MXPA97008123A - Luma / chrome separation filter with derressive element co - Google Patents

Abstract

The present invention relates to a method for reducing noise and separating the components of a composite video input signal, comprising the steps of: combining said composite video input signal with a noise reduction signal to form a first reduced noise video signal, delaying said reduced noise video signal to form a plurality of delayed video signals having different delays respectively, forming said combined noise reduction signal said composite video input signal with a selected pair of said delayed signals, and combining said first reduced noise video signal with an additional one of said delayed video signals to form chrominance component and reduced noise luminance separated signals.

Description

FILTER OF SEPARATION OF LUM A / CROMA WITH ELEMENT OF COMMON DELAY DESCRIPTION OF THE INVENTION This invention relates generally to circuits for separating luminance and chrominance components from composite video signals and particularly to signal separation circuits of the type employing comb filters. In video apparatus, such as television receivers or monitors, video recorders, video recorders of discs / reproduction units, and the like, there is often a need to separate a composite video signal into luminance and chrominance components. A simple way to provide such separation is to filter in low pass the composite signal to obtain a luminance component and filter in high pass or band pass, the composite signal to obtain the chrominance component. This technique, however, does not efficiently recover the components and results in a loss of sharpness in presented images. Since the composite components are spectrally interspersed, the comb filtration provides a more efficient separation and thus a clearer presented image. The advantages of the comb filter separation of the luminance (Y) and chrominance (C) components of a video signal are well known. In the most elementary "one line" or "1-H" form of the comb filter, the image elements (hereinafter referred to as "pixels") temporarily separated to a line are added to produce a separate luminance component. and subtracted to produce a separate chrominance component. Such filters provide superior image details, as compared, for example, with the low pass / high pass filter method of the Y / C separation, but tend to exhibit visual artifacts (e.g., hanging dots) for certain characteristics of images. Some two-line (2-H) comb filters provide improved performance with respect to the 1H comb filter described above, "adapting" the filter to the detail of the image and thus obtaining a desirable reduction in visual artifacts. In principle, this is done by combing the composite signal twice in the vertical direction to produce two combinated chrominance signals and then selecting them or "mixing" them with a "soft switch" based on the analysis of image characteristics to select the signal ( or "mixing" signals) that have the latest visual artifacts. The chrominance signal thus produced is then subtracted from the composite signal to provide a separate luminance output signal. This form of chrominance / luminance signal separation is commonly known as "2-H" or combed in two lines and provides reduced visual artifacts as compared to 1-H combing and improved image detail as compared to the filtration method low pass / high simple step of Y / C separation. A 2-H comb filter Y / C separation circuit is described, for example, by McNeely and Willis in the US patent. No. 4,786,963, entitled Adaptive Y / C Separation Apparatus For TV Signals (Adaptable Y / C Separation Apparatus for TV Signals), which was issued on November 22, 1988. The present invention is directed to satisfy the need of the noise reduction in a television set employing line comb filters for luminance / chrominance signal separation. A method, according to the invention, for reducing noise and separating the components of a composite video input signal, comprises combining the composite video input signal with a noise reduction signal to form a first reduced video signal. of noise; delaying the first reduced noise video signal to form a plurality of delayed video signals having respectively different delays; forming the noise reduction signal by combing said composite video input signal with a selected pair of said delayed video signals; and combing the first reduced noise video signal with one more of the delayed video signals to form chrominance output signals and reduced noise luminance. The apparatus, according to the invention, for reducing the noise and separating the luminance and chrominance components of a composite video input signal, comprises a source for providing the composite video input signal and first circuit means, coupled to the source, to combine the composite video input signal with a noise reduction signal thereto to form a first video signal of reduced noise. Second circuit means, coupled to the first circuit means, delays the first reduced noise video signal to form a plurality of delayed video signals having respectively different delays. Third circuit means, coupled to the source and the second means, comb the composite video input signal with a selected pair of delayed video signals to form the noise reduction signal for the first circuit means. Fourth circuit means, coupled to the first and third circuits, comb the first reduced noise video signal with one more of the delayed video signals to form chrominance output and reduced noise luminance signals. The foregoing and other aspects of the invention are illustrated in the accompanying drawings, wherein like elements are denoted with like reference numbers, and wherein: Figure 1 is a block diagram of a television receiver including a signal separator of luminance / chrominance modalizing the invention; Figure 2 is a block diagram illustrating a modification of the receiver of Figure 1; Figure 3 is a block diagram illustrating another modification of the receiver of Figure 1; Figure 4 is a diagram illustrating the comb filtration in the receivers of Figures 1, 2, and 3; Figures 5 and 6 are average circuit block diagrams suitable for use in the examples of Figures 1,2, and 3; Figure 7 is a block diagram of a "soft" or "mixed" switch suitable for use in the example of Figure 3; Figures 8 and 9 are limiting circuit transfer function diagrams suitable for use in the examples of Figures 1, 2, and 3; Figures 10A and 10B are block diagrams of limiter and core circuits for providing the transfer function of Figure 9; Figure 11 is a spatial diagram illustrating line and pixel relationships (smaller element in the video display screen) in operation of the example of Figure 3, when transverse gradient processing is being used to control the mixture; Figure 12 is a block diagram of a transverse gradient processor suitable for use in the example of Figure 3; and Figure 13 is a block diagram illustrating a modification of the apparatus of Figure 1. The luma / chroma separation filters of the present invention are of general utility and can be used in any normal video signal processing application. NTSC that requires the separation of a composite video signal to its luminance and chrominance components and where noise reduction is desired. Advantageously, as explained below, the separator of the present invention reduces the noise in both the chrominance and luminance components of the separated signals. In addition, the noise reduction provided includes a two-dimensional spatial effect. Specifically, the main noise reduction is provided in the vertical direction of both the luminance and chrominance components, and is provided through an IIR (infinite impulse response) or "repetitive" filter. The additional noise reduction in the horizontal direction is provided by an FIR comb filter (finite impulse response) that is "nested", to talk like that, inside a feedback loop of the IIR filter. A further advantage of the complete filter is that there is a desirable participation of delay elements for the noise reduction and separation functions. Actually, the participation of elements is so efficient that the benefits of two-dimensional noise reduction are achieved with substantially the same amount of memory as could otherwise be required to provide only the signal separation function without any noise reduction. An important technical advantage of combining the noise reduction and separation functions, according to the invention, is that a synergistic effect is received, where the reliability is improved. This improvement is due to a reduction in the parts account as compared, for example, with the separate noise reduction and cascade separation filters. In the examples of the invention, which are presented below, those of Figures 1 and 2 illustrate two different examples of the invention employing a main delay of a horizontal line ("1-H" hereafter). The example of Figure 3 is more complex, requiring a major net delay of about two horizontal lines ("2-H" hereafter) in an "adaptive" comb filter configuration, which desirably suppresses certain artifacts characteristic of conventional 1-H comb signal separators (for example, "hanging points"), which may occur under certain image conditions. The television receiver 10 of Figure 1 comprises a tuner unit 12 for tuning an RF input signal S1A and providing a composite video output signal of the baseband S2. A switch 14 is provided for selecting the baseband composite video signal S2 or an auxiliary baseband video input signal S1B to provide a composite video signal source of baseband S3, which is processed, as will be described, to be presented through a kinescope 18 (or other suitable display device, such as a CCD screen). To process signal S3 for presentation, receiver 10 includes a luminance / chrominance signal separator 20A (imaginatively highlighted) having an input 22 for receiving composite video signal S3 and outputs 24 and 26 for providing, respectively, a reduced noise luminance output signal S14 and a reduced noise chrominance component video signal S13. The components S14 and S13 are applied to a luminance / chrominance matrix and processing unit 16, which provides conventional functions such as brightness and contrast control, hue and saturation control, and the like and generates a video output signal of RGB component for the kinescope display unit 18. The signal separation and noise reduction unit 20A, in this example of the invention, is of digital construction and includes an analog to digital converter 23 timed to four times the sub clock -porter 25 (ie, 4-fsc sampling) to convert the composite video signal S3 to the digital form (S4) with a resolution of, illustratively, eight bits per sample. It will be understood that the invention can be well practiced using analogous components and if analogous processing is performed then the clock 25 and the converter 23 can be omitted. Also, in this example of the invention, it is assumed that the Y / C matrix and processing circuit is of the digital type and for this reason no digital to analog converters are required for the component output signals S 13 and S 14. Said output converters can be added, of course, if the processing and matrix unit is of the analogous type which requires analog input signals. For normal NTSC video signals, the use of a 4-fsc sampling rate produces a signal that has 910 samples per horizontal line and 4 samples per full-color sub-carrier cycle (360 degrees), which corresponds to two clock pulses or samples per half cycle (180 degrees) of the color subcarrier. These parameters must be appropriately adjusted in cases where a different sample rate is selected (such as a 3-fsc rate or three times the frequency of the color sub-carrier) or where the signal S3 is of the transmission format PAL that has a different line rate and a color subcarrier frequency. After conversion to the digital form, the composite video signal S4 is combined by an adder 34 with a noise reduction signal S5 to form a first video signal of reduced noise S6, then the signal S6 is applied to a multiple delay circuit comprising delay units 38, 42 and 40, which are connected in cascade, in the named order. The unit 38 provides a delay of a minor half cycle of the horizontal line of the color subcarrier period. In terms of digital samples, this delay corresponds to 910 samples (full line) minus 2 clock periods (180 degrees) for a net time delay in unit 38 of 908 samples or clock periods. The units 42 and 40, both have half-cycle time delays of the color sub-carrier, which, for this digital mode at a given sample rate (4-fsc), each corresponds to 2 clock periods. Since the delay units 38, 42 and 40 are connected in cascade, the net time delays are 908 clocks, 910 clocks and 912 clocks. The multiple delays thus provided by units 38, 42, and 40 in this manner correspond to multiple periods of (i) one line minus one medium color cycle, (ii) one line exactly, and (iii) one line plus half cycle color. The delayed signals are identified as those of the video signals S7, S8, and S9 respectively. After the formation of the multiple delays described above, the shorter S7 and the longer S9 are averaged in an averager 36 to provide an averaged S10 signal. The averaged value of signal S10 is equal to the sum of signal S9 plus signal S7 by two (for example, [S7 + S9] / 2). Figures 5 and 6 provide examples of suitable analog and digital signal averagers. In Figure 5 the analog signals S7 and S9 are combined in an adder circuit 500, the output of which is attenuated by 6 decibels in an attenuator 502. Figure 6 is similar, except that the signals are digital and are added in an adder digital 600 and are divided by two in a divisor 602. As a practical way, the division between two in a binary digital processing does not require "hardware". All that is required is a 1-bit offset of the binary point, which can be achieved by lowering the output of LSB (least significant bit) of the adder. The averaged signal S10 provided by the averager 36 is used to form the noise reduction video signal S5 by subtractively combining the composite video input signal S4 with the averaged signal S10 in subtractor 30 and limiting the resulting difference signal S11 with a limiter 32. Figures 8 and 9 are examples of suitable limiter transfer functions. In Figure 8, the limiter 32 exhibits a linear transfer function 802 (i.e., a constant gain) between the negative limitation level 800 and the positive limiting level 804 of the output signal S5. The slope "m" of the linear region is selected to be less than unity (m <1) to ensure convergence to zero in the absence of an input signal, as will be explained later. A preferred gain scale is between 0.5 and 0.9. The higher gain results in more pixels on average and in this way, more noise reduction, but an increased delay or adaptation to change the scene content. The lower gain provides a faster response at the expense of fewer pixels that are averaged for noise reduction purposes. The preferred scale of 0.5 < m < 0.9 represents a desirable compromise between noise reduction efficiency and response to scene changes. Figure 9 illustrates a preferred limiter transfer function, which includes a "dead zone" or "core" region 904 between the linear regions 902 and 906 of the limiter transfer function. As in the example of Figure 8, the slopes ("m" values) of the linear segments 902 and 906 of the transfer function are less than or equal to unity and, preferably, are in the range of 0.5 to 1.0. Advantageously, the core region 904 provides a further improvement of the noise immunity in addition to the previously discussed vertical and horizontal noise reduction effects of the entire system. Essentially, the "core" region suppresses small signal perturbations so it maintains said "fine detail" or recirculation noise in the repetitive filtering loop. The core aspect can be added to the example of Figure 1 either by direct implementation in the limiter or by connecting the cascade limiter with the core formed. The last aspect is shown in Figure 10A, where the core former 1000 is connected in series with the limiter 1002. The option of direct implementation of the core in the limiter can be implemented as shown in Figure 10B using a ROM (access-only memory) 1004 for the limiter and storing the transfer function of Figure 9 in the ROM. During operation, of the portion of the separator 20A discussed above, the output S11 of the subtractor 30 represents the difference between the video input signal S4 and the reduced noise averaged video signal S10 of the previous line. Accordingly, the coherent signal components of S4 and S10 tend to cancel and the difference between the respective noise components of S4 and S10 appear as S11 at the output of the subtractor. The phase of this noise difference signal is opposite to that of the input signal noise signal due to the subtraction and thus when S11 is finally added to S4 in the adder 34, there is a reduction in the noise level S6. The reduced signal of noise S6 is delayed by a line and feedback to form S10 and the noise reduction procedure repeats recursively with the additional noise that is being removed from each new received line. To ensure that the recirculation video signal S10 is finally reduced to zero when the input signal S4 is set to zero, the gain of the limiter 32 is selected to be less than unity, as discussed previously. The main function of the limiter 32 is to minimize the effects of the noise reduction system on the vertical detail of the reduced noise signal S6. Evoke that the subtractor 30 compares a present pixel with an averaged one of a previous line. If there is a horizontal line in the presented image, then the pixels on both sides of the line will differ greatly in amplitude and the difference signal S11 will often be as large as the noise difference component intended to represent. The limiter 32 prevents large values of the signal S11 from reaching the adder 34. The limit level is, however, high enough to allow the noise to pass through the adder 34 to achieve noise reduction. Illustratively, for this purpose, a limit level on the fair scale of some units of IRÉ signal level is desirable. It has been found in visual tests that limiting levels around 2 or 3, IRÉ units provide adequate noise reduction with acceptably low visual artifacts. In addition to limiting the noise reduction signal S5, as discussed above, it has also been found desirable to form a core to the noise reduction signal, as discussed above, with respect to Figures 9, 10A and 10B. the core formation advantageously avoids the removal of vertical details of low amplitude of the presented images by blocking the passage of small values of the difference signal S11. A level of core training adequate for this purpose must be less than the value of the limit level. Illustratively, a core training level of approximately 1 IRÉ unit has been found adequate (or 1 or 2 LSBs in a digital system) with a limit on a scale of 2 or 3 IRÉ units. The averager 36 performs an important part of the repetitive noise reduction loop, since it determines the phase of the chrominance component that is recirculated within the repetitive loop (feedback) to allow the noise reduction system to be used with composite video signals . Emphasize that the object of the present invention is to concurrently separate and reduce noise in composite video signals. However, in composite NTSC video signals there is not an entire ratio of color cycles per line. Therefore, if the feedback signal S10 were exactly delayed by a line with a conventional 1-H delay, the color phase would be inverted in the feedback signal caused by a low amplitude chroma to drive the limiter to the limitation even when no detail is present. In view of the foregoing, a correction is made in the color phase of the feedback signal S10 by averaging a short line (i.e., a line minus one half of a color cycle) with a longer line (i.e. line plus one half of a color cycle). The resulting signal S10 is delayed on average by one line, which is necessary for a good repetitive noise reduction, and the average has produced a color phase in the feedback signal that coincides with the color phase of the incoming signal. . Consequently, the color component of the signal is reduced in noise, as well as the luminance signal component. An additional advantage of the average of short and long lines to produce the delayed line feedback signal S10, is that a horizontal filter is formed by two half-cycle delays and the averager 36. The effect is to remove the additional high-frequency noise, In this way, noise is reduced by filtering in two dimensions, horizontally by the averaging procedure and vertically by repetitive filtering. The remaining elements of the separator 20A in Figure 1 comprise two subtractors 50 and 56 and a chroma pass filter 52, which forms a "combination circuit". This is the output portion of the separator, which combines the reduced noise video signal S6 provided by the adder 34, with the reduced noise and delayed video signal of a line S8 provided at the output of the delay unit of half-cycle 42 to produce the luminance and chrominance signal components S14 and S13. Specifically, the line delayed signal S8 is subtracted from the signal S6 in the subtractor 50 and the resulting difference signal is filtered by bandpass in the chroma bandpass filter 52 to produce the separate chrominance component S13.

The line delay 38, the half-cycle delay 42, the subtractor 50 and the chroma bandpass filter 52 form a chrominance comb filter 1-H (one line) that produces the separated reduced noise chroma signal. S13, as noted above. The amplitude response or transfer function for this filter is illustrated in Figure 4, through the amplitude characteristic 402. The width of the pass band of the chrominance signal is determined by the bandpass filter 52. Within the chrominance pitch band, recurrent amplitude peaks occur at frequencies, which are odd multiples of half the horizontal line regime. The separate luminance component S14 is produced by the subtractor 56, which subtracts the separate chrominance component S13 from the reduced noise non-delayed video signal S6. As shown in Figure 4 by the amplitude response 400, it produces an amplitude response with a unit gain for all luminance components below the lower band edge of chroma filter 52 and produces a signal response of luminance that is filtered in a comb filter through the frequency scale of the bandpass filter 52. The peaks of the luminance response occur in multiples of the horizontal line regime and coincide with the valleys of the signal response of chrominance as illustrated by the response characteristics 400 and 402. Figure 2 illustrates a modification of the separator 20A in the example of Figure 1, where a different comb 1-H filter filter configuration is used to combine the signals output S6 of the adder 34 with the output signal S8 of the half cycle delay unit 42 to form the separate output signals S13 and S14. As will be explained, the resulting reduced noise combined responses for luma and chroma are the same as shown in Figure 4 for the embodiment of Figure 1, although, in the example of Figure 2, the luma component is combined through of the entire luminance signal as an intermediate step in processing. In detail, in the separator 20B of Figure 2, the combination elements of the output signal 50, 52, and 56 of Figure 1 are replaced by two adders 200 and 206, a subtractor 202, a low pass filter 204. and a chroma bandpass filter 208. The separated chrominance signal is produced in the same manner as in the example of Figure 1, mainly, the subtractor 202 subtracts the delayed reduced noise signal of line S8 from the signal of reduced non-delayed noise S6 and the resulting signal S12 is filtered by a band pass filter, by the filter 208, to provide the reduced noise chrominance signal S13. The response of the filter is shown by the response 402 in Figure 4 and is the same as the previous example. The difference between the separators 20A and 20B of Figures 1 and 2 is in the separation of the luminance signal component. Emphasize that in the separator 20A of Figure 1, the luminance component was obtained by subtracting the separated chrominance output signal S13 from the reduced noise composite video signal S6 producing the response of the comb filter 400 of Figure 4. In the separator 20B of Figure 2, the luminance component is separated by the adder 200, which adds the composite video signal S6 to the delayed composite video signal of line S8. This forms a line comb filter, in which the resulting luminance signal S15 is combined across its entire bandwidth. Since the luminance region below the chroma band conveys the vertical detail, the combined luminance signal S15 lacks vertical detail. However, the combined vertical detail is available at the output of the subtractor 202 and is separated from the chrominance component through a low pass filter 204, which passes the signal components below the chrominance band. The resulting vertical detail signal S16 is restored to the luminance signal via the adder 206 thus replacing the vertical detail lost due to the combination at the output of the adder 200 and producing the luminance video output signal S14. As shown in Figure 4 by the luma 400 response, the luminance component is combined only in the upper region in the chroma band and there is no loss in vertical detail. As previously noted, the same total result exists in the example of the separator 20A, but with a different circuit topology. In the previous examples of Figures 1 and 2, delays 38 and 42 were shown to provide 100% of the delay required for signal separation and 99.79% of the delay required for vertical and horizontal noise reduction (ie, 910 out of 912 delayed clock cycles). Accordingly, by "sharing" these two specific circuit elements, the addition of noise reduction to a 1-H comb filter (as shown in Figures 1 and 2) only requires 2 additional clock delays, out of 910 that it could be otherwise required for comb filter signal separation of a line. In this way, in terms of video signal memory requirements, one can add noise reduction to the comb filter separators of a line at a modest memory cost, only requiring approximately two tenths of a percent (i.e. , 0.2%) of more memory (delay) than a comb separator of a line without noise reduction. As explained above, 2- H comb separators provide more efficient separation with fewer visual artifacts as compared to single-line comb filter separators. In the example of Figure 3, as will be described, it will be shown that the benefits of noise reduction and comb separation of two lines can be obtained with an even higher memory efficiency. Specifically, for the comb separator 2-H shown, with sample at an assumed rate of four times the frequency of the color subcarrier, 1820 pixe or delay clock cycles are required for line combining purposes and 6 pixe Delay are used for processing "transverse gradient" giving a total of 1826 pixels for the comb separation function without noise reduction. With noise reduction, only two additional pixels of delay (corresponding temporarily to half a color cycle) are required over the 1826 needed for 2-H separation. In the separator 20C of Figure 3, the reduced noise composite video signal S6 and the delayed reduced noise composite video signal S8 are combined to form the separate component signals S13 and S14 through the elements marked 302 a 318. As a brief summary, these elements form, in effect, an adaptive comb filter that combines a given pixel with a mixture of pixels or "mix" of an anterior line and a posterior line. The mixture or "combination" of the pixels, in turn, is controlled by a switch control unit 308, which analyzes diagonals of a pixel array (see Figure 11) to produce a chrominance signal S21 having visual artifacts. reduced. The luminance signal, S14, is separated by subtracting the recovered chrominance signal S21 from the video signal composed of reduced noise delayed on line S8. The control of the mixture is by diagonal gradient measurements from line to line or "transverse gradients", as described in the patent of E.U.A. 4,786,963 of McNeely and Willis. To complete, Figure 12 of this provides details of the switch control signal generator or "mixed" signal generator from McNeely et al. In Figure 3, the video signal composed of reduced noise S6 provided by the adder 34 and the delayed reduced noise composite video signal of a line are applied to respective bandpass filters 306 and 302 having bandwidths inclusive of the chrominance signal band (illustratively, 1.0-4.2). MHz). The filter 306 thus produces a band-limited signal Bb corresponding to the bottom line of the pixels g, hei of Figure 11. The filter 302 produces pixels corresponding to ad, e and f, which occur at a line earlier than the line. line N and thus form the middle line or Mb signal. To provide the pixels a, b, and c for the line N-2 (the upper line of Figure 11), the signal Mb of the filter 302 is delayed by a full line in a line delay unit 304. The separated chrominance signal is formed through a subtractor 312, which subtracts either the upper line signal Tb or the bottom line signal Bb or a mixture of the two from the Mb video line signal, thus forming an adaptive comb filter. The adaptation of this filter is controlled by the switch control unit 308, which receives signals Tb, Mb and Bb, analyzes its diagonal differences and generates a control signal K to control the mixer switch 310. Briefly stated, the unit control 308 analyzes "ai" pixels to find the best combination of N and N-2 lines to be combined with the N-1 line in order to reduce visual artifacts such as hanging dots. In the McNeely et al. Apparatus, this analysis is based on diagonal measurements of the nine pixels, a-i, shown in Figure 11 through a processor shown in Figure 12 of the present, and discussed below. The image analysis provided by unit 308 is described through the following three relationships: XU = MAX. { ABS (a-f), ABS (c-d)} (1) XL = MAX. { ABS (d-i), ABS (f-g)} (2) K = XL / [XL + XU] (3) In the expression (1) above, XU is defined as a "superior transverse gradient" and, with reference to Figure 11, lines N-2 and N-1, this gradient is equal to the maximum (MAX) of two functions, mainly , the absolute value of the pixel difference af and the absolute value of the pixel difference cd. This function, in other words, assigns a value to the larger of the two diagonal gradients a-f and c-d that occur between the upper and middle lines, N-2 and N-1. In expression (2) above, XL is defined as a "lower transverse gradient" and, with reference to Figure 11, lines N-1 and N, this gradient is equal to the maximum value (MAX) of two functions, mainly, the absolute value of the pixel difference di and the absolute value of the pixel difference fg. This function, in other words, assigns a value to the largest of the diagonal gradients di and fg that occur between the middle line N-1 and the bottom line N. The decision to mix is made according to equation 3, the which provides that K, the control signal, is set equal to the value of the lower transverse gradient divided by the sum of the two transverse gradients. For example, if XL is zero (indicating no difference between the middle and background lines) and XU is different from zero, then K will be equal to zero and will cause the switch 310 to select the bottom line signal Bb for the subtraction of the midline signal in the subtractor 312 thus producing the combined chroma signal S21 (after scaling by 6 dB provided by the divider 314). Conversely, if the lower gradient XL is not zero and the upper gradient is zero, K will be equal to the unit or "1". This indicates that the upper and middle lines provide a better choice of combination than the middle and background lines and thus the switch 310 will select the upper line signal Tb for the application to the substracyor 312.

A third condition occurs when neither XL nor XU is zero. For this case, the switch control unit 308 causes the switch 310 to mix the top and bottom lines according to equation (3) above and by doing so favors the signal that has the last line towards the line difference and thus providing minimal visual artifacts. The recovery of the luminance component in the example of Figure 3 is achieved by the subtractor 316, which subtracts the separately-combinely filtered chrominance component, separated from the full broadband reduced noise signal of the midline S8 provided by the medium cycle delay unit 42. Since the chrominance signal S21 is of band limited to the region of 1.0 to 4.2 MHz by filters 302 and 306, the vertical detail component of the luminance signal is not degraded, since signal S14 is not filtered by comb below 1.0 MHz. Therefore, no vertical detail restoration is needed. And because the bandwidth of signal S21 is a little wider than normal, an additional bandpass filter 318 is provided to limit the bandwidth of the chrominance signal to about 3.58 MHz plus or minus about half MHZ. It will be appreciated that the development of the control signal "K" in the example of Figure 3 necessarily introduces a processing delay. In the application of the principles of the invention, it is desirable that this processing delay be compensated by introducing similar delays in the signal paths leading to the "soft switch" 310 and the subtractor 312. Since the correction for the processing delay in the delays of the chrominance signal path S13, which is used, in this example, to develop the luminance signal S14, one can add an additional compensation delay in the luminance signal path leading to the subtractor 316 to thereby ensure the luma / chroma record. As previously noted, Figure 12 provides a currently preferred implementation of the switch control unit 308, which implements equations 1, 2 and 3. The pixels c, f, i, are formed by delaying the signals Tb, Mb, and Bb in the two pixel delay units 1202, 1204 and 1206, respectively. The diagonal differences c-d, a-f, f-g and d-i are provided by subtractors 1208, 1210, 1212 and 1214, respectively, which subtract d from c, f from a, g from f, and i from g, respectively. The absolute values of the outputs of the subtractor are formed by the absolute value circuits 1216, 1220, 1222 and 1226, respectively. The signal XU is provided by the maximum value detector 1218, which passes the maximum of the inputs of the absolute value circuits 1216 and 1220. The signal XL is provided by the maximum value detector 1224, which passes the maximum of the signals of the absolute value circuits 1222 and 1226. Equation (3) is implemented by the adder 1228, which adds the maximum values and the divisor 1232, which divides the output of the maximum value circuit 1224 between the output of the adder 1228 to form the control signal "K". The low pass filters 1230 and 1234 eliminate high frequency variations. The operation of the transverse gradient processor 308B in the system of Figure 3 is as previously described in detail in the discussion of the switch control unit 308 and expressions 1-3. Figure 7 provides an implementation of the soft switch 310, which includes a subtractor, which subtracts the bottom line signal Bb from the upper line signal Tb. A multiplier 704 multiplies the output of the subtractor 702 by the control signal K and the result is added to the signal Bb to form the mixed output signal S20. During operation, when K is zero, the bottom line signal Bb is passed to the subtractor 312 for combination to form the output signal of the chrominance component. When K is a unit, the signal Bb is inverted by the subtractor 702 and thus canceled in the adder 706, leaving the upper line signal Tb to combine the subtractor with the midline signal to form the chrominance component output signal S21 For the values of K between zero and unity, the upper and background line signals are mixed in proportion to the value of "K", so that S20 is equal to K times b plus the quantity (1-K) of times of the signal Bb. Figure 13 illustrates a modification of the separator of figure 1, which provides an alternative to obtain the difference signal S11 of the composite video input signal S4 and the average feedback signal S10. Note that in the example of Figure 1 the signal S11 was obtained by the subtractor 30, which subtracts the composite video input signal S4 from the averaged signal S10. In the modified circuit of Figure 13, the input terminals of the subtractor 30 are inverted. Accordingly, in this example, the difference signal S11 is obtained by subtracting the averaged signal S10 from the composite video signal S4. Since the above change will effectively invert the signal S11 and the limited signal S5, a second version is provided to ensure that the phase of the noise reduction signal relative to the composite video input signal is of appropriate polarity to provide the noise reduction when the noise reduction signal is combined with the composite video input signal S4 to produce the reduced noise video signal S6. This function is provided in the example of Figure 13, replacing adder 34 of Figure 1 with subtractor 1300. The subtractor is connected in order to subtract the limited difference signal S5 from the composite video input signal, advantageously , these modifications provide useful topological alternatives to the design engineer to form the noise reduction signal without altering the total operation of the modified circuit. As used throughout the above description, the term "color cycle" has been used to represent a period equal to a complete cycle of the chrominance signal sub-carrier. This period corresponds to two radians Pi of angle or t360 electrical degree of rotation. Therefore, the term "half color cycle" corresponds to the length of time for the color subcarrier to pass through radians Pi of revolution or 180 electrical degrees, for sampled digital systems a velocity of four times the sub- color carrier, a color cycle corresponds to a period of four image elements ("pixeis") and half color cycle corresponds to a range of two pixels. The realization of the sample at the speed of four times that of the color subcarrier in digital systems is preferred, due to the relative ease at which delays of a half color cycle can be achieved (2 pixels = 0.5 color cycles). ). However, digital implementations of the invention can be achieved with other sample rates by using interpolation to obtain medium cycle delays.

Claims (5)

1. - A method for reducing noise and separating components of a composite video input signal (S4), comprising: combining (34) said composite video input signal (S4) with a noise reduction signal (S5) to form a first reduced noise video signal (S6); further comprising: delaying (38,42,40) said first reduced noise video signal (S6) to form a first delayed video signal (S7) having a one-line delay minus half a color cycle, one second delayed video signal (S8) having a one-line delay, and a third delayed video signal (S9) having a one-line delay plus half a color cycle; forming said noise reduction signal (S5) by combining (30) said composite video input signal (S4) with the first (S7) and the third (S9) delayed video signals; and combining (50, 52, 56) said first reduced noise video signal (S6) with the second delayed video signal (S8) to form chrominance component output signals (S13) and reduced noise luminance (S14). . 2 - A method according to claim 1, wherein the step of forming said noise reduction signal comprises: averaging (36) said first and third delayed video signals; and forming (30) a difference signal (S11) between the resulting averaged signal (S10) and the composite video input signal (S4). 3 - A method according to claim 2, wherein the first combining step (34) for forming said first reduced noise video signal (S6) comprises one selected from (i) adding (34) said input signal of composite video (S4) to said noise reduction signal (S5); u (ii) subtracting (1300) said noise reduction signal (S5) from the composite video input signal (S4) 4 - A method according to claim 1, wherein the step of forming a signal reduction of noise (S5) comprises averaging (36) said first (S7) and third (S9) delayed signals to provide an averaged signal (S10), forming (3) a difference signal (S11) between the averaged signal (S10) and the composite video input signal (S4); and limiting (32) said reference signal (S11) 5 - A method according to claim 2, further comprising applying a core (1000) to said noise reduction signal (S5) before combining it with the composite video input signal (S4) 6 - A method according to claim 2, wherein the last combination step mentioned (50, 52, 56) comprises combining (50) said first video signal (S6) with the second delayed video signal (S8) to produce said chrominance component video output signal (S13); and subtracting (56) said first video signal chrominance component signal (S13) from the first video signal (S6) to produce said separate luminance component signal and reduced noise (S14). 7. A method according to claim 2, wherein said second combining step comprises: delaying (304) said second delayed video signal (S8) by a line to form a fourth video signal; combining (310, 312) said first video signal (S6), said second delayed video signal (S8) and said fourth delayed video signal (Tb) as an image detail function to form said video output signal of reduced noise chrominance components (S13); and subtractively combining said reduced noise chrominance component video output signal (S13) with said second delayed signal (S8) to form said video output signal of reduced noise luminance components (S14). 8. A method according to claim 2, wherein the second combining step comprises: combining (202,204,208) said first video signal (S6) with said second video signal delayed to produce the component video output signal Reduced noise chrominance (S13) and a vertical detail video signal (S16); and combining (200) said first video signal (S6) with the second delayed video signal (S8) to produce a luminance component signal having decreased vertical detail (S15); and combining (206) said luminance component signal having decreased vertical detail (S15) with the vertical detail video signal (S16) to form said reduced noise luminance component video output signal. 9. A method according to claim 2, further comprising: forming the third delayed video signal (S9) by delaying (40) said second delayed video signal (S8) by means of a color cycle; forming the second delayed video signal (S8) by delaying (42) said first video signal (S7) by means of a color cycle; and forming said first delayed video signal (S7) delaying (38) the first reduced noise video signal (S6) by one line minus half color cycle. 10 - An apparatus for reducing the noise and separating the luminance and chrominance components of a composite video input signal (S4), comprising: a source (12, 14, 23, 25) for providing said video input signal compound (S4); and a first combination circuit (34), coupled to the source (12, 14, 23, 25), to combine the composite video input signal (S4) with a noise reduction signal (S5) supplied thereto to form a first reduced noise video signal (S6), wherein: a delay circuit (38, 42, 40), coupled to the first combination circuit (34), for delaying said first reduced noise video signal (S6) to form (i) a delayed first video signal ( S7) having a delay of one line minus half color cycle, (ii) a second delayed video signal (S8) having a one-line delay, and (iii) a third delayed video signal (S9) having a delay of one line plus half of the color cycle; a second combination circuit (36, 30, 32), coupled to said source (12, 14, 23, 25) and to the delay circuit (38, 42, 40), to combine the composite video input signal (S4) ) with the first and third delayed video signals (S7, S9) to form said noise reduction signal (S5); and a third combination circuit (50, 52, 56), coupled to the first and second combination circuits (34; 36, 30, 32) for combining the first reduced noise video signal (S6) with the second signal of delayed video (S8) to form chrominance and luminance output signals of separated and separated network noise (S 13, S14). 1 - An apparatus according to claim 10, wherein said second combination circuit (36, 30, 32) comprises: an averager (36), coupled to the delay field (38, 42, 40) for averaging said first and third delayed video signals (S7, S9), and a subtractor (30), coupled to the source (12, 14, 23, 25) and the averager (36) to form a difference signal (S11) between the resulting averaged signal (S10) and the composite video input signal (S4). 1

2. An apparatus according to claim 10, wherein said first combination circuit (34) comprises one selected from: (i) an adder (34) to sum said composite video input signal (S4) to the signal Noise reduction (S5); (ii) a subtractor (1300) for subtracting the noise reduction signal (S5) from the composite video input signal (S4). 1

3. An apparatus according to claim 10, wherein the second combination circuit (36, 30, 32) for combining the noise reduction signal (S5) comprises: an averager (36) for averaging the first and third delayed video signals (S7, S9) to provide an averaged signal (S10); a subtractor (30) for forming a difference signal (S11) between the averaged signal (S10) and the composite video input signal (S4); and a limiter (32) for limiting said difference signal (S11). 1

4. An apparatus according to claim 10, which further comprises: a core circuit (1000) for applying a core to the noise reduction signal (S5) before combining it with the composite video input signal (S4). 1

5. An apparatus according to claim 10, wherein the third combination circuit (Figure 1:50, 52,56) comprises: a first subtractor (50), coupled to the first combination circuit (34) and the circuit delay (38, 42, 40), for combining the first reduced noise video signal (S6) with the second delayed video signal (S8) to produce said reduced noise chrominance component video output signal (S13) ); and a second subtractor (56), coupled to the first subtractor (50) and to the first combination circuit (34), for subtracting the chroma component video signal output separated from reduced noise (S13) provided by the first sub-processor (50) of the first reduced noise video signal ( S6) provided by the first combination circuit (34) to produce said separate luminance component signal and reduced noise (S14). 16 - An apparatus according to claim 10, wherein said third combination circuit (Figure 3: 302, 304, 306, 308, 310, 312, 314, 316, 318) comprises: a delay circuit (304) to delay the second delayed video signal (S8) on one line to form a fourth delayed video signal (Tb); a fourth combination circuit (308, 310, 312) combining the first reduced noise video signal (S6), the second delayed video signal (S8) and the fourth delayed video signal (Tb) as a detail function of image to form said reduced noise chrominance component video output signal (S13); and a subtractor (316) for subtractively combining said reduced noise chrominance component video output signal (S13) with the second delayed signal (S8) to form said reduced noise luminance component video output signal (S14). ). 17. An apparatus according to claim 10, wherein said third combination circuit (Figures 2: 200, 202, 204, 206, 208) comprises: a subtractor (202) for combining said first video signal of reduced noise (S6) with the second delayed video signal (S8) to produce the reduced noise chrominance component video output signal (S13) and a vertical detail video signal (S16); a first adder (202) for combining the first reduced noise video signal (S6) with the second delayed video signal (S8) to produce a luminance component signal having reduced vertical detail (S15); and a second adder (206) for combining the luminance component signal having the reduced vertical detail (S15) provided by the first adder (202) with the vertical detail video signal (S16) to form the output signal of Reduced noise luminance component video (S14). 18. An apparatus according to claim 10, wherein the delay circuit (38,42,40) comprises: a first delay unit (40) coupled to receive said second delayed video signal (S8) to form said delay third delayed video signal (S9) delaying the second delayed video signal (S8) by half color cycle; a second delay unit (42) coupled to receive said first delayed video signal (S7) to form the second delayed video signal (S8) by delaying the first delayed video signal by half color cycle; and a third delay unit (38) coupled to receive the first reduced noise video signal (S6) to form the first delayed video signal (S7) by delaying the first reduced noise video signal (S6) by a line less than half color cycle.