WO2004102975A1 - Composite video signal y/c separation circuit and y/c separation method - Google Patents

Abstract

A FIR comb filter (2) is configured by use of a unit filter having, as a filter coefficient, a basic numeral sequence ({3, -8, -12, 72, -110, 72, -12, -8, 3}/256). The delay amount of a plurality of delay elements (12-1 to 12-8) constituting the comb filter (2) is set to one horizontal scanning interval of an image. In this way, a comb filter characteristic can be realized which has a larger attenuation amount in the frequency band of a brightness signal frequency-interleaved with a color signal, so that the color and brightness signals can be substantially completely separated from the composite video signal without being influenced by the status or motion of the input image signal, thereby significantly reducing image interferences such as cross color and dot crawl.

Description

 Description YC separation circuit and YC separation method for composite video signal

 The present invention relates to a YC separation circuit and a YC separation method for a composite video signal, and more particularly to separating a color video signal and a luminance signal from a composite video signal formed by frequency-multiplexing a color signal and a luminance signal. It is about the method of doing. Background art

 Generally, a color video signal is composed of a color signal (C signal) and a luminance signal (Y signal).

) Are frequency-multiplexed and are called composite video signals. As shown in Fig. 1, the chrominance signal is orthogonally modulated from a 3.58 MHz (in the case of NTSC signal) reference signal called a color subcarrier (color subcarrier) and multiplexed with the luminance signal. . When a color signal is multiplexed with a luminance signal, a burst signal that is a phase reference during color demodulation is also multiplexed at the same time.

In the case of an NTSC signal, the frequency f _{S (;} of the color subcarrier is determined to be 255 × 2 times the horizontal scanning frequency f _H , and the frequency spectrum is, as shown in FIG. The chrominance signal is in the form of a frequency interleave inserted between the luminance signals, ie, the luminance signal has an energy peak at every integral multiple of the horizontal scanning frequency f _H , and 0.5 from that peak. The energy is minimized at the point where the frequency is shifted by f _{H. The} interference of the spectrum of the color signal and the spectrum of the luminance signal is suppressed by fitting the color signal at the minimum energy. As shown in Fig. 3, the phase of the color subcarrier is The phases are opposite both between the inside and between the frames.

 The video signal processing on the receiver side is performed separately for the chrominance signal and the luminance signal, converted to RGB signals and output to the display device. That is, on the receiver side, the color signal band is extracted from the frequency band characteristics of the composite video signal, and color demodulation is performed on the axes of two color difference signals (R-Y signal and B-Y signal). Then, a G_Y signal is generated by matrix synthesis from the demodulated RY signal and BY signal. In addition, a luminance signal is added to these three color difference signals: a signal, a G signal, and a B signal are generated.

 In order to perform the above-described video signal processing on the receiver side, it is necessary to separate the chrominance signal component and the luminance signal component from the composite video signal. This separation is generally called YC separation, and has a significant effect on video signal performance after color signal processing and luminance signal processing. If the YC component is not completely performed, image interference such as cross-color and dot crawl will occur due to interference between the color signal and the luminance signal.

 Cross color refers to crosstalk in which a luminance signal leaks into a chrominance signal. When this occurs, moire occurs in the vertical stripes with fine luminance. On the other hand, dot crawl is caused by crosstalk (dot interference) in which a chrominance signal is leaked into a luminance signal. They say they look moving. In order to avoid such image disturbance, it is required to perform YC separation precisely.

The methods of YC separation are roughly classified into three categories: one-dimensional YC separation, two-dimensional YC separation, and three-dimensional YC separation. FIG. 4 is a diagram showing the configuration of these three types of YC separation circuits. The one-dimensional YC separation focuses on the fact that the color signal is near 3.58 MHz, as shown in Fig. 4 (a), and extracts the color signal with a band-pass filter that passes the band of the color signal. Other signals are referred to as luminance signals. It is a way to This method was adopted in the early days of the appearance of color video signals, and cross-color appears in the striped part of the image, and dot crawls appear in areas where the color is discontinuous.

 The two-dimensional YC separation utilizes the fact that the phase of the color subcarrier is inverted every line due to frequency interleaving. In other words, as shown in Fig. 4 (b), the signal of one line is subtracted from the signal of a certain line to extract the color signal, and the color signal is further extracted by a band pass filter of 3.58 MHz. Take out. In this method, the frequency characteristic of the filter becomes a comb shape, and vertical stripes having a luminance of 3.58 MHz do not appear in the color signal output. Therefore, the cross-color phenomenon is reduced compared to the one-dimensional YC separation, and the horizontal resolution is increased. However, dot crawls remain in areas where colors are discontinuous vertically.

 The three-dimensional YC separation further utilizes the fact that the phase of the color subcarrier is inverted every frame. That is, as shown in Fig. 4 (c), the three-dimensional YC separation circuit subtracts the signal between two frames to extract a color signal and the above-mentioned one-dimensional and two-dimensional filter. Are configured in combination. According to this method, in the case of a still image, not only the cross color but also the dot clock can be reduced. However, YC separation cannot be completely performed for videos. For this reason, motion adaptive processing is often applied, such as using three-dimensional YC separation for stationary parts of the input video and using two-dimensional YC separation for moving parts.

As the current YC separation method, the three-dimensional YC separation method, which has superior separation characteristics compared to one-dimensional and two-dimensional, is predominant. However, even in this three-dimensional YC separation method, the attenuation characteristic of the portion from which the luminance signal is extracted by the comb filter has not been able to realize a sufficient reduction amount, and the YC separation is not completely performed. There was a problem that it was not possible. On the other hand, there is also a problem that motion-adaptive YC separation must be performed for moving images because they cause interference in the images. Disclosure of the invention

 The present invention has been made to solve such a problem, and is capable of almost completely separating a color signal and a luminance signal without being affected by the state or movement of an input video signal. It is intended to provide a separation method. The YC separation circuit for a composite video signal according to the present invention performs digital filter processing on the digital composite video signal, and converts one of a color signal and a luminance signal having a frequency interleave relationship. A comb filter to be taken out; and a subtractor to subtract one of the signals extracted by the comb filter from the digital composite video signal to take out the other signal. A digital filter that multiplies a signal of each tap in a tapped delay line composed of a plurality of delay devices by a filter coefficient composed of a predetermined basic numerical sequence, and then adds and outputs the multiplied signal. The delay amount of the delay unit is set to one horizontal scanning period or one vertical scanning period.

Another embodiment of the present invention is characterized by further comprising a band pass filter for extracting the one signal. This band-pass filter uses, for example, a filter that uses the above-described predetermined basic numerical sequence as a filter coefficient as a unit filter, and inserts delays for n clocks between taps corresponding to the filter coefficient. Thus, the filter is designed by cascading at least one of the filter whose pass frequency band is adjusted and at least one of the unit filters. Also, this The band pass filter may be configured to take out one of the signals by performing a predetermined operation using signals of adjacent pixels located before and after n clocks away from the pixel of interest.

 In addition, the YC separation method of a composite video signal according to the present invention is characterized in that a signal of each tap in a tapped delay line composed of a plurality of delay units is calculated by a filter coefficient composed of a predetermined basic numerical sequence. A digital filter that multiplies, adds, and outputs a color signal using a comb filter in which the delay amounts of the plurality of delay units are set to one horizontal scanning period or one vertical scanning period, respectively. One signal is extracted from a digital composite video signal in which the luminance signal and the luminance signal are in a frequency interleaved relationship, and the other signal is extracted by subtracting the one signal from the digital composite video signal It is characterized by doing so.

 As described above, according to the present invention, a comb filter is configured using a unit filter that uses a predetermined basic numerical sequence as a filter coefficient, and the delay amounts of a plurality of delay units configuring the comb filter are imaged. 1 horizontal scanning period or 1 vertical scanning period, so that a comb filter characteristic with extremely large attenuation in the frequency portion of the luminance signal that has a relationship between the color signal and the frequency interleave is obtained. be able to. By extracting color signals using this comb filter, the color and luminance signals can be almost completely separated from the composite video signal without being affected by the state or movement of the input video signal. This makes it possible to significantly reduce image disturbance such as cross-color / dot crawl. BRIEF DESCRIPTION OF THE FIGURES

 FIG. 1 is a diagram showing a waveform of a composite video signal.

Figure 2 shows the frequency spectrum of a composite video signal. You.

 FIG. 3 is a diagram showing the phase of the color subcarrier in a composite video signal.

 FIG. 4 is a diagram showing the configuration of three types of YC separation circuits according to the conventional method. FIG. 5 is a block diagram showing the overall configuration of the YC separation circuit according to the first embodiment.

 FIG. 6 is a block diagram showing a detailed configuration example of the comb filter according to the first embodiment.

 Fig. 7 is a diagram showing the circuit configuration of the unit filters L10 and HI0 and a numerical sequence of filter coefficients.

 FIG. 8 is a diagram illustrating the frequency characteristics of the single-pass unit filters L10 and L11.

 FIG. 9 is a diagram illustrating the frequency characteristics of the high-pass unit filters H10 and H11.

 FIG. 10 is a diagram showing a partially enlarged Y / C separation filter characteristic according to the first embodiment.

 FIG. 11 is a block diagram showing the overall configuration of the three-dimensional Y C separation circuit according to the first embodiment to which the motion adaptive processing is applied.

 FIG. 12 is a block diagram showing the overall configuration of the YC separation circuit according to the second embodiment.

FIG. 13 is a diagram illustrating the frequency characteristics of the one-pass unit filter (L 10) ^m .

FIG. 14 is a diagram illustrating a frequency characteristic of the high-pass unit filter (H 10) ^π .

Fig. 15 shows a partially enlarged YZC separation filter characteristic according to the second embodiment. FIG.

 FIG. 16 is a diagram illustrating filter coefficients of the basic low-pass filter L 4 an.

 FIG. 17 is a diagram illustrating a hardware configuration example of the basic low-pass filter L4a4.

 FIG. 18 is a diagram illustrating a frequency characteristic of the basic-port one-pass filter L4a4.

 FIG. 19 is a diagram illustrating frequency-gain characteristics of the basic low-pass filter L 4 an.

 FIG. 20 is a diagram illustrating filter coefficients of the basic low-pass filter L an.

 FIG. 21 is a diagram illustrating a hardware configuration example of the basic low-pass filter La4.

 FIG. 22 is a diagram illustrating a frequency characteristic of the basic mouth one-pass filter La4. FIG. 23 is a diagram illustrating a frequency-gain characteristic of the basic low-pass filter Lan.

 Fig. 24 shows the filter coefficients of the basic high-pass filter H4sn.

'is there.

 FIG. 25 is a diagram illustrating a hardware configuration example of the basic high-pass filter H4s4.

 FIG. 26 is a diagram illustrating the frequency characteristics of the basic high-pass filter H 4 s 4.

 FIG. 27 is a diagram illustrating frequency-gain characteristics of the basic high-pass filter H 4 sn.

Figure 28 shows the filter coefficients of the basic high-pass filter H sn. You.

 FIG. 29 is a diagram illustrating a hardware configuration example of the basic high-pass filter Hs4.

 FIG. 30 is a diagram illustrating a frequency characteristic of the basic high-pass filter Hs4. FIG. 31 is a diagram illustrating a frequency-gain characteristic of the basic high-pass filter Hsn.

 FIG. 32 is a diagram illustrating a frequency-gain characteristic of the basic high-pass filter H msn where m is a parameter.

 FIG. 33 is a diagram showing an optimum value of the parameter n with respect to the parameter m.

 FIG. 34 is a diagram showing a relationship between the parameter m and the optimum value of the parameter n for the parameter m and a relationship between the parameter m and the parameter X for the parameter m.

 FIG. 35 is a diagram illustrating an impulse response of the basic high-pass filter H msn. BEST MODE FOR CARRYING OUT THE INVENTION

 (First Embodiment)

 Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 5 is a block diagram showing the overall configuration of the YC separation circuit according to the first embodiment. As shown in FIG. 5, the YC separation circuit of the present embodiment includes an AZD converter 1, a comb filter 2, a node-pass filter (BPF) 3, a subtractor 4, and a delay circuit 5. Have been.

The A / D converter 1 converts an analog composite video signal input to the YC separation circuit of the present embodiment into a digital signal. This input The composite video signal is generated, for example, by detecting the video signal by video intermediate frequency processing. The digital composite video signal output from the A / D converter 1 is input to the BPF 3 and to the subtractor 4 via the delay circuit 5.

 The BPF 3 performs a predetermined band limitation on the composite video signal output from the A / D converter 1 so that the frequency band component of the color signal is converted from the composite video signal. Extract. The comb filter 2 is configured based on a theory unique to the present embodiment, and extracts a color signal having a relationship between a luminance signal and a frequency interleave using a pass band of the comb filter characteristic. . In other words, the luminance signal is attenuated using the cutoff region of the comb filter characteristics. Details of the comb filter 2 will be described later.

 The color signal output from the comb filter 2 is the desired color signal (C signal). The color signal output from the comb filter 2 is also input to the subtractor 4. The subtractor 4 subtracts a chrominance signal from the composite video signal output from the A / D converter 1 and delayed by a predetermined amount by the delay circuit 5 to obtain a luminance signal (Y signal). Take out. The delay circuit 5 has a delay amount equivalent to the total delay amount of the comb filter 2 and the BPF 3 in order to match the phases of the composite video signal and the color signal in the subtractor 4.

The details of the comb filter 2 will be described below. FIG. 6 is a block diagram showing a detailed configuration example of the comb filter 2. Fig. 7 is a diagram showing unit filters L1n and HIn, which are the basis of the comb filter 2 shown in Fig. 6, where (a) shows the circuit configuration and (b) shows the filter configuration. The figure shows a numerical sequence of coefficients. The letter “n” after the unit filter sign indicates the number of delay clocks inserted between each tap. However, FIG. 7 shows the case where n = 0.

First, the unit filters L 10 and HI 0 in FIG. 7 will be described. As shown in FIG. 7 (a), unit filter L 1 0 of the present embodiment, H 1 0 At, cascaded eight D-type flip-flop 1 0 2 to 1 0 2 - ₈ to the input by signal Are sequentially delayed by one clock CK. Then, the signal extracted from the input and output taps of the D-type flip-flop 1 0 2 to 1 0 2 _8, the filter coefficient hi ~ shown in FIG. 7 (b)! 9 is multiplied by _nine coefficient units 13_, to 13-9, respectively, and the multiplied results are all added by _eight adders 14-, to 14-8 and output.

The two types of circuit configuration of a unit Tofiru evening L 1 0, HI 0, both being adapted in FIG. 7 (a), the filter coefficient (the coefficient unit 1 3 _, ~ 1 3 - ₉ of the multiplier values hi H9) are different as shown in Fig. 7 (b).

 As can be seen from Fig. 7 (b), the filter coefficients of the low-pass unit filter L10 are extremely simple numerical sequences {3, 8, -12, -72, -1110,-7 2,-12, 8, 3} Z 2 56. Such filter coefficients have the property that the numeric sequence is symmetric, the sum of the numeric sequence is non-zero, and the jumps of the numeric sequence have the same sign and are equal to each other ( 3 — 1 2 — 1 1 0-1 2 + 3 = _ 1 2 8, 8-7 2-7 2 + 8 =-1 2 8).

 Also, the filter coefficient of the high-pass unit filter H10 is an extremely simple numerical sequence {3, —8, -12, 72, -110, 72, —12, -18, 3}. / Consists of 56. Such filter coefficients have the property that the numerical sequence is symmetrical, the total value of the numerical sequence is zero, and the one-to-one total value of the numerical sequence is equal to each other with the opposite sign (3 — 1 2-1 1 0-1 2 + 3 = _ 1 2 8, 1 8 + 7 2 + 7 2 — 8 = 1 2 8).

A unit filter that uses a numerical sequence with such properties as the filter coefficient Taking the impulse response of L 10 and H 10, the sample has a finite value other than “0” only when the sample position along the time axis is within a certain section, and the value is It is a function that is all "0", that is, a function whose value converges to "0" at a predetermined finite sample position. The case where the value of the function has a finite value other than “0” in a local region and becomes “0” in other regions is called “finite base”.

 This finite function is a sampling function that has a maximum value only at the center sample position and a value of "0" at other sample positions. And all the sample points necessary to obtain smooth waveform data pass.

Next, a case will be described in which the number of clocks n of the delay inserted between the taps of the D-type flip-flops 10 2, to 10 2 _-8 is n ≥ l. For example, the filter coefficient of the mouth-pass unit filter L11 is generated by inserting one "0" between the filter coefficients of the low-pass unit filter L10. Similarly, the filter coefficient of the mouth-pass unit filter L in (n = 2, 3, ...) is obtained by inserting n "0" s between each filter coefficient of the mouth-pass unit filter L10. Generated by doing

 FIG. 8 is a diagram illustrating frequency-gain characteristics obtained as a result of performing FFT (Fast Fourier Trans fer) on the numerical sequence of the low-pass unit filters L10 and L11. Here, the gain frequency is normalized by "1". As can be seen from Fig. 8, in the low-pass unit filter L10, the gain is 0.5 at the center frequency, and there is no overshoot in the low-frequency region and no ringing in the high-frequency region. Filter characteristics are obtained.

Fig. 7 (b), which is the basis for realizing such mouth-to-pass filter characteristics ) Above, the numerical sequence {3, 8, -12, -72,-1110, -72, 1-12, 8, 3} is based on the basics of It is. The sampling function that has been generally used in the past converges to "0" at the sampling position of t == ± ∞, whereas the sampling function obtained from the above-mentioned numerical sequence shows that "0" at the finite sampling position. Converges to.

 Therefore, when the numerical sequence in the upper part of FIG. 7 (b) is subjected to FFT conversion, only data corresponding to a local range having a finite value other than "0" is significant. Data corresponding to values outside this range should not be neglected, although it should be considered, and there is no need to consider it theoretically, so no truncation error occurs. Therefore, if the numerical sequence shown in the upper part of Fig. 7 (b) is used as a filter coefficient, it is not necessary to cut off the coefficient using a window function, and a good low-pass filter characteristic can be obtained. .

 Also, as can be seen from FIG. 8, if the number of “0” s inserted between the filter coefficients for the basic mouth-pass unit filter L 10 is n, the frequency-gain The frequency axis (period in the frequency direction) of the characteristic is 1 / n. In other words, the frequency characteristic has a comb shape in which the number of passbands and the number of cutoffs are repeated according to the value of n. In this case, there is no overshoot or ringing in any frequency domain, and a good comb filter characteristic in which a passband and a cutoff band appear at equal intervals is obtained.

Similarly to the low-pass unit filter L11, the filter coefficient of the high-pass unit filter HI1 is generated by inserting one "0" between each filter coefficient of the high-pass unit filter HI0. You. Similarly, the filter coefficients of the high-pass unit filter H ln (n = 2, 3,...) Are n “0” s between each filter coefficient of the high-pass unit filter H 10. Generate by entering. FIG. 9 is a diagram showing the frequency-gain characteristics of the high-pass unit filter H10, Η11. Here, the gain and frequency are normalized by "1". As can be seen from Fig. 9, in the high-pass unit filter 010, the gain is 0.5 at the center frequency, and there is also an overshoot in the high-frequency region and a ringing in the low-frequency region. Good high-pass filter characteristics can be obtained.

 The numerical sequence shown in the lower part of FIG. 7 (b), which is the basis for realizing such a high-pass filter characteristic, is also the basis of a finite-unit sampling function. Therefore, by using this numerical sequence as a filter coefficient, it is not necessary to cut off the coefficient using a window function, and a good eight-pass filter characteristic can be obtained.

 Also, as can be seen from FIG. 9, if the number of “0” s inserted between the filter coefficients for the basic high-pass unit filter H 10 is n, the frequency-gain characteristic The frequency axis (period in the frequency direction) is 1 / n. In other words, the frequency characteristic has a comb shape that repeats a number of passbands and cutoff bands according to the value of n. In this case, there is no overshoot 'or ringing in any of the frequency regions, and a good comb filter characteristic in which the passband and the cutoff band appear at equal intervals is obtained.

Comb of Figure 6 shown above filter 2, eight D-type flip as tapped delay line flop 1 2 - 1 2 - using ₈ connected in cascade, each D-type flip Tsu flop 1 2 -, - The delay amounts of 1 2 to ₈ are each set for one horizontal scanning period (1 H) (η = 1 Η-1). Incidentally, the D-type flip-flop 1 2 ~ 1 D-type flip-flop 1 1 preceding the 1 2 _8, Ru der intended for temporarily retaining the digital Composite preparative video signal input .

In the case of NTSC signal, one frame (two fields) scans 5 2 5 lines Since the frequency f _sc of the color subcarrier is determined to be 450 to 2 times the horizontal scanning frequency f _H , the delay amount of each D-type flip-flop 1 2 _, to 1 2 _ ₈ Is 1 H = 4 5 5 clocks. Thus, each D-type unfavorable flop 1 2 -, by constituting the comb filter 2 delay amount of ~ 1 2 ₈ As 1 H min, YC separation circuit shown in FIG. 5 is a two-dimensional YZ C It becomes a separation type.

In the example of FIG. 6, the coefficient multiplier 1 3, as the filter coefficient of ~ 1 3 _9, and using the coefficient values of the high-pass unit filter H 1 0 shown in the lower in Figure 7 (b). If using the coefficient values of the high-pass unit Tohu filters HI n as a filter coefficient of the coefficient unit 1 3 to 1 3 _9, configured as eta = 1 .eta. 1, the comb filter 2, the 7 8 7 5 · H z The filter has a comb-type filter characteristic in which the pass band and the cut-off band are repeated at the pitch, and the peak of the pass band and the peak of the energy of the color signal appearing at every integral multiple of the horizontal scanning frequency f _H substantially match. You.

 FIG. 10 is a diagram showing a partially enlarged YZC separation filter characteristic when the comb filter 2 is configured as a 1H delay type. In the case of the conventional two-dimensional YC separation circuit shown in Fig. 4 (b), the type with a one-line delay line can only attenuate the cutoff band of about-18dB. Also, the attenuation is about -35 dB even in the 2-line type, whose attenuation characteristics are steeper than in the 1-line type. On the other hand, if the comb filter 2 of the present embodiment is used, it is possible to obtain a large attenuation of about 160 dB as shown in FIG. In addition, the bandwidth of the cutoff area is wider than that of the conventional method.

Therefore, by using the comb filter 2, the luminance signal component can be greatly attenuated from the composite video signal and only the color signal component can be extracted. In other words, the present embodiment has significantly better YC separation characteristics than the conventional method, and is almost completely independent of the state and movement of the video signal. It is possible to perform YC separation.

In the example shown in FIG. 6, the delay amount of _eight D-type flip-flops 12 to 12-8 is set to one horizontal scanning period (1H), but one vertical scanning period (IV) is used. ) Minutes are fine. For NTSC signals, slow Noberyou of the D-type flip-flop 1 2 to 1 2 _ ₈ needed to comb filter 2 and 1 V delay type, a 1 V = 5 2 5 X 9 1 0 clock Become.

Thus, coefficient unit 1 3 _, ~ 1 3 - closed filter coefficient of ₉ using the coefficient values of Haipasuyuni Tsu preparative filters HI n and, n = comb filter 2 configured as l V _ 1 is, 3 0 H The comb filter characteristic is such that the pass band and the cut band are repeated at a pitch of z, and the peak of the pass band and the peak of the energy of the color signal appearing at every integral multiple of the vertical scanning frequency f _v are almost the same. Have.

 By using both the 1 V delay type comb filter having such filter characteristics and the above-mentioned 1 H delay type comb filter, it is possible to configure a three-dimensional YC separation circuit. In the case of a three-dimensional YC separation circuit, as shown in FIG. 11, when the motion of the input video is detected and the motion value is larger than a certain level (when the correlation between frames is smaller than a predetermined value), the switch is switched. Select the delay circuit 6 with the switch 7 to select the 2D YC separation type.If the motion value is smaller than a certain level, select the 1 V delay type comb filter 2-2 and select the 3D YC It is preferable to apply motion adaptive processing so as to be separated.

As described in detail above, in the first embodiment, the unique comb filter 2 is configured using the high-pass unit filter H 1 n using a predetermined basic numerical sequence as a filter coefficient, and the comb filter is used. The delay amount of each of the D-type flip-flops 1 2—, and 1 2 to ₈ constituting filter 2 was set to one horizontal scanning period or one vertical scanning period of video. As a result, an extremely large amount of attenuation is obtained in the frequency portion of the luminance signal, which is in a relationship between the color signal and the frequency interleave. A comb filter characteristic can be obtained.

 Therefore, by forming the YC separation circuit using the comb filter 2 having such characteristics, the color signal and the luminance signal can be almost completely separated without being affected by the state or movement of the input video signal. Will be able to As a result, image disturbance such as cross-color dot crawl can be significantly reduced.

In the first embodiment, an example is described in which the coefficient values of the high-pass unit filter HI n are used as the filter coefficients of the coefficient units 13 _, to 13 _ ₉ constituting the comb filter 2. Not limited. For example, by using the coefficient value of the low-pass unit filter L 1 n as the filter coefficient of the coefficient 13-i to l 3 _ ₉ , a comb-type in which the pass band and the cut-off By obtaining the frequency characteristics described above and frequency-shifting them by a predetermined amount, the peak of the pass band and the peak of the color signal may be matched.

In the first embodiment, an example has been described in which a color signal is extracted by the comb filter 2 and a luminance signal is extracted by subtracting the extracted color signal from a digital composite video signal. On the other hand, by using the coefficient value of the low-pass unit filter L 1 n as the filter coefficient of the coefficient units 13 _, to 13 _ ₉ constituting the comb filter 2, the frequency is reduced as compared with the above example. A comb-shaped frequency characteristic in which the band and the cutoff region are inverted is obtained. Then, the luminance signal may be extracted by the comb filter 2 having such frequency characteristics, and the color signal may be extracted by subtracting the extracted luminance signal from the digital composite video signal.

Further, in order to take advantage of the YC separation characteristic shown in the first embodiment, the number of processing bits corresponding to a large attenuation in the cutoff region of the comb filter 2 is required. Is required. Since the number of pits increases during the process of increasing the resolution of an image, a higher quality image can be obtained by performing Y'C separation processing after the high definition processing.

(Second embodiment)

 Next, a second embodiment of the present invention will be described. FIG. 12 is a block diagram showing the overall configuration of the YC separation circuit according to the second embodiment. Note that, in FIG. 12, components having the same functions as the components illustrated in FIG. 5 are denoted by the same reference numerals.

As shown in FIG. 12, the YC separation circuit according to the second embodiment includes an AZD converter 1, a comb filter 2, a BPF 3 ', a subtractor 4, and a delay circuit 5'. In the present embodiment, the BPF 3 ′ is configured by a high-pass unit filter (H 11) ^m (the details of this configuration will be described later). In addition, the delay amount of the delay circuit 5 ′ is set to the same delay amount as the total delay amount of the comb filter 2 and the BPF 3 ′. In FIG. 12, the analog composite input to the YC separation circuit · The video signal is converted to digital data by the AD converter 1. The digital composite video signal output from the AD converter 1 is input to the BPF 3 'and the delay circuit 5'. The BPF 3 ′ extracts the frequency band component of the color signal from the composite video signal and outputs it to the comb filter 2. The comb filter 2 extracts a color signal having a relationship between a luminance signal and a frequency interleave using a pass band of the comb filter characteristic.

The comb filter 2 is configured in the same manner as in FIG. 6, and by setting the delay amount of the D-type flip-flop 1Sls-s to one horizontal scanning period (1H), Dimensions can be YZC separated. Further, by forming the comb filter 2 by combining the 1 H delay type and the IV delay type, it is possible to use a three-dimensional YZC separation type.

 The color signal output from the comb filter 2 is the desired color signal. The color signal output from the comb filter 2 is also input to the subtractor 4. The subtractor 4 extracts a luminance signal by subtracting the color signal from the composite video signal input from the AZD converter 1 via the delay circuit 5 '.

The details of the BPF 3 ′ (high-pass unit filter (H 11) ^m ) will be described below. Here, the cascade connection of the unit filters L 1 n and H 1 n described in the first embodiment will be considered. For example, by cascading the unit filters L ln and H n, the coefficients of each unit filter L 1 n and H n are multiplied and added to create a new filter coefficient. Hereinafter, for example, when the number of cascades of the low-pass unit fill L 1 n is m, this is described as (L in) ⁿ .

FIG. 13 is a diagram showing the frequency-gain characteristics of the low-pass unit filters L 10, (L 10) ² , (L 10) ⁴ , and (L 10) ⁸ . In Fig. 13 as well, the gain and frequency are normalized by "1". When there is only one mouth-to-pass unit filter L10, the clock at the position where the amplitude is 0.5 is 0.25. On the other hand, when the cascade number m increases, the pass band width of the filter becomes narrower. For example, when m = 8, the clock at the position where the amplitude is 0.5 is 0.125.

FIG. 14 is a diagram illustrating the frequency-gain characteristics of the high-pass unit filters H 10, (HI 0), (H 10), and (HI 0) ⁸ . In Fig. 14 as well, the gain and frequency are normalized by "1". If there is only one high-pass unit filter H10, the clock at the position where the amplitude becomes 0.5 is 0.25. On the other hand, when the cascade number m increases, the pass band width of the filter becomes narrower. For example, when m == 8, the clock at the position where the amplitude is 0.5 is 0.375.

The high-pass unit filter (H11) ^π shown in Fig. 12 has two passbands as shown in Fig. 9, and the band width of the passband is adjusted to be narrow as shown in Fig. 14. It has filter characteristics. In the present embodiment, this is used as a band-pass filter for extracting only the color signal near 3.58 {}.

FIG. 15 is a diagram showing a partially enlarged YZC separation filter characteristic when the BPF 3 ′ is configured as described above. As shown in FIG. 15, according to the present embodiment, in addition to the same structure as the comb filter 2 shown in FIG. 6, a high-pass unit filter (Η) having better attenuation characteristics than the conventional BPF 3 is provided. 1 1) By using ^m as the BPF 3 ′, an even larger out-of-band attenuation characteristic of about 72 dB can be obtained.

As described in detail above, in the second embodiment, instead of the conventional BPF 3 shown in FIG. 5, an eight-pass unit filter (H11) ^B having good attenuation characteristics in the frequency portion of the luminance signal is used. Since it is used as the BPF 3 ′, a larger out-of-band attenuation characteristic can be obtained as compared with the first embodiment. Therefore, the YC separation can be performed in a more complete manner without being affected by the state or movement of the input video signal, and video interference such as cross-color / dot crawl can be significantly reduced.

In addition, the fact that the band width of the passband becomes narrower when a plurality of unit filters are connected in cascade can be applied as follows. Generally, the bandpass of the chrominance signal is narrower than the bandpass of the luminance signal. Therefore, the band width of the pass band of the color signal is adjusted to be narrow by cascade-connecting several stages of the comb filters 2. In this way, the frequency characteristic of the comb filter 2 is It can be adjusted to the band width of the signal pass band, and it becomes possible to separate only the color signal more precisely.

 In addition, by removing the comb filter 2 from the configuration of FIG. 12, it is also possible to configure a one-dimensional YC separation circuit. In this case, B PF 3 ′ can obtain a larger out-of-band attenuation characteristic than the conventional B PF 3. Therefore, compared to the conventional one-dimensional YC separation circuit, the YC separation can be performed in a more complete form, and image disturbance such as cross-color / dot crawl can be significantly reduced. In the first and second embodiments, the example in which the numerical sequence shown in FIG. 7B is used as the filter coefficient of the comb filter 2 has been described, but other numerical sequences may be used. The meaning of the numeric sequence shown in Fig. 7 (b) and other possible numeric sequences are described in detail below. In the following description, the parameters m and n are used, but have completely different meanings from those described above.

 Here we consider several types of basic filters with specific impulse responses. Basic filters include two types: basic low-pass filter and basic high-pass filter. Hereinafter, these basic filters will be described.

The basic mouth-one-pass filter L man (where m and n are variables and n is a natural number)> The filter coefficient of the basic mouth-one-pass filter L man is based on the numerical sequence of “1 1, m, 1 1” It is obtained by a moving average calculation in which the original data before the calculation and the previous data before the calculation by a predetermined delay amount are sequentially added.

Figure 1.6 shows the filter coefficients of the basic low-pass filter L 4 an (when m = 4). In Fig. 16, from the top of n columns by moving average calculation; When calculating the th filter coefficient, the original data refers to the j-th data from the top of the (n-1) column. The previous data is', (n Indicates the (j-1) th data from the top of the 1 (1) th column.

 For example, the first numerical value “—1” from the top of the basic mouth one-pass filter L 4 a 1 is obtained by adding the original data “1 1” and the previous data “0”, and the second numerical value The numerical value "3" is obtained by adding the original data "4" and the previous data "1-1". The third number "3" is obtained by adding the original data "-1" and the previous data "4", and the fourth number "-1" is obtained by adding the original data "0" and the previous data "4". — 1 "is obtained by adding

 Each of the filter coefficients of the basic low-pass filter L 4 an shown in FIG. 16 has the property that its numerical sequence is symmetrical, and that the sum of every other digit in the numerical sequence has the same sign and is equal to each other. (For example, in the case of the basic low-pass filter L 4 a 4, −1 + 9 + 9 + (— 1) = 16, 0 + 16 + 0 = 16). The numerical sequence of the basic one-pass filter L4a4 is the same as the numerical sequence shown in the upper part of FIG. 7 (b).

The numerical sequence of "1-1, m,-1" is generated based on the original numerical sequence "-1, N". The basic unit filter using this numerical sequence "—1, N" as a filter coefficient has one or two taps (one when N = 0, and two taps otherwise). Note that the value of N does not necessarily have to be an integer. Since the basic unit filter having this numerical sequence "1,, Ν" as a filter coefficient is an asymmetric type, in order to make it a symmetric type, it is cascaded evenly. Must be used. For example, when two cascaded, the numerical sequence "A 1, New" by convolution of the filter coefficient is "A New, New ² + 1, one New". In here, when the ^{(N 2 + l) ZN =} m, when m is an integer, N = (m + (m 2 - 4), / 2) a / 2.

Assuming m−4 as in the example of FIG. 16, Ν = 2 + ΛΓ3. Sandals That is, the coefficient of the basic unit filter is "1-1, 3.732" (in this case, three decimal places are displayed). Also, the filter coefficient when this basic unit filter is connected in cascade in two stages is "_3.732, 14.92, -3.732". This numeric sequence has the relationship _1: 4: -1.

 If this numerical sequence is actually used as a filter coefficient, the value sequence of the filter coefficient can be calculated by dividing each value of the numerical sequence by 2 Ν (= 2 * (2 + 3) = 7.464). Set the gain to "1" so that the amplitude when FF Τ conversion is "1". That is, the numerical sequence of the filter coefficients actually used is “— 1/2, 2, _ 1/2”. The actual numerical sequence "1-12, 2,-1 2" is obtained by multiplying the original numerical sequence "-1, 4, 1 1" by X times (X = 1 / (m-2)). Equivalent to.

 When the scaled numerical sequence is used as a filter coefficient, the filter coefficients of the basic one-pass filter Lman are such that the sum of the numerical sequence is "1", and the sum of the numerical sequence is "1". It has the property that the total value is equal to each other with the same sign. .

FIG. 17 is a diagram illustrating a hardware configuration of the basic low-pass filter L4a4 (when m = 4 and n = 4). As shown in FIG. 17, the basic port—passfill L 4 a 4 is composed of a FIR operation unit 101 having a numerical value sequence “1 1 2, 2,-1/2” as a starting point as a filter coefficient, And a moving average calculation unit 201 for performing a moving average calculation on the numerical value sequence. Of these, the FIR operation unit 101 consists of two cascade-connected D-type flip-flops 7 1 to 7 1 _-2 , 3 coefficient units 7 2-, to 7 2 _-3, and 2 subtractors. It is composed of a to a.

Two D-type flip-flop 7 1-7 1 _ ₂ sequentially delays the input data by one clock CK. Three coefficient units 7 2 -, ~ 7 2 - _3, each D-type flip The signals extracted from the input / output taps of the flip-flops 7 1 7 1 ₂ are multiplied by filter coefficients of 1/2 and 2 1Ζ2, respectively. The first subtractor 73_, subtracts the multiplication result of the first coefficient unit 7 from the multiplication result of the _second coefficient unit 72_2. The second subtracter 7 3 _ ₂ subtracts a third coefficient multiplier 7 2 ₃ multiplication results from the first subtracter 7 3 subtraction result of.

In addition, the moving average calculation unit 201 is configured by cascade-connecting four integrators 7 4 7 4 „ ₄ each having the same configuration, for example, the first-stage integrators 7 4 —, The D-type flip-flop 75 that delays the input data by one clock, the original data that does not pass through the D-type flip-flop 75, and the previous data that is delayed through the D-type flip-flop 75. It is composed of an adder 76_ for addition, and an adjuster 7 for returning the amplitude of the addition result to the original value.

In the configuration of the basic low-pass filter L 4 a 4 shown in FIG. 17, the coefficient unit 7 2 Ί 2 _ ₃ for multiplying the filter coefficient and the data outlet to the coefficient unit 7 2 7 2 ₃ An output tap is required only for the first-stage FIR operation unit 101. , And the number is only three.

Further, since the value of the filter coefficient is 1 2, 2 1 2, coefficient unit 7 2 7 2 - ₃ can be constituted by a bit Toshifu bets circuit. Further, the four integrators 7 4 7 4 ₄ comprises regulator 7 7 7 7 _{- 4} also can be configured in pit Toshifu bets circuit. Even if the value of n is set to a value other than 4 and the number of adjusters changes, all of the adjusters can be configured by bit shift circuits. Therefore, no multiplier is required in the hardware configuration of the basic low-pass filter L 4 an.

Although the case where m = 4 has been described here, if m = 2 ¹ (i is an integer), all coefficient units and adjusters can be configured by bit shift circuits. No multiplier is needed. FIG. 18 is a diagram illustrating frequency characteristics (frequency-gain characteristics and frequency-phase characteristics) obtained by performing FFT transform on a numerical sequence of filter coefficients of the basic low-pass filter L4a4. Here, the gain is represented by a linear scale, and the standardized gain is shown as 32 times. On the other hand, the frequency is standardized at "1".

 As can be seen from FIG. 18, the frequency-gain characteristics show that the passband is almost flat and the cutoff band has a gentle slope. In addition, an almost linear characteristic is obtained in the frequency-phase characteristic. In this way, by simply configuring as shown in FIG. 17, it is possible to obtain a single-pass filter having an excellent frequency characteristic without over-sling.

 Fig. 19 is a diagram showing frequency-gain characteristics of the basic mouth one-pass filter L4 an using n as a parameter. (A) represents the gain on a linear scale, and (b) represents the gain on a logarithmic scale. It is represented by From FIG. 19, it can be seen that the greater the value of n, the steeper the slope of the cutoff region.

In FIG 1 7, D-type flip-flop 7 1 ~ 7 1 _ _2, 7 5, are all one clock delay amount of 1-7 5 _4, but for example, all of these one horizontal scanning period ( By using 1 H), a filter similar to the comb filter 2 shown in FIG. 6 (except that a high-pass unit filter type is used as the filter coefficient in FIG. 6) can be configured. . In other words, by setting the delay amount to 1H, the frequency characteristic becomes a comb shape that repeats a passband and a cutoff band for 1H, and there is no overshoot or ringing in any frequency region. In addition, a good comb filter characteristic in which the pass band and the cutoff region appear at equal intervals is obtained.

Figure 20 is a diagram showing the filter coefficients of the basic one-pass filter Lan when N = 0 in the numerical sequence '' —1, N ”of the basic unit filter. When two unit filters are connected in cascade, the filter coefficient is "0, 1, 0". Therefore, the filter coefficient of the basic D-pass filter Lan is obtained by a moving average operation in which the original data and the previous data are sequentially added starting from "1".

 Each filter coefficient of the basic one-pass filter Lan shown in Fig. 20 has the property that its numerical sequence is symmetrical, and that the total value of each skip of the numerical sequence is the same sign and equal to each other. (For example, in the case of the basic mouth one-pass filter La4, 1 + 6 + 1 = 8, 4 + 4 = 8).

FIG. 21 is a diagram illustrating a hardware configuration of the basic low-pass filter La 4 (when n = 4). Here, since the starting point is a single "1", the two D-type flip-flops 7 1 to 7 1 shown in Fig. 25 have three coefficient units 7 2-\ ~. Ί 2 — ₃ and two subtractors 7 3 7 3 — ₂ are not required. That is, four integrators 7 4 _ the second half illustrated in Figure 1 7, 1-7 4 - can constitute the basic low-pass filter L a 4 by simply cascading _4. Therefore, the number of taps is 0, and no multiplier is required.

 FIG. 22 is a diagram illustrating frequency characteristics obtained by performing FFT conversion on a numerical sequence of filter coefficients of the basic low-pass filter La4. Here, the gain is represented by a linear scale, and the normalized gain is shown by 16 times. On the other hand, the frequency is normalized by "1".

 As can be seen from FIG. 22, the almost flat passband in the frequency-gain characteristic is narrower than that in FIG. 18, but the slope of the cutoff region has a gentle characteristic. In addition, almost linear characteristics are obtained in the frequency-phase characteristics. Thus, a low-pass filter having good frequency characteristics with no overshoot or ringing can be obtained only by configuring as shown in Fig. 21.

FIG. 23 is a graph showing the frequency-gain characteristics when n of the basic low-pass filter Lan is set as a parameter, and (a) shows the gain by a linear scale. (b) shows the gain on a logarithmic scale. From FIG. 23, it can be seen that the greater the value of n, the steeper the slope of the cutoff region.

<Basic high-pass filter Hm sn (m and n are variables, n is a natural number)> The filter coefficients of the basic high-pass filter H msn are the original data before the calculation, starting from the numerical sequence of "1, m, 1" And a moving average calculation that sequentially subtracts the previous data that is a predetermined amount of time before the data.

 FIG. 24 is a diagram showing filter coefficients of the basic high-pass filter H 4 sn (when m = 4). In Fig. 24, when calculating the j-th filter coefficient from the top of the n-th column by the moving average operation, the original data refers to the:) '-th data from the top of the (n-1) th column. Also, the previous data refers to the (j1) th data from the top of the (n-1) th column.

 For example, the first numerical value “1” from the top of the basic high-pass filter H 4 s 1 is obtained by subtracting the previous data “0” from the original data “1”, and the second numerical value “3” is the original data “ Obtained by subtracting the previous data "1" from 4 ". The third numerical value “—3” is obtained by subtracting the previous data “4” from the original data “1”, and the fourth numerical value “1 1” is obtained by subtracting the previous data “1” from the original data “0”. "Is obtained by subtracting

In the basic high-pass filter H 4 sn shown in Fig. 24, when n is an even number, for any filter coefficient, the numerical sequence is symmetric, and the total value of the numerical sequence jumps is equal to each other with the opposite sign. (For example, in the case of the basic high-pass filter H 4 s 4, 1 + (— 9) + (— 9) + 1 =-16, 0 + 16 + 0 = 16). The numerical sequence of this basic high-pass filter H 4 s 4 is the same as the numerical sequence shown in the lower part of FIG. 7B. On the other hand, when n is an odd number, the numerical sequence has a symmetric absolute value, and the first half and the second half have opposite signs. Also, one jump in the numeric sequence Have the property that the sum of the values is equal to each other with the opposite sign.

 The numerical sequence of "1, m, 1" is generated based on the original numerical sequence "1, N". The basic unit filter that uses this numerical sequence "1, N" as a filter coefficient has one to two taps (one for N = 0, two otherwise). Note that the value of N need not be an integer.

Since the basic unit filter having this numerical sequence "1, N" as a filter coefficient is asymmetric, it is necessary to use an even number of cascades to make it symmetric. For example, when two cascaded by narrowing seen tatami numerical sequence "1, N", the filter coefficient is ^{"N, N 2 + 1,} N". Here, the ^{(N 2 + 1) / N} = a m, when m is an integer, N = - a ^{(m + (m 2 4)} 1/2) / 2.

 When m = 4 as in the example of FIG. 24, Ν = 2 + ΛΓ3. That is, the coefficient of the basic unit filter is "1, 3.73.2" (three decimal places are displayed here). Also, the filter coefficients when the two basic unit filters are connected in cascade are “3.732, 14.928, 3.732”. This number sequence has a 1: 4: 1 relationship. If this number sequence is actually used as a filter coefficient, each value of the number sequence is 2 Ν (= 2 * (2 + 3) = 7. By dividing by 4 6 4), the amplitude when the numerical sequence of filter coefficients is FFFFconverted becomes “1”, and the gain is normalized to “1”. That is, the numerical sequence of the filter coefficients actually used is "1Ζ2, 2, 1/2". This actually used numerical sequence "1 no 2,2,1 2" also corresponds to the original numerical sequence "1,4,1" multiplied by X (X = 1 / (m-2)).

When a numerical sequence standardized in this way is used as a filter coefficient, the filter coefficients of the basic high-pass filter H msn are equal to the total of the numerical sequence. The sum is "0", and it has the property that the total value of one jump of a numerical sequence is equal to the opposite sign.

FIG. 25 is a diagram illustrating an eight-one hardware configuration of the basic high-pass filter H 4 s 4 (when m = 4, n = 4). As shown in Fig. 25, the basic high-pass filter H 4 s 4 is composed of a FIR operation unit 102 having a starting value sequence “1/2, 2, 1 Z 2” as a filter coefficient, And a moving average calculating section 202 for calculating a moving average of a column. Of these, the FIR operation unit 102 is composed of two cascaded D-type flip-flops 8 1 _, to 8 1 _ ₂ , three coefficient units 8 2 _, to 8 2 _ ₃ , and two of the adder 8 3, composed of Ri by the and - 8 3 _2.

The two D-type flip-flops 8 1-1, to 8 _1-2 sequentially delay the input data by one clock CK. Three coefficient units 8 2-8 2 _ _3, to the signal taken out from the D-type flip-flop 8 1 ~ 8 1 ₂ of the input and output taps, 1/2, 2, 1/2 filter coefficients Are respectively multiplied. Two adders 8 3-8 3 - _2, the coefficient units 8 2, to force out by adding all multiplication results of 1-8 2 _ _3.

In addition, the moving average calculation unit 202 that performs a moving average calculation on the above-described numerical sequence “1 Z 2, 2, 1/2” has four differentiators 84 „, to 84 _, each of which has the same configuration. ₄ constructed by cascading. for example the first stage of the differentiator 8 4 or a D-type flip-flop 8 5 for delaying one clock input data does not pass through the D-type flip-flop 8 5 yuan It is composed of a subtracter 86_, which subtracts the previous data delayed through the D-type flip-flop 85 from the data, and an adjuster 87 for restoring the amplitude of the subtraction result.

In the configuration of the basic high-pass filter H 4 s 4 shown in FIG. 25, the coefficient units 8 2 _, to 8 2 and the coefficient units 8 2 to 8 Only the first-stage FIR operation unit 102 needs an output tap, which is an outlet for data to 8 2-3. And the number is only three.

Furthermore, the values of the filter coefficients 1 2, '2, 1 because it is Z 2, coefficient multipliers 8 2-8 2 _ ₃ can be constituted by a bit Toshifu bets circuit. Further, the four differentiator 8 '4 _ ~ 8 4 - ₄ adjuster 8 7-8 _7-4 provided even can be configured with bit Bok shift circuit. Even if the value of n is set to a value other than 4 and the number of adjusters changes, all of the adjusters can be configured by bit shift circuits. Therefore, no multiplier is required even in the hardware configuration of the basic high-pass filter H 4 sn.

Although the case where m = 4 has been described here, if m = 2 ¹ (i is an integer), all coefficient units and adjusters can be configured by bit shift circuits. No multiplier is needed.

 FIG. 26 is a diagram illustrating a frequency characteristic obtained by performing FFT conversion on a numerical sequence of filter coefficients of the basic high-pass filter H4s4. Here, the gain is shown by a linear scale, and the normalized gain is shown by 32 times. On the other hand, the frequency is normalized by "1".

 As can be seen from FIG. 26, the frequency-gain characteristics show that the passband is almost flat and the cutoff band has a gentle slope. In addition, almost linear characteristics are obtained in the frequency-phase characteristics. Thus, a high-pass filter having good frequency characteristics without over-shoot ringing can be obtained only by configuring as shown in FIG. 25.

 Fig. 27 shows the frequency-gain characteristics of the basic high-pass filter H4sn with n as a parameter. (A) shows the gain on a linear scale, and (b) shows the gain on a logarithmic scale. ing. From FIG. 27, it can be seen that the greater the value of n, the steeper the slope of the cutoff region.

In FIG 2 5, D-type flip-flop 8 1-8 _1.2 8 5 ~ 8 5 _ ₄ Are all one clock, but by setting them all for one horizontal scanning period (1H), a filter having a function equivalent to the comb filter 2 shown in Fig. 6 is constructed. be able to. In other words, by setting the delay amount to 1H, the frequency characteristic becomes a comb shape that repeats a passband and a cutoff band for 1H, and there is overshoot ringing in any frequency region. A good comb filter characteristic is obtained, in which the passband and the cutoff band appear at equal intervals.

 Fig. 28 is a diagram showing the filter coefficients of the basic high-pass filter H sn when N = 0 in the numerical sequence "1, N" of the basic unit filter. When two stages are connected in cascade, the filter coefficient is “0, 1, 0.” Therefore, the filter coefficient of the basic high-pass filter H sn is obtained by sequentially subtracting the previous data from the original data starting from “1”. It is obtained by moving average calculation.

In the basic high-pass filter H sn shown in Fig. 28, when n is an even number, for any filter coefficient, the numerical sequence is symmetric, and the total value of each skip in the numerical sequence is equal to each other with the opposite sign. (For example, in the case of the basic high-pass filter Hs4, 1 + 6 + 1 = 8,-4 + (-4) =-8). When n is an odd number, the numerical sequence has a symmetric absolute value, and the first half and the second half have opposite signs. Figure 29 shows the eighty-one window configuration of the basic high-pass filter H s 4 (assuming n = 4). FIG. Here, since the numerical sequence as the starting point is a single "1", 2 two D type shown in 5 flip-flop 8 1 ~ 8 1 - _2, three coefficient units 82 -, ~ 8 _2-3 and two adders 8 3 _, ~ 8 3 _ ₂ are unnecessary. That is, the differentiator in the latter half shown in Figure 25 4 _ ₄ simply cascaded at can constitute a basic highpass filter H s 4. Therefore, the number of taps is 0, and no multiplier is required.

 FIG. 30 is a diagram illustrating frequency characteristics obtained by performing FFT conversion on a numerical sequence of filter coefficients of the basic high-pass filter Hs4. Here, the gain is represented by a linear scale, and the normalized gain is shown by 16 times. On the other hand, the frequency is normalized by "1".

 As can be seen from FIG. 30, the passband that is almost flat in the frequency-gain characteristic is narrower than that in FIG. 26, but the slope of the cutoff region has a gentle characteristic. In addition, almost linear characteristics are obtained in the frequency-phase characteristics. Thus, a high-pass filter having good frequency characteristics with no overshoot or ringing can be obtained only by configuring as shown in FIG. 29.

 Fig. 31 shows the frequency-gain characteristics of the basic high-pass filter Hsn with n as a parameter. (A) shows the gain on a linear scale.

(b) represents the gain on a logarithmic scale. From FIG. 31, it can be seen that as the value of n increases, the slope of the cutoff region becomes steeper.

Of parameter values m and n on characteristics>

 First, the effect of changing the number of stages n of the moving average calculation will be described. For example, as shown in FIG. 19, in the basic mouth one-pass filter L ma n, when the value of n is increased, the slope of the cutoff band becomes steeper, and the bandwidth of the pass band becomes narrower. When the value of n is small, both ends of the top of the frequency characteristic rise. As the value of n increases, the top gradually approaches flat, and becomes completely flat at n = 4. As the value of n increases, both ends of the top become lower than the median. This tendency can be similarly applied to the basic high-pass filter H msn (see FIG. 27).

On the other hand, a basic roper configured with the coefficient value of the basic unit filter N = 0 As shown in FIG. 23 and FIG. 31 for the filter L an and the basic high-pass filter H sn, both ends of the top are lower than the median in any case of the value of n. When the value of n is increased, the slope of the cutoff band becomes steeper, and the bandwidth of the passband becomes narrower, as in the case of the basic mouth-one-pass filter L man and the basic high-pass filter H msn where N ≠ 0. .

 Next, the effect of changing the value of m will be described. FIG. 32 is a diagram illustrating a frequency-gain characteristic of the basic high-pass filter H msn where m is a parameter. From FIG. 32, it can be seen that when the value of m is reduced, the slope of the cutoff region becomes steeper, and the bandwidth of the passband becomes narrower. Although not shown here, the same can be said for the basic low-pass filter Lman.

 Fig. 32 also shows the optimal value of parameter n for parameter m (the value of n at which the top of the frequency characteristic becomes flat). That is, when m = 4, the optimal value is n = 4, when m = 3.5, the optimal value is n = 6, when m = 3, the optimal value is n = 8, when m = 2.5 The optimal value of is n = l6. Figure 33 is a graphical representation of this. As can be seen from FIG. 33, the optimum value of the parameter n for the parameter m becomes larger as the value of m becomes smaller.

 This will be described in more detail with reference to FIG. FIG. 34 is a diagram showing, in a table form, the relationship between the nighttime m and the optimum value of the parameter n. In addition, FIG. 34 also shows the relationship between “no” and “parameter X” to “lame night” m.

As described above, the optimal value of the parameter n with respect to the parameter m increases as the value of decreases. Here, when m = 2, the filter characteristics change significantly, and good frequency characteristics cannot be obtained. Conversely, if m> 2, the delay introduced between taps is increased. Even without this, good filter characteristics with a narrow band width in the pass band can be obtained. Meanwhile, no. The optimum value of the parameter n decreases as the value of the parameter m increases, and n = l when m = 10. In other words, when m = 10, the number of stages for the moving average calculation may be one. For this reason, it is preferable that the parameter m be used under the condition of 2 <m≤10.

 The value of parameter n is adjusted to the frequency characteristic as shown in Fig. 19 and Fig. 23 by using an arbitrary value selected in a certain range before and after the optimum value shown in Fig. 34. be able to.

 FIG. 35 is a diagram showing the impulse responses of the four types of basic high-pass filters Hmsn shown in FIG. The impulse response having the waveform shown in Fig. 35 has a finite value other than "0" only when the sample position along the horizontal axis is constant, and the value in other regions Are all "0" functions, that is, finite functions whose values converge to "0" at a given sampling position. Although not shown here, the impulse response of the basic high-pass filter Hsn and the basic low-pass filters Lman and Lan is of a finite range, as in the case of the Gran.

 As described above, in such a finite impulse response, only data in a local region having a finite value other than "0" is significant. Data outside this area should not be neglected, although it should be considered, and there is no need to consider it theoretically, so there is no truncation error. Therefore, if the numerical sequences shown in FIGS. 16, 20, 24, and 28 are used as the filter coefficients, it is not necessary to cut off the coefficients for ^; Obtainable.

Here, an example was described in which the FIR calculation unit was configured by cascading the basic unit filters in two stages, but even stages other than two stages, such as four stages, six stages, eight stages, and so on, were connected. To compose the FIR operation unit You may.

 Also, here, an example was described in which the comb filter 2 was configured (in the form of the basic filters Lman, Lan, Hmsn, and Hsn) by cascading the FIR calculation unit and the moving average calculation unit. However, one digital filter is simply composed of a plurality of D-type flip-flops, a plurality of coefficient units, and a plurality of adders, and one of the numerical values shown in FIG. 16, FIG. 20, FIG. 24, and FIG. The comb filter 2 may be configured such that a row is set in a plurality of coefficient units in the digital filter.

 In the first and second embodiments, the NTSC system has been described as an example of a video signal. However, the present invention can be similarly applied to other signal standards such as the PAL system and the SECAM system.

 Further, in the first and second embodiments, an example has been described in which the technique of YC separation is realized by a hardware configuration. However, it is also possible to realize the technique by DSP software. For example, when implemented by software, the YC separation device of this embodiment is actually configured with a computer CPU or MPU, RAM, ROM, etc., and the program stored in RAM or R〇M is stored. It can be realized by operating.

Therefore, the present invention can be realized by recording a program that causes a computer to perform the functions of the above embodiments on a recording medium such as a CD-ROM, and reading the program into a computer. As a recording medium for recording the above program, a flexible disk, a hard disk, a magnetic tape, an optical disk, a magneto-optical disk, a DVD, a nonvolatile memory card, and the like can be used in addition to the CD-ROM. In addition, the above program can be realized by downloading the program via a network such as the Internet at a short time. Further, in order to realize the function of YC separation according to each of the above embodiments in a network environment, all or some of the programs may be executed on another computer.

 Further, not only the functions of the above-described embodiments are realized by the computer executing the supplied program, but also the OS (operating system) in which the program is running on the computer. If the functions of each embodiment are realized in cooperation with other application software, etc., or if all or part of the processing of the supplied program is performed by a computer function expansion port or function expansion unit, Such a program is also included in the embodiments of the present invention when the functions of each embodiment are implemented by the program.

 In addition, each of the first and second embodiments is merely an example of a concrete embodiment for carrying out the present invention, and the technical scope of the present invention is interpreted in a limited manner. It must not be. That is, the present invention can be implemented in various forms without departing from the spirit or main features thereof. Industrial applicability

 INDUSTRIAL APPLICABILITY The present invention is useful for a composite video signal YC separation circuit and YC separation method for separating a color video signal and a luminance signal from a composite video signal formed by frequency-multiplexing a color signal and a luminance signal. It is.

Claims

Range of request

1. Digital filter processing of digital composite video signals to extract one of the color signal and the luminance signal, which are in a frequency-in-leave relationship,

 A subtractor for subtracting the one signal extracted by the comb filter from the digital composite video signal and extracting the other signal,

 The comb filter multiplies each signal of each tap in a delay line with taps composed of a plurality of delay devices by a filter coefficient composed of a predetermined basic numerical sequence, and then adds and outputs the result. A YC separation circuit for a composite video signal, wherein the delay amounts of the plurality of delay units are set to one horizontal scanning period or one vertical scanning period, respectively. .

 2. The filter coefficients consisting of the predetermined basic numerical sequence are such that the numerical sequence is symmetrical, the total value of the numerical sequence is non-zero, and the total value of each of the numerical sequence is the same. 2. The composite video signal YC separation circuit according to claim 1, wherein the values are set so that the signs are equal to each other.

 3. The above-mentioned basic basic numerical sequence is the sum of the original data before the operation and the previous data that is a predetermined number before the numerical sequence consisting of the ratio of “—1, m, — 1”. 3. The composite video signal YC separation circuit according to claim 2, wherein the YC separation circuit is a numerical sequence obtained by repeatedly performing a moving average operation for adjusting an amplitude n times.

4. The numerical sequence of the ratio '' — 1, m, — 1 ”is characterized by multiplying the numerical sequence of“ 1-1, m, — 1 ”by 1 (m—2). Billing YC separation circuit for composite video signal according to range 3.

 5. The filter coefficients consisting of the predetermined basic numerical sequence are such that the numerical sequence is symmetrical, the total value of the numerical sequence is zero, and the total value of the numerical sequence is one step. 2. The YC separation circuit for composite video signals according to claim 1, wherein the values are set so that the signs are equal to each other.

 6. The predetermined basic numerical sequence is the amplitude adjustment by subtracting a predetermined number of previous data from the original data before calculation from the original data before the operation for the numerical sequence consisting of the ratio of "1, m, 1". 6. The YC separation circuit for composite video signals according to claim 5, wherein the YC separation circuit is a numerical sequence obtained by repeatedly performing a moving average operation n times.

 7. The numerical sequence having a ratio of "1, m, 1" is obtained by multiplying the numerical sequence of "1, m, 1" by 1 (m-2). The composite video signal YC separation circuit described in section 6.

 8. The YC separation circuit for a composite video signal according to claim 3, wherein the number n of repeating the moving average calculation is 8 (m−2) times.

 9. The composite video signal YC separation circuit according to claim 6, wherein the number n of repeating the moving average calculation is 8 (m−2) times.

 10. The composite video signal YC separation circuit according to claim 8, wherein the value of m satisfies the condition of 2 <m≤10.

11. The composite video signal YC separation circuit according to claim 1, wherein the comb filter is configured by cascading a plurality of stages.

12. A YC separation of the composite video signal according to claim 1, further comprising a band-pass filter for extracting one of the color signal and the luminance signal. circuit.

 13. The band-pass filter uses a filter that uses the predetermined basic numerical sequence as a filter coefficient as a unit filter, and delays n clocks between taps corresponding to the filter coefficient. The filter is designed by arbitrarily connecting at least one of the filter whose pass frequency band has been adjusted by introducing a filter and at least one of the unit filters. The described composite video signal YC separation circuit.

 14. A digital filter that multiplies the signal of each tap in the delay line with taps consisting of a plurality of delay units by a filter coefficient consisting of a predetermined basic numerical sequence, and then adds and outputs the result. Then, using a comb filter in which the delay amount of each of the plurality of delay units is set to one horizontal scanning period or one vertical scanning period, the color signal and the luminance signal have a frequency interleaving relationship. A composite video, wherein one signal is extracted from the digital composite video signal and the other signal is extracted by subtracting the one signal from the digital composite video signal. YC separation method of signal.