FI75964C - APPARAT FOER KAMFILTRERING AV EN SAMMANSATT TV-SIGNAL FRAON BILDAREA TILL BILDAREA. - Google Patents

Description

75964

An apparatus for comb filtering the combined TV signal from frame to frame

The present invention relates to combi-to-picture comb filters for separating the luminance (Y) and chrominance (C) components of television signals, and in particular to an apparatus for correcting signal distortion due to motion between successive video areas 10 in said comb filter system.

Electronic signals of a periodic nature can advantageously be processed by storing versions of the signal that differ in time by a repetition period, and then combining the stored versions to add information to the signal. For example, in conventional television transmission systems (and most video recording / playback systems), a large portion of the brightness (luminance) information contained in an image is represented by sig-20 signal frequencies concentrated on whole multiples of the horizontal line scan frequency. The color (chrominance) information is encoded or placed in a portion of the luminance signal spectrum that is close to frequencies midway between multiples of 25 line scan frequencies (i.e., odd multiples of half the line scan frequency halves).

The chrominance and luminance information can be separated and the detail information can be added by suitably combing the combined signal spectrum. Known comb filtering arrangements take advantage of the fact that the odd multiple ratio between the chrominance signal components and the new half of the line sweep frequency causes the chrominance signal components to be in the corresponding image areas in successive streaks at 2 75964 consecutive phases at 180 ° C. The luminance components in the respective image areas on the successive lines are substantially in phase with each other (and are 5 so-called non-interlaced components).

The comb filter system produces one or more variations of the combined image representative signal that are delayed with respect to time by at least one line sweep period (so-called -10 tu 1-H delay). The signals from one line are summed with the signals from the previous line, which results in the cancellation of interlaced frequency components (e.g., chromaticity) while amplifying non-interlaced frequency components (e.g., luminance).

15 Subtracting the signals by two consecutive lines (for example, inverting the signals of one line and then adding them to those two) cancels the non-interlaced signals while amplifying the interlaced signals. Thus, the luminance and chrominance signals 20 can be comb-combined together and at the same time advantageously separated.

The comb filtering procedure described above is possible due to the large inherent level of signal surplus from line to line in a typical video scene. Examining the stationary image, and a particular pixel of this particular scene, it can be seen that the level of the excess signal representing that pixel from the image area to the image area is substantially one hundred percent. Thus, when performing comb filtering from the image area 30 to the image area, the proportion of the residual luminance signal in the comb-filtered chrominance signal is significantly reduced from the line-to-line comb-filtered signal. Correspondingly, the proportion of the residual chrominance signal in the comb-filtered luminance signal decreases significantly from line 35 to the comb-filtered signal.

3,75964

More specifically, the comb-filtered chrominance signal, for example, is not accompanied by any vertical detail. The chrominance signal of the combined signal of the NTSC system is synchronized to be 180 ° in a phase-to-frame phase ratio of 5, allowing comb filtering based on the image area. The light path component can therefore be separated by linearly increasing the combined signal from the following successive image areas, and the chrominance component can be separated by subtracting the combined signal linearly from the following image areas.

The response characteristic of a line-to-line comb filter has zeros or toothing located in line frequency or 15,734 kHz periods, while the image-to-image filter response has teeth in 30 Hz periods. The smaller distance between the teeth or zeros in the comb area filtering of the image area results in more complete comb filtering processing in all spatial directions than in the internal comb filtering of the line.

For example, GB-1 402 609 and DE-30 07 520 disclose comb filtering over the entire image area, in which image objects which are not stationary cause distortions in the reproduced images. These distortions are due to the fact that in the image areas where motion occurs, the blanking of the brightness components in the comb-filtered color signal and the blanking of the color components in the comb-filtered brightness signal are not complete, and in addition there is an effective reduction of the brightness signal bandwidth. Visible movement of an image is when differences in the contents of Figure 30 occur in 1/30 second or less as a result of movement or oscillations of the object.

In comb filtering covering the entire image area, the distortions caused by the movement are two-dimensional and are entirely due to the movements of the image that take place from one image area to another. Distortions appear both horizontally and vertically in the image plane and appear as duplicate or ghost images in playback images. Duplicate images are spaced at a distance corresponding to the speed of movement, and may also be accompanied by the wrong hue for the edge areas of moving objects. In order to avoid these phenomena, DE-30 07 520 proposes banded comb filtration.

In-line comb filters, on the other hand, produce one-dimensional distortions due to the vertical structure of the scene, even if the scene 10 is stationary. Distortions following in-line comb filtering are manifested as an effective reduction in the bandwidth of the signal representing the vertical structure of the scene and a softening of the vertical edges of the scene content.

It is an object of the present invention to reduce the perceptible effects of motion in an image-to-image comb filtration system.

A preferred embodiment of the present invention includes a comb filter for processing television signals (TV), the filter including a delay device for delaying the TV signal by one frame. A first device is provided for combining a TV signal and a delayed TV signal from a delay device to form a sum corresponding to the first 25 components obtained from the TV signal. A second device is provided for combining a TV signal and a delayed TV signal from a delay device to form a difference corresponding to the second component obtained from the TV signal. A linear low pass filter is provided, the input terminal of which is connected to a second combiner device for receiving the second component of the TV signal. The low-pass filter has an output terminal and essentially blocks the frequency spectrum normally charged by the second component of the TV signal. A third device is provided which connects to the first combiner device and the output terminal of the low-pass filter to combine the first component and the TV signal in a low-pass filtered signal to correct errors of the first component of the signal due to intra-picture movement.

According to a further aspect of the invention, the motion detection device is connected to the output terminal of the low-pass filter and is responsive to the low-pass filtered signal for detecting the occurrence of motion within the field of view. The motion detection device generates a control signal responsive to that motion. A fourth signal combining device is provided, the first input terminal of which is connected to the second combining device to receive the second component. The fourth signal combiner has a second input terminal and an output terminal. The switching device is coupled to the motion detecting device and is responsive to the control signal for selectively coupling the first component to the second input terminal of said fourth signal combiner when the intra-scene scene motion is detected.

According to a further aspect of the invention, a linear phase bandpass filter is connected in series to the output terminal of the fourth signal combiner. The bandpass filter is designed to pass only the frequency spectrum of the chrominance signal.

In the drawings: Fig. 1 is a block diagram of a frame-by-frame comb filter known in the art of TV signal processing # Figs. 2a-d show amplitude waveforms with respect to time associated with the circuit of Fig. 1, while Fig. 30e shows a waveform; obtained when the circuit of Figure 1 is converted; Fig. 3, including Figs. 3a-b, shows a block diagram of a frame-by-frame comb filter according to an embodiment of the present invention.

Referring to Figure 1, there is shown a field-to-field comb filter known in the art. For the purpose of printing, a device is shown in an embodiment in which the luminance and chrominance signals are separated from the combined TV signal. However, the device is not limited to this specific application. The device of Figure 1 can be used, for example, to comb filter a luminance signal to place a chrominance signal to form a video signal connected thereto.

The capacity of the memory element 10 of Fig. 1 is sufficient to store an amount of electrical data of one image area of the video signal (for example, two fields of the NTSC video format). The memory member 10 sequentially accepts the input signal at the input terminal 11 and outputs the same signal to the output terminal 14 with a delay of one frame area period. The member 10 may comprise a digital memory, for example direct memory (RAM), including relevant support electronics, or an analog memory, for example a serial charge transfer device (CTD) or charge coupled device (CCD), including the necessary support circuitry. As such, it does not matter whether the body 10 20 is a digital or analog circuit, however, it is assumed that it is digital in nature. Other relevant circuit elements are also assumed to be digital in nature in order to avoid digital-to-analog conversion of the signal between circuit elements.

The input signal from terminal 11 and the delayed signal from terminal 14 are summed together in adder circuit 12. The luminance components of the signal are summed in phase, while the chrominance components are canceled 180 ° relative to each other. subtraction circuit 13, in which the chrominance components of the two image areas are amplified exponentially, while the luminance components are canceled from the luminance corap-35 to produce a substantially free chrominance signal to terminal 16 (at least when successive image area signals are at the same stationary point).

Referring to Figure 2, the effects of movement 5 on the image-to-image comb filtering operation described above are shown graphically. Figure 2a represents a portion of the combined signal applied to terminal 11 over a random period of time. Figure 2b represents the combined video signal from the same scene exactly one period of 10 frames later and in which some motion occurs. In waveform 2a, it is assumed that the signal component between time points T1 and T3 represents an object whose luminance and chrominance parameters are L2 and C2, respectively. The subject is assumed to be located within the content of the uniform scene content with the luminance and chrominance parameters being Li and Cl, respectively.

Between image areas, the object moves spatially, causing the object's video signal representation to change with time, i.e., to fall between time points T2 and T4 in Figure 2b. The relative signal parameters L and C of the luminance and chrominance components of the signal remain the same between the waveforms of Figures 2a and 2b, although there is a time shift with respect to the target represented by L2, C2. However, the chrominance component of the signal is 180 ° in 25 phase relationship between waveforms 2a and 2b (i.e., between image areas).

The sum and difference of the waveforms in Figures 2a and 2b are represented by waveforms 2c and 2d, respectively. It is observed that for those time blocks in which the absolute absolute value of the chrominance vector component 30 between waveforms 2a and 2b is the same, the sum of these two waveforms (Figure 2c) results in a substantially complete reversal of the chrominance component. Correspondingly, when the amplitude of the luminance component of the signal is the same between the two waveforms (2a and 2b) 35, the reduction of these two waveforms e 75964 results (Fig. 2d) resulting in a substantially complete reversal of the luminance component. Figures 2a / 2b, 2c and 2d represent signals present at terminals 11, 14, 15 and 16 of the circuit of Figure 1, respectively.

5 During periods when motion occurs between image areas, i.e., at times T1 and T2 and at times T3 and T4, signals of different sizes are partially summed and result in incomplete cancellation of the unwanted component.

10 In addition, motion distorts the desired signal. Motion causes the signal to be averaged over the time periods in which motion occurs. These averages are shown in Figure 2c by signal blocks denoted by (LI + L2), which is equivalent to 15 (2L 1 + 2L 2) / 2. Normally, the signal from the two image areas is weighted by a factor of 1/2 before combining them. The amplitude of the comb-filtered signal is half of the displayed value and the signal occurring during the motion period is (Li + L2) / 2. The averages of the luminance signal 20 (L1 + L2) tend to reproduce the signal offsets to produce a double or ghost image surrounding the moving subject. The averages (C1 + C2) of the comb-filtered chrominance signal during periods of intra-image motion tend to distort the color tone around the moving subject.

If a low-pass filter is provided at terminal 16 of Figure 1, the comb-filtered chrominance signal of Figure 2d produces the signal shown in Figure 2e. In Fig. 2e, the spectrum of the chrominance signal is removed, leaving only the unmodified luminance component due to the intra-field motion. If the signal of Fig. 2e is summed or subtracted from the signal of Fig. 2c, the edges of the luminance signal are amplified, producing a luminance component equivalent to that which is not comb filtered.

The summation of the signal of Fig. 2e amplifies the light signal of Fig. 2c between times T1-T2 to the level 2La and reduces the luminance signal between times T3 and T4 to 2L1, as shown by the dashed lines.

The circuitry shown in Figure 3a is an image-to-image comb filter arranged in accordance with the principles of the present invention, which is capable of correcting distortions, such as ghosting or haze, caused by movement between successive image areas. In Fig. 3, the members indicated by 10 with the same numbers as the members of Fig. 1 perform the same functions as in Fig. 1. Thus, comb-filtered light-density and comb-filtered chrominance signals are produced at the circuit terminals 15 and 16 of Fig. 3a, respectively. -15 incompletely comb-filtered signal components due to internal motion.

The comb-filtered chrominance signal at terminal 16 is applied to a linear-phase low-pass filter 27 to remove the spectrum of the chrominance signal, leaving unreversible pin-frequency luminance components. The low-pass filtered signal from the filter 27 is summed back in the adder circuit 31 to the comb-filtered luminance signal from the terminal 15 to produce a corrected luminance signal Yc at the terminal 32.

The luminance signal thus corrected is generated at terminal 32 in accordance with the principles of the present invention. The remaining uncompromising chrominance component from the interface 15 in the comb-filtered luminance signal only tends to create a second-order distortion that is not detectable except for close examination of the image shown.

The low pass filter 27 must be in linear phase to correctly amplify the motion detail in the comb-filtered luminance signal. The transient response 35 is adversely affected if the motion detail 10 75964 is not properly amplified due to phase differences between the blocks subjected to the motion of the comb-filtered luminance signal and the detail of the motion separated from the comb-filtered chrominance signal.

5 Residual chromaticity (C2-C1, C1-C2) due to intra-image movement in the luminance channel does not produce significant distortions in the reproduced image. However, the averaged or shifted chrominance signal (CI + C2) in the comb-filtered chrominance signal component that occurs during motion cycles may produce noticeable color distortion. The chrominance signal is a vector quantity. Thus, the vector sum C2 + Cl during the periods of motion can produce colors that are completely different from those represented by the second signal from C1 or C2. In order to maintain color purity at the edges of moving objects, it is often desirable for the chrominance component to be corrected against motion errors that develop in comb filtration processing.

The information required to correct the comb-filtered chrominance signal is derived from the comb-filtered luminance signal. This information can be repositioned in a manner similar to a method for restoring motion detail to a comb-filtered luminance signal. However, color restoration should not be a continuous procedure, but should be performed during movement cycles. This is due to the relatively high energy density of the luminance signal compared to the cross-components of the chrominance signal. Continuous summation of the comb-filtered luminance signal to the comb-filtered brightness signal would significantly impair processing. However, summing the comb-filtered luminance signal to the comb-filtered chrominance signal during periods of motion produces a preferred signal without resetting the chromaticity.

It is usually necessary to bandpass filter the comb-filtered chrominance signal to eliminate the residual luminance signal 75964, i.e., the motion detail of the luminance, the comb-filtered luminance component summed into the comb-filtered chrominance signal is substantially eliminated from the corrected chrominance signal. Only the luminance components from the frequency band of the chrominance-5 signal remain in the corrected chrominance signal, and the energy of the luminance signal in this band is relatively small.

The luminance signal consists of the basic parts of the chrominance signal 1Q, i.e. the red, blue and green signals, and thus the amplitude of the luminance signal also gives an indication of the amplitude of the chrominance signal. When the signal from the low-pass filter 27 exceeds a predetermined value, it indicates the presence of motion, and this amplitude of the signal indicates the amplitude of the errors in chrominance. Errors in the chrominance signal large enough to cause detectable errors in the reproduced image can thus be detected by measuring the amplitude of the luminance signal 20 emitted by the low-pass filter 27.

In Fig. 3a, a motion detector 28 connected to the output terminal of the low-pass filter 27 compares the low-pass filtered chrominance signal with a threshold value. The threshold is set to a level corresponding to the signal level required to generate detectable chromaticity distortions. When the low-pass filtered signal from the filter 27 exceeds a threshold value, the motion detector 28 generates a control signal that closes the switch 29.

Figures 3b and 3c are exemplary embodiments of an analog and digital motion detector, respectively. The analog version 28 'comprises a high gain differential amplifier 40 and a reference voltage source 41. The signal from the filter 27 is applied to the non-inverting input terminal 42 of the amplifier 40 and the signal from the reference source 41 35 is applied to the inverting terminal of the amplifier 40. When the voltage at terminal 42 exceeds the reference value, the output signal from amplifier 40 becomes large, in which case it remains small. Connected to the output terminal of the amplifier 40 is an N-type channel transistor 5 29 ', which can be used as the switch 29 in Fig. 3a.

In Figure 3c, the digital version 28 "of the motion detector comprises a binary reference source 45, a subtraction-10 circuit 43 (comparator), a character detector 44, and an AND-circuit switch 29". (In the drawing, double-line connections indicate parallel bit lines). The subtractor 43 subtracts the reference number (45) from the binary signal at the output 42 of the filter 27. The difference value tut-15 is touched by the character circuit 44 to determine whether it is positive or negative. If the difference is positive, the SGN circuit produces a positive output pulse that allows the AND circuit 29 "to pass a signal to its data lines. Note that if the subtraction circuit performs two complementing arithmetics, the signal circuit can be eliminated and replaced with the inverter circuit directly connected to the most signal output terminals. between a bit, i.e. a character bit, and an AND circuit.

The circuits of Figures 3b and 3c are, as shown, responsive to input signals of only one polarity, however, it is known in circuit art to place two circuits of that type in parallel to be responsive to both polarities. Alternatively, compared and can be of the "window comparator" type.

The switch 29 selectively switches the comb-filtered luminance signal from the interface 15 to one input of the adder circuit 30. The second input of the adder circuit 30 is directly connected to receive the comb-filtered chrominance signal from terminal 16. When detecting a motion detail exceeding the threshold, the switch 29 is closed and the comb-filtered 13 75 9 64 chrominance and luminance components are summed in the suramain circuit 30. When no motion detail is detected, the switch 29 is open and the adder circuit outputs the comb-filtered chrominance signal from the terminal 16.

The output signal from the adder circuit 30 is coupled to a linear phase bandpass filter 34 which passes only the frequency spectrum of the chrominance signal to its output terminal 35.

The circuit members 31 and 30 may be either adder or subtractor 10, as indicated by (+) signs at their input terminals in Figure 3a. However, they must both be circuits of the same type, i.e., both are either adder circuits or subtraction circuits. Whether they are adder circuits or subtraction circuits determines whether the signal blocks distorted by forward or reverse motion are amplified or canceled, respectively.

Claims (10)

A cam flutter for processing television signals, which filter includes a delay means (10) for delaying said television signal an image period, first means (12) * for combining said television signal and the delayed television signal from said delay means to form their sum corresponding to a delay signal. first component (Y) in said television signal and second means (13) for combining said television signal and the delayed television signal from said delay means to form their difference corresponding to a second component (C) in said television signal, said cam filter acknowledging characterized in that a low pass filter (27) is linear phase and with an input terminal coupled to said second compression means (13) for receiving said second component in the television signal and with an output terminal, said low pass filter substantially discarding the frequency spectrum which normally is taken up by said second component, and third means 20 (31) coupled to said first combining means (12) and to said output terminal of said low-pass filter to combine the first signal component with the low-pass filtered signal to correct the first signal component for errors in the same depending on motion between images. A comb filter according to claim 1, characterized in that motion detection means (28) are coupled to said output socket of said low-pass filter (27) and actuable in response to said low-pass filtering signal to detect the occurrence of movement between images and to detect due to such movement, a control signal generates a fourth signal combiner (30) with an input terminal coupled to said second combiner (13) to receive said second component, said fourth signal combiner having a second input terminal and an output terminal, and switching means (29) coupled to said motion detection means and operable in dependence on said control signal to selectively copy said first component to said second input terminal of said fourth signal combiner when motion between images is detected. A comb filter according to claim 2, characterized in that said television signal Mr a composite signal including chrominance and luminance resistors, said first component (Y) corresponds to the luminance signal, said second component (C) corresponds to said second component (C). against said chrominance signal and said delay means comprises a memory means (10) for storing at least one image of the video signal. A comb filter according to claim 3, characterized in that a linear phase bandpass filter is connected in series with said output terminal of said fourth signal combiner, said bandpass filter being so designed that it only transmits chrominance signal frequency spectrum. A comb filter according to claim 2, characterized in that said motion detection means includes said low pass filter (27), a reference (41: 45), and comparison means (28 ': 28 ") for comparing one of said positions. -pass filter transmitted signal with said reference and to generate said control signal whenever the low-pass filtered signal differs from the reference by a predetermined amount. Comb cam according to claim 2, characterized in that said third combining means and / or said fourth combining means is a signal adder circuit. Comb filter according to claim 2, characterized in that said third combining means and / or said fourth combining means is a signal subtractor circuit. 8. A method of comb filtering television signals using the apparatus of claim 1 to separate chrominance and luminance components, characterized in that the television signal is delayed with the duration of an information image, that delayed and non-delayed television signals are combined. subtractive to allow