signal processing circuit,
• - The intermediate signal is reduced in noise with respect to the input signal when the basic detail detector ( 8 ) detects an input signal whose temporal fluctuation with respect to a basic time constant (τ₁) is small,
• - The output signal is reduced in noise with respect to the intermediate signal when the additional detail detector ( 12 ) detects an intermediate signal, the temporal fluctuation with respect to an additional time constant (τ₂) is small,
• - The basic time constant (τ₁) is different from the additional time constant (τ₂).

13. Video signal processing circuit according to claim 12, characterized in that

• - That the intermediate signal coincides with the input signal substantially or is enhanced in detail with respect to the input signal when the basic detail detector ( 8 ) detects an input signal whose temporal fluctuation with respect to the basic time constant (τ₁) is large, and
• - That the output signal coincides with the intermediate signal substantially or is enhanced in detail with respect to the intermediate signal when the additional detail detector ( 12 ) detects an intermediate signal, the temporal fluctuation with respect to the additional time constant (τ₂) is large.

14. A video signal processing circuit according to claim 13. characterized in that the change from the noise-reduced intermediate or output signal that with the input or intermediate signal essentially matching intermediate or output signal or to the regarding the input or intermediate signal detailed intermediate or output signal in time stretched.   15. Video signal processing circuit according to claim 13 or 14, characterized, that the change from the noise-reduced intermediate or Output signal to that with the input or intermediate signal essentially identical intermediate or Output signal or to the input or Intermediate signal detailed intermediate or Output signal is stretched in relation to signal fluctuation. 16. Video signal processing circuit according to one of claims 12 to 15, characterized in that the time constants (τ₁, τ₂) of the detail detectors ( 8 , 12 ) are determined such that the pixels whose signals are processed by the detail detectors ( 8 , 12 ), are arranged linearly for each detail detector ( 8 , 12 ). 17. Video signal processing circuit according to claim 8, 9, 10, 11 or 16, characterized in that the pixels, the signals of which are processed by the basic detail detector ( 8 ), are arranged horizontally next to one another, vertically one below the other or diagonally in a video image. 18. Video signal processing circuit according to claim 10, 11, 16 or 17, characterized in that the pixels, the signals of which are processed by the additional detail detector ( 12 ), are arranged vertically one below the other or diagonally or in successive video images in a video image. 19. Video signal processing circuit according to one of claims 8 to 12, characterized in that the cross-fade circuits ( 7 , 11 ) each have a comb filter or a low-pass filter ( 6 , 10 ) for generating the noise-reduced signals. 20. Video signal processing circuit according to one of claims 8 to 19, characterized in that the detail detectors ( 8 , 12 ) have noise level detectors ( 28 ) and that the input or the intermediate signal are classified as slightly fluctuating if the temporal fluctuation of the input or the intermediate signal is less than the detected noise level. 21. Video signal processing circuit according to one of claims 8 to 19, characterized in that the basic detail detector ( 8 ) and the basic crossfade circuit ( 7 ) or the additional detailed detector ( 12 ) and the additional crossfade circuit ( 11 ) have a common circuit part ( 14 ).