JPH09224186A - Video camera and control correcting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To add a detail without deteriorating an S/N ration near black with respect to a video signal. SOLUTION: BPF 26 is varied within a range in which a central frequency is 2 to 8MHz. The central frequency of BPF 35 is made lower (about 2MHz) than that of BPF 26. LPF 28 extracts a vertically low band and horizontally low band signal VLHL and supplies this signal to a nonlinear circuit 36 to control the mixing ratio of the output signal of BPF 26 and that of BPF 35 by means of the output of the nonlinear circuit 36. The mixing ratio of the output of BPF 35 is increased near black. Thereby the detail is added to the neighborhood of black without increasing noise. A mixer 30 mixes horizontal detail components from a mixer 37 and vertical components from a multiplier 27, and a detail signal DTL is taken out to an output terminal 31 from the mixer 30.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contour correction device capable of varying the frequency characteristic of a detail signal according to the level of a luminance signal, and a video camera provided with the contour correction device.

[0002]

2. Description of the Related Art Conventionally, a contour correction device used in a signal processing circuit of a video camera has a structure as shown in FIG. In FIG. 1, 0H (which means that the delay amount is 0), 1H, and 2H delayed outputs of the red signal R are supplied to the HPF (high-pass filter) 21a and the LPF (low-pass filter) 21b in the vertical direction, respectively. Also, the 0H, 1H, and 2H delay outputs of the green signal G are respectively the HPF 22a and the LPF in the vertical direction.
22b. Further, the blue signal B has 0H and 1
The H and 2H delay outputs are the HPF 23a in the vertical direction.
And LPF 23b.

The HPFs 21a, 22a, 23a generate high frequency signals in the vertical direction of the respective color signals, and the HPF2
The output signals of 1a, 22a and 23a are mixed by a mixer 24a at a predetermined ratio. The vertical high frequency signal VH from the mixer 24 a is passed through the horizontal LPF 25 to generate a vertical high frequency and horizontal low frequency signal VHHL, and this signal is supplied to the mixer 30.

Vertical LPFs 21b, 22b, 23b
Generates a low-frequency signal in the vertical direction of each color signal, and L
The output signals of the PFs 21b, 22b, and 23b are mixers 24b.
Are mixed in a predetermined ratio. The vertical low-pass signal VL from the mixer 24b is passed through a horizontal BPF (bandpass filter) 26 to generate a vertical low-pass and horizontal high-pass signal VLHH. The BPF 26 is configured so that the center frequency (usually about 2 to 8 MHz) can be changed by manual adjustment according to the user's preference and the characteristics of the image being captured. This signal VLHH is supplied to the mixer 30. The mixer 30 adds the vertical high band / horizontal low band signal VHHL to the vertical low band / horizontal high band signal VLHH, and the added output signal is supplied to the multiplier 27.

In addition, the vertical low frequency signal V from the mixer 24b
L is supplied to the horizontal LPF 28, and the LPF 28 generates a vertical low band and horizontal low band signal (DC component) VLHL, and this signal VLHL is supplied to the non-linear circuit 29. The non-linear circuit 29 has, for example, a comparator, and has a non-linear input-output characteristic. The non-linear circuit 29 performs non-linear processing such that 0 is provided near the black portion of the input signal and 1 is provided when the input signal has a predetermined magnitude or more. Such processing is level dependent. The output signal of the nonlinear circuit 29 is supplied to the multiplier 27, and the multiplier 27 controls the gain of the output signal of the mixer 30. Then, the detail signal DTL is taken out from the output terminal 31.

Since the 1H delay line has a large circuit scale and is expensive, it is used only for the G channel or only for the G and R channels, so that the vertical high band signal, horizontal low band signal VHHL, and vertical low band signal, horizontal high band signal can be obtained. VLHH may be generated. In such a case, the channel without the 1H delay line is not provided with the LPF in the vertical direction, but is directly supplied to the mixers 24a and 24b for the horizontal detail.
In FIG. 1, blocks 21a, 21b and 2 indicated by broken lines
Reference numerals 3a and 23b denote circuit blocks that can be omitted when only the G channel is used.

As described above, in the conventional contour correction device, the gain of the detail signal near the black level is reduced by the non-linear circuit 29 to prevent the deterioration of the S / N ratio. But,
In a digital signal processing camera, when noise near black is removed, the detail disappears, and when the detail is not erased, the noise becomes conspicuous.

In video cameras, a gamma correction circuit is usually provided. The input-output characteristic of the gamma correction circuit has a large gain for low-level inputs, so
In the image pickup signal, the input video level near black is amplified about 4 times by gamma correction. The same applies when the detail signal generated by the contour correction device passes through such a gamma correction circuit. As a result, the S / N ratio deteriorates, and noise near black is most noticeable to human eyes.

Therefore, in the above-described conventional contour correcting device, the DC component that has passed through the non-linear circuit 29 is supplied to the multiplier 27 to control the gain of the detail signal, so that the detail gain near black is reduced. Was there. Such a process is called level dependent. Also, in an analog circuit, the product of G (gain) and B (bandwidth) is generally constant, so if the gain of the gamma correction circuit increases near black, the bandwidth becomes narrower and naturally high-frequency components also occur. descend. Therefore, it was not necessary to make the level dependency so effective.

[0010]

However, unlike the analog circuit, in the digital circuit, the frequency characteristic does not change even if the gain near black is increased by the gamma correction circuit. That is, the high frequency component does not decrease and the high frequency noise increases as the gain increases, so that the level dependency must be strongly applied. However, as a result of the level dependency being applied too strongly, the detail signal in the vicinity of black is almost eliminated, resulting in a very unnatural image. For example, when a bright image and a dark image exist, the detail signal may be added only to the bright image.

Therefore, an object of the present invention is to provide S / near black.
An object of the present invention is to provide a video camera and a contour correction device capable of adding details without deteriorating the N ratio.

[0012]

The present invention is directed to an image pickup means for picking up an image of a subject and outputting a video signal, a signal level detecting means for detecting the signal level of a DC component of the video signal, and a video signal from the video signal. Based on the output of the detail signal generating means for generating a detail signal for correcting the contour of the image represented by the video signal by extracting a part of the frequency components, and the DC component based on the output of the signal level detecting means. A control means for controlling the detail signal generating means and a mixing means for mixing the detail signal with the video signal are provided so that the frequency component of the detail signal generated by the detail signal generating means becomes lower as the signal level of It is a video camera characterized by that.

Further, according to the present invention, an image pickup means for picking up an image of a subject and outputting a video signal, a signal level detecting means for detecting a signal level of a DC component of the video signal, and a detail for generating a detail signal from the video signal. Control for controlling the detail signal generating means based on the outputs of the signal generating means and the signal level detecting means such that the boost frequency of the detail signal generated by the detail signal generating means becomes lower as the signal level of the DC component becomes lower. A video camera comprising means and mixing means for mixing a detail signal with a video signal.

Further, according to the present invention, a signal level detecting means for detecting a signal level of a direct current component of a video signal, and a part of frequency components of the video signal are extracted from the video signal to be represented by the video signal. Based on the output of the detail signal generating means for generating the detail signal for correcting the contour of the image and the signal level detecting means, the frequency of the detail signal generated by the detail signal generating means decreases as the signal level of the DC component decreases. A contour correction device for a video signal, comprising: a control means for controlling the detail signal generation means so that the component becomes low; and a mixing means for mixing the detail signal with the video signal.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. FIG. 2 shows a circuit configuration of a DSP (Digital Signal Processing) camera to which the present invention can be applied, that is, a video camera which performs digital signal processing. In FIG. 1, reference numerals 1a, 1b, and 1c denote CCDs as R, G, and B image pickup devices, respectively. CCD1
The output signals of a, 1b and 1c are the video amplifier 2 respectively.
a, 2b, 2c. And video amplifier 2
a, 2b, and 2c respectively amplify to a predetermined level to achieve white / black balance. Video amplifier 2a,
The output signals of 2b and 2c are A / D converters 3a and 3a, respectively.
It is supplied to 3b and 3c and converted into a digital signal.
The respective color signals which are digital signals are line memories 4a, 4c, 4e having a delay amount of 1H (1 horizontal section).
And the contour correction device 5 as well.

1 by the line memories 4a, 4c, 4e
The H-delayed signal is supplied to the contour correction device 5 and is also supplied to the contour correction device 5 via the line memories 4b, 4d, and 4f. Line memories 4b, 4d,
By 4f, 2H delay output of each of the R, G, and B image pickup signals can be obtained. The contour correction device 5 uses the detail signal DTL, which is a high-frequency component in the horizontal and vertical directions for contour enhancement.
Is generated. The present invention mainly relates to the contour correction device 5
It relates to the circuit configuration of. In addition, the line memory 4
The 1H delayed outputs of the R, G, and B signals of a, 4c, and 4e are supplied to the linear matrix circuit 6, and the hue and saturation are adjusted.

The detail signals from the contour correction device 5 are added to the output signals from the linear matrix circuit 6 corresponding to the respective color components of the three primary colors by the adders 7a, 7b and 7c, respectively. Then, the knee / gamma correction circuits 8a and 8
In b and 8c, non-linear processing is performed. What is Knee
This is to expand the apparent dynamic range by compressing the level of the high brightness part. What is gamma?
The CRT has an inverse characteristic of the voltage-luminance characteristic. The color signals subjected to the non-linear processing by the knee / gamma correction circuits 8a, 8b and 8c are supplied to the matrix circuit 9. The matrix circuit 9 converts the three primary color signals into Y, RY, and BY luminance / color difference signals.

Y, RY, B- from the matrix circuit 9
The Y luminance / color difference signal is output to the parallel / serial converter 10
Thus, the parallel data is converted into the serial data, and the digital serial component output signal is taken out from the output terminal 11. In addition, the matrix circuit 9
Luminance / color difference signals from the D / A converters 12a, 12
b and 12c respectively convert into analog signals and output as analog component signals. That is, the Y signal is taken out to the output terminal 13a, the RY signal is taken out to the output terminal 13b, and the BY signal is taken out to the output terminal 13c.

Further, the luminance / color difference signal from the matrix circuit 9 is converted into a digital composite signal by the encoder 14, and the digital composite output signal is taken out from the output terminal 15. Also, the digital composite output from the encoder 14 is the D / A converter 1.
An analog VBS (Video Burst Sync) output signal is extracted from the output terminal 17 by being converted into an analog signal by 6.

FIG. 3 is a block diagram of a contour correcting device according to an embodiment of the present invention. The contour correcting apparatus of this embodiment corresponds to the contour correcting apparatus 5 in FIG. In FIG. 3, 0H (meaning that the delay amount is 0), 1H, and 2H delayed outputs of the red signal R are supplied to the vertical HPF (high-pass filter) 21a and LPF (low-pass filter) 21b, respectively. Also, the green signal G
0H, 1H, 2H delay output of
It is supplied to the PF 22a and the LPF 22b. further,
The 0H, 1H, and 2H delayed outputs of the blue signal B are supplied to the HPF 23a and the LPF 23b in the vertical direction, respectively.

The HPFs 21a, 22a, 23a generate high frequency signals in the vertical direction of the respective color signals, and the HPF2
The output signals of 1a, 22a and 23a are mixed by a mixer 24a at a predetermined ratio. The vertical high frequency signal VH from the mixer 24 a is passed through the horizontal LPF 25 to generate a vertical high frequency and horizontal low frequency signal VHHL, and this signal is supplied to the mixer 30 via the multiplier 27. .

Vertical LPFs 21b, 22b, 23b
Generates a low-frequency signal in the vertical direction of each color signal, and L
The output signals of the PFs 21b, 22b, and 23b are mixers 24b.
Are mixed in a predetermined ratio. The vertical low-pass signal VL from the mixer 24b is supplied to a horizontal BPF (bandpass filter) 26, a horizontal BPF 35, and a horizontal LPF 28. The BPF 26 generates a vertical low band / horizontal high band signal VLHH1, and the BPF 35 generates a vertical low band / horizontal high band signal VLHH2.

Since the 1H delay line has a large circuit size and is expensive, it is used only for the G channel or only for the G and R channels, so that the vertical high band signal, horizontal low band signal VHHL and vertical low band signal, horizontal high band signal are used. VLHH may be generated. In such a case, the channel without the 1H delay line is not provided with the LPF in the vertical direction, but is directly supplied to the mixers 24a and 24b for the horizontal detail.
In FIG. 3, blocks 21a, 21b and 2 indicated by broken lines
Reference numerals 3a and 23b denote circuit blocks that can be omitted when only the G channel is used.

In FIG. 3, the horizontal BPF 26 is
It is a variable coefficient filter, and is adapted to be variable by manual adjustment within a center frequency range of 2 to 9 MHz according to the user's preference and the characteristics of the picture being imaged. The output signal VLHH1 of the BPF 26 is supplied to the mixer 37 and mixed with the output signal VLHH2 of the BPF 35. The output signal VLHH of the mixer 37 and the output signal of the multiplier 27 are supplied to the mixer 30, and the detail signal DTL for contour enhancement is taken out to the output terminal 31 of the mixer 30.

The horizontal output signal (DC component) VLHL of the LPF 28 is supplied to the non-linear circuits 29 and 36. The output of the non-linear circuit 29 controls the mixing ratio of the mixer 30, and the output of the non-linear circuit 36 controls the mixing ratio of the mixer 37.

The center frequency of the BPF 35 is lower than that of the BPF 26 (about 2 MHz). Further, the non-linear circuits 29 and 36 have, for example, comparators and have non-linear input-output characteristics, and the output signal is 0 near the black of the input signal and is 1 above the predetermined magnitude. Non-linear processing (level dependent processing) is performed.

The vertical low band and horizontal high band signal VLHH from the mixer 37 is a horizontal component of the detail signal DTL. Therefore, the closer the center frequency of the BPF 26 is to the higher frequency side, the thinner the contour is emphasized.
The effect of noise is large. Therefore, in the contour correction apparatus shown in FIG. 3, the center frequency of the other horizontal BPF 35 is kept low (about 2 MHz), and the horizontal component of the detail signal is generated without making noise noticeable. On the other hand, the signal VHHL, which is the vertical component of the detail signal, is not significantly affected by noise because the horizontal LPF 25 is provided.

Therefore, the vertical low band and horizontal low band signals (DC component) VLHL are supplied to the non-linear circuit 36, and the non-linear circuit 3
The mixer 37 is controlled by the output of 6, and the BPF in the horizontal direction is controlled.
The mixing ratio of the output signal VLHH1 of 26 and the output signal VLHH2 of the BPF 35 is changed. When the mixture ratio is represented by α,
The output VLHH of the mixer 37 is (VLHH = (1-α)
-It is represented by VLHH1 + αVLHH2. This mixing ratio α
The control of (1) can control the boost frequency of the vertical low-frequency and horizontal high-frequency signals VLHH.

When controlling the mixture ratio α, the influence of noise is greater near black, so the mixture ratio α is increased near black and the ratio of the output of the BPF 35 is increased.
That is, the boost frequency is lowered. As a result, the outline becomes thicker, but it is possible to add details to the vicinity of black without increasing noise. In addition, the embodiment of the present invention has an advantage that it does not give a feeling of strangeness, because the same effect as that of the phenomenon in which the frequency characteristic is deteriorated is obtained in a portion where the gain near black is high in the analog signal processing.

FIG. 4 shows a more detailed circuit configuration of the contour correcting device (FIG. 3) according to one embodiment of the present invention. In FIG. 4, the delay outputs R0H, G of the three primary color signals R, G, B are respectively applied to the HPFs 21a, 22a, 23a in the vertical direction.
0H, B0H, R1H, G1H, B1H, R2H,
G2H and B2H are supplied, and the high frequency component in the vertical direction is extracted. Since an increase in the number of 1H delay circuits causes an increase in circuit scale and cost, it is impossible to increase the order in the vertical direction. Therefore, HPF 21a, 2
2a and 23a are fixed coefficient filters of 3 taps (for example, the coefficients are -1/2, -1/2),
In addition, there is a zero in DC.

Vertical LPFs 21b, 22b, 23b
With respect to the delayed outputs R0 of the three primary color signals R, G, B respectively
H, G0H, B0H, R1H, G1H, B1H, R
2H, G2H, B2H are supplied, and the low frequency component in the vertical direction is extracted. As described above, since the order cannot be increased in the vertical direction, the LPFs 21b, 22
b and 23b are 3-tap fixed coefficient filters (for example, the coefficients are set to 1/4, 1/2, and 1/4), and a zero point exists at 1/2 of the sampling frequency in the vertical direction. To do. When the red and blue 1H delay circuits are not used, the channels without the 1H delay circuit are LPF 21b,
22b and 23b are not provided, the mixer 24a,
24b.

The output R of the HPF 21a is fed to the mixer 24a.
The output GVH of the VH and HPF 22a and the output BVH of the HPF 23a are supplied. Since the gain of the mixer 24a is 1,
The sum of the coefficients is 1, and the number of multipliers in the mixer 24a is two. Also, the mixing ratio is manually adjusted according to the user's preference and the characteristics of the image being captured,
Generally, the S / N ratio deteriorates in the order of G, R, and B. Therefore, usually, the mixing ratio of the green signal G is set to be large and the blue signal B is set to be small. Since the mixer 24a mixes the AC components, as shown in FIG. 5, instead of the configuration of FIG. 4, the absolute value conversion circuits 40, 41, 42, 43 and the comparators 44, 4 are used.
5 and switches 46, 47 can be used. However, the configuration of FIG. 5 cannot be used in the case of the mixer 24b in which the signals to be mixed also include the DC component.

In FIG. 5, the signal RVH is supplied to the absolute value conversion circuit 40, and the absolute value converted signal is supplied to the comparator 44. Further, the signal GVH is supplied to the absolute value conversion circuit 41, and the absolute valued signal is supplied to the comparator 44.
The comparator 44 compares the amplitudes of the absolute values of the signal RVH and the signal GVH, and the switch 46 selects the one having the larger absolute value. The signal selected by the switch 46 is supplied to the absolute value conversion circuit 42, and the absolute value converted signal is supplied to the comparator 45. Further, the signal BVH is supplied to the absolute value conversion circuit 43, and the absolute value converted signal is output to the comparator 4.
5 is supplied. The comparator 45 compares the amplitudes of the absolute values of the signal selected by the switch 46 and the signal BVH, and the switch 47 selects the one with the larger absolute value and outputs the signal with the larger absolute value. It

Returning to FIG. 4, the mixer 24b has an output RVL of the LPF 21b and an output G of the LPF 22b.
The output BVL of the VL and LPF 23b is supplied. Since the gain of the mixer 24b is 1, the sum of the coefficients is 1, and the number of multipliers in the mixer 24b is only two. The mixing ratio is variable according to the user's preference and the characteristics of the image being captured, but generally the S / N ratio becomes worse in the order of G, R, and B. Therefore, usually, the mixing ratio of the green signal G is set to be large and the blue signal B is set to be small.

The LPF 25 band-limits the high frequency component VH in the vertical direction in the horizontal direction. Here, the high frequency component in the horizontal direction is a high frequency component in the diagonal direction because the high frequency component VH in the vertical direction is input. Usually, it is often not so noticeable even if the resolution in the diagonal direction is reduced.
Bandwidth is limited to gain N ratio. But NTSC
In the case of the method, 3.58 MHz becomes the frequency of the subcarrier in the horizontal direction, and attenuation in the vicinity thereof is increased in order to prevent an increase in cross color due to edge enhancement. Also, PA
In the case of the L method, the frequency of 1/2 in the vertical direction is the frequency of the subcarrier, and there is also an offset of 25 Hz, so there is no effect in preventing cross color.
The S / N ratio and the diagonal resolution limit the band to an appropriate band.

An example of the LPF 25 has a sampling frequency of 36.
In the case of the NTSC system of MHz, the attenuation around 3.58 MHz is -40 dB, and the attenuation above 4 MHz is -20 dB.
belongs to. This LPF25 has 15 taps,
The coefficients are (8, 5, 5, 8, 8, 12, 11, 14, 1)
1, 12, 8, 8, 5, 5, 8) / 128.

The BPF 26 extracts a horizontal high frequency component from a vertical low frequency component to extract a vertical low frequency and horizontal high frequency signal VLHH1 which is a horizontal detail component. The center frequency (boost frequency) in the frequency characteristics of the BPF 26 is variable according to the user's preference and the characteristics of the image being captured. FIG. 6 shows an example of the coefficient when the sampling frequency is 36 MHz. As shown in FIG. 6, the BPF
26 is 21 taps and has a center frequency of 2 MHz to 9
It is possible to switch the coefficient set by manual adjustment within the range of MHz. The difference in center frequency causes the impulse response waveform width to become thicker / thinner on the time axis.

The BPF 35 extracts a horizontal high frequency component from the vertical low frequency component, and extracts a vertical low frequency and horizontal high frequency signal VLHH2 which is a horizontal detail component. Although it is desired to set the boost frequency sufficiently low so as not to be affected by noise as much as possible, if the set frequency of the BPF 26 is too high, there is too much difference and a feeling of strangeness occurs. Therefore, it is practical to set the BPF 35 to about 1/2 of the set frequency of the BPF 26. Also, if it is set too low, it will be far from the function of edge enhancement, so it is preferable not to set it below 2 MHz.

The coefficient of the BPF 35 is basically the BPF 26.
However, as shown in FIG. 7, normally, the amplitude of the input signal decreases as the frequency becomes higher due to the characteristics of the lens, the optical LPF, the CCD, the prefilter of the AD converter, and the like. In this way, since the frequency characteristics of the input signal are not flat, BP
It is necessary to change the gain with F26. Therefore, although the gain at the boost frequency is usually matched, it may be varied according to the characteristics of the image being captured.

The LPF 28 further limits the band in the horizontal direction for the low frequency component VL in the vertical direction. Since the band-limited signal VLHL is used for controlling the level dependent, it outputs a signal (DC component) VLHL limited to the band that is not extracted as a detail signal. Therefore, the band limitation is performed by the LPF 28 below the frequency boosted by the BPF 35. Sampling frequency 36MHz, cutoff frequency 1MHz, 2MH respectively
FIG. 8 shows an example of filter coefficients when z and 3 MHz are set. In FIG. 8, there are 21 taps at 1 MHz, 17 taps at 2 MHz, and 11 taps at 3 MHz.

The mixer 37 outputs horizontal detail signals VLHH1 and VLH from the BPF 26 and the BPF 35.
Mix H2. The boost frequency is varied by mixing these detail signals and changing the mixing ratio. The gain of the mixer 37 is 1, and the sum of the coefficients is 1. Therefore, only one multiplier in the mixer 37 is required. A hold circuit 38 is connected between the multiplier and the output of the level-dependent nonlinear processing circuit 36. FIG. 9 shows waveforms for explaining the hold circuit 38.
Shown in

In FIG. 9, FIG. 9A shows BPFs 26 and B.
An example (step waveform) of the waveform of the input signal VL input from the mixer 24b to the PF 35 and the LPF 28 is shown. For such an input waveform, the output signal VLHH1 from the BPF 26 becomes that shown in FIG. 9B, and BP
The output signal VLHH2 from F35 is as shown in FIG. 9C.

Further, the LPF 28 outputs a signal having the waveform shown in FIG. 9D, and the nonlinear circuit 36 outputs the mixing ratio signal LD1 (LD1 is the same as the above mixing ratio α) shown in FIG. 9E. Occur. In FIG. 9D, when the mixing ratio output LD1 is directly supplied as the coefficient (mixing ratio) in the mixer 37, the output VLHH having a discontinuous waveform as shown in FIG. 9F is generated. To prevent this,
A hold circuit 38 is provided to hold the mixing ratio LD1 at a constant value for one pulse of the detail signal. An example of the circuit configuration for realizing the hold circuit 38 is shown in FIG. 10A.

As shown in FIG. 10A, the hold circuit 38
Has a shift register including a plurality of registers 51a to 51g, a shift register including a plurality of registers 53a to 53f, and a shift register including a plurality of registers 54a to 54d. Each register has a delay amount of one pixel (one sample). In the shift register including the registers 51a to 51g, switch circuits 58a to 58g are provided on the input side of delay elements other than the delay element 51d. The mixing ratio LD1 (input signal) or the output signal of the preceding delay element is supplied to one input terminal of the switch circuits 58a to 58g, and the output signal of the delay element 51c is supplied to the other input terminal. There is. The switch circuits 58a to 58g are normally connected to the lower side and perform a shift operation.

The switch circuits 58a to 58g are controlled by the outputs of the AND gates 52a to 52g. The boost frequency is lower than that of the shift register including the registers 53a to 53f and the absolute value conversion circuit 55.
The signal VLHH2 from 5 is supplied. This signal VLH
H2 is parallelized by the shift register and supplied to one input of the exclusive OR gate. Register 5 as the other input of the exclusive OR gate
The output of 3c is commonly supplied. The output of the exclusive OR gate is inverted, and when the logical values of the two inputs are the same, a high level output is generated.

The absolute value of the signal VLHH2 from the absolute value conversion circuit 55 is supplied to the shift register consisting of the registers 54a to 54d, and the central pixel of the three consecutive pixels taken out by this shift register and the pixels before and after it. Each of the pixels is supplied to the subtractors 56a and 56b. Therefore, the subtracters 56a and 56b form the respective differences between the pixel at the center of the three consecutive pixels and the pixels before and after the pixel.

The output signals of the subtractors 56a and 56b are supplied to the comparators 57a and 57b and compared with the threshold values, respectively. The output signals of the comparators 57a and 57b are supplied to the AND gate 57c. Comparators 57a, 57b,
By the AND gate 57c, as shown in FIG. 10B,
If the two differences are both larger than a predetermined size (threshold value), it is detected as a maximum and the output of the AND gate 57c becomes high level. This threshold value is held in an internal register and supplied from an external control device to the comparators 57a and 57b through a control bus.

BP, which is the horizontal detail component of the thicker one
Output of F35 (vertical low band, horizontal high band signal VLHH2)
Is supplied to the absolute value conversion circuit 55, and is also supplied only to the codes to the shift registers (53a to 53f), and the output signal of the absolute value conversion circuit 55 is shifted to the shift registers (54a to 54f).
d). The shift registers (54a to 54d) are composed of the comparators 57a and 57b and the AND gate 57c.
When it is detected that the output of the register 54c has a maximum value, the sign of the signal VLHH2 at the maximum point is generated in the output of the shift register 53c, so the exclusive OR gates and AND gates 52a to 52a.
Depending on g, the outputs of the registers 53a to 53f, which are the signs of VLHH2 of each pixel around the maximum point, and the register 5
Compare the outputs of 3c.

When it is detected by the exclusive OR gates and AND gates 52a to 52g that the sign matches the maximum value of the signal VLHH2, the switch circuits 58a to 58g are connected to the upper side in FIG. 10A.
As a result, the value of the mixing ratio LD1 at the maximum point of the signal VLLH2 is set in the registers 51a, 51b, 51c, 51.
.., 51g, and the contents of these registers are all controlled to the same value.

FIG. 11 shows the operation of the hold circuit 38 described above. The signal VLHH2 shown in FIG. 11A is used.
FIG. 11B is a timing chart when the power is supplied. That is, the register 5 is shown at the top of FIG. 11B.
The time change of the contents of 1a to 51g is shown. Below that, the time changes of the outputs of the AND gates 52a to 52g are shown. Further, the time changes of the contents (signs) of the registers 53a to 53f are shown, and the time changes of the contents (absolute value) of the registers 54a to 54d are shown. Furthermore, the outputs of the subtracters 56a and 56b and the switch circuits 58a to 58g.
Output is shown.

In this way, the hold circuit 38 shown in FIG. 10A holds the value of the mixing ratio LD1 at the maximum point during the pulse width of the signal VLHH2, and as a result, the mixing ratio of the mixer 37 is also held. .

The width of the regions on both sides of the maximum point for holding the mixing ratio is within the range of about one impulse response of the BPF 35 for generating the signal VLHH2, and the portion where the polarity of the detail signal is the same as the maximum point is searched for, Try to consider it as the range of the waveform of the pulse. Further, the impulse response of the signal VLHH1 from the BPF 26 falls within the impulse response of the signal VLHH2. When the input is a thin impulse, the polarities of the response may be reversed between the BPF 26 and the BPF 35. However, the LPF 28 for the level dependent
There is no problem because the response of is almost 0.

When the vertical low-frequency signal VL from the mixer 24b has a step waveform as shown in FIG. 12, or when this signal VL is a thick pulse, the signal VLH
The waveform of the signal VLHH1 enters inside the waveform of H2. Further, when the output VL from the mixer 24b is a narrow pulse, the signal VLHH1 may not enter inside the waveform of the signal VLHH2, but since the output VLHL (DC component) is almost 0, the change of the mixing ratio also occurs. There is not much, and it does not matter.

The non-linear circuit 36 is a mixing ratio LD of the signals VLHH1 and VLHH2 which are horizontal detail signals.
It is a circuit that outputs 1, and by calculating from the horizontal low-frequency and vertical low-frequency signals VLHL that do not include details,
A mixing ratio LD1 is generated. As a general circuit, RA
It is composed of a lookup table using M. The non-linear circuit 36 is connected to a control bus, and a table is loaded from an external control device. As an example of the table, the one obtained by optimally normalizing the slope of the input / output characteristic of the gamma correction circuit as shown in FIG. 13A (FIG. 13B) can be used.

In FIG. 13A, assuming that the function of the gamma correction circuit is g, the output is the equation y = g (x). Then, if the derivative of g is set to g ′, LD1 = [g ′ (VLH
L) -1] / [g '(0) -1] can be used for normalization. FIG. 13B shows the characteristic of the mixing ratio LD1 thus normalized. However, it is limited to 0 ≦ LD1 ≦ 1. The output VLHH of the mixer 37 is the mixing ratio LD1
And (VLHH =
It is represented by (1-α) · VLHH1 + αVLHH2. Therefore, the larger the mixing ratio LD1 from the nonlinear circuit 36 is,
The ratio of the signal VLHH2 in the output is supposed to be large.

If the circuit scale of the RAM becomes too large, the non-linear circuit 36 uses a configuration using a comparator. In this case, as shown in FIG.
The nonlinear circuit 36 has a characteristic of a broken line. In this characteristic, there is discontinuity near the change point, so it is preferable to set the width of the transition region where the mixing ratio LD1 changes as wide as possible.

The non-linear circuit 29 is a circuit for determining the gain of the vertical high band and horizontal high band signals VHHL, and calculates the gain LD2 from the horizontal and vertical low band signals VLHL. As a general circuit, a look-up table configuration using a RAM is used. The non-linear circuit 29 is connected to the control bus, and the contents of the table are loaded from an external control device. In the analog circuit, only switching was performed at a predetermined level, but the characteristics when it is turned off are shown in FIG.
It is rounded as shown in. Input / output characteristics that mimic the change curve of the gain LD2 when this analog circuit is used,
It is made to write on the table.

Further, as shown in FIG. 16, the nonlinear circuit 2
9 can also be a comparator configured by a subtractor 61, a multiplier 62, and a clip circuit 63. The plus side input of the subtracter 61 is + IN, and the minus side input is −
When the coefficient is set to IN and the coefficient for the multiplier 62 to which the output of the subtractor 61 is supplied is sloped, the output OUT is as shown in FIG.
It is represented by what is shown in A.

In this embodiment, VLHL is supplied as the positive input of the subtractor 61, and the beginning XP of the transition region is supplied as its negative input. The output of the subtractor 61 is
(VLHL-XP), and the gain LD2 is 0 when the signal VLHL is XP or less. When a slope is applied to the multiplier 62 as a slope, the characteristic shown in FIG. 17B is obtained. Since the gain LD2 can take only a value of 0 to 1, the smaller the inclination, the wider the transition range in which the gain LD2 changes. If XP is set too large at the beginning of the transition region, the waveform of the detail signal will be distorted.

For example, as shown in FIG. 18A, an input signal (IE IN) and a detail signal DTL '(multiplier 27
Input), the signal VLHL in the case of the waveform of the signal VLHL
When the detail signal DTL 'below the level of the broken line is cut, the output waveform is distorted as shown in FIG. 18B. Therefore, it is preferable not to make XP too large at the beginning of the transition region.

The mixer 30 is a mixer of vertical detail components (vertical high band, horizontal low band signal) VHHL and horizontal detail components (vertical low band, horizontal high band signal) VLHH. The mixer 30 has a gain of 1 and the sum of the coefficients is 1. Therefore, the number of multipliers in the mixer 30 is only one. The detail signal DTL is taken out from the mixer 30 to the output terminal 31.

The mixers 24a, 24b, 3 shown in FIG.
At 0, the mixing ratio, the coefficient in the variable coefficient filter, etc. are held in a register (not shown) and set from an external control device through the control bus. The value of this coefficient depends on the user's preference and the characteristics of the image being captured.
It is variable. Further, the coefficient of the filter needs to be changed even if the frequency characteristic is not changed if the sampling frequency is changed due to replacement of the CCD or the like.

[0063]

As described above, the present invention improves the S / N ratio by shifting the frequency characteristic of the detail signal to a low frequency range near black, which has a high gain in the gamma correction circuit, and improves the S / N ratio. A sufficient amount of detail signal can be added. Further, since the high frequency gain of the analog circuit drops near black, where the gain is high, the operator can accept the digital circuit with a similar effect to that of the analog circuit.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a circuit configuration of a conventional contour correction device.

FIG. 2 is a block diagram showing a circuit configuration of an example of a video camera to which the present invention can be applied.

FIG. 3 is a block diagram showing a circuit configuration of a contour correction device according to an embodiment of the present invention.

FIG. 4 is a block diagram showing a more detailed circuit configuration of a contour correction device according to an embodiment of the present invention.

FIG. 5 is a block diagram showing an example of a mixer according to an embodiment of the present invention.

FIG. 6 is a schematic diagram for explaining a BPF in one embodiment of the present invention.

FIG. 7 is a schematic diagram for explaining a BPF according to an embodiment of the present invention.

FIG. 8 is a schematic diagram for explaining an LPF according to an embodiment of the present invention.

FIG. 9 is a waveform diagram for explaining a hold circuit according to an embodiment of the present invention.

FIG. 10 is a block diagram showing an example of a hold circuit in one embodiment of the present invention, and a waveform diagram for explaining detection of a maximum value.

FIG. 11 is a waveform chart and a timing chart for explaining an operation of an example of the hold circuit according to the embodiment of the present invention.

FIG. 12 is a waveform diagram for explaining a mixer according to an embodiment of the present invention.

FIG. 13 is a waveform diagram for explaining a non-linear circuit in one embodiment of the present invention.

FIG. 14 is a waveform diagram for explaining a non-linear circuit in one embodiment of the present invention.

FIG. 15 is a waveform diagram for explaining a non-linear circuit in one embodiment of the present invention.

FIG. 16 is a block diagram showing an example of a non-linear circuit according to an embodiment of the present invention.

FIG. 17 is a waveform diagram for explaining a non-linear circuit in one embodiment of the present invention.

FIG. 18 is a waveform diagram for explaining a non-linear circuit in one embodiment of the present invention.

[Explanation of symbols]

21a, 22a, 23a ... HPF in the vertical direction, 21
b, 22b, 23b ... Vertical LPF, 24a,
24b, 30, 37 ... Mixer, 25, 28 ... Horizontal LPF, 26, 35 ... Horizontal BPF,
27 ... Multiplier, 29, 36 ... Non-linear circuit, 31
... Output terminals

Claims (11)

[Claims] 1. An image pickup means for picking up an image of a subject and outputting a video signal, a signal level detection means for detecting a signal level of a DC component of the video signal, and a part of the video signal from the video signal. A detail signal generating means for generating a detail signal for correcting the contour of the image represented by the video signal by extracting a frequency component, and a signal of the DC component based on the output of the signal level detecting means. A control means for controlling the detail signal generating means and a mixing means for mixing the detail signal with the video signal so that the frequency component of the detail signal generated by the detail signal generating means becomes lower as the level is lower. Video camera characterized by having. 2. The video camera according to claim 1, wherein the detail signal generating means compares a first filter for extracting a relatively high first frequency component from the video signal with the video signal. A second filter for extracting a second frequency component having a relatively low frequency, an output of the first filter, and an output of the second filter with a mixing ratio determined based on a control signal given from the control means. A video camera comprising a frequency component mixing means for mixing. 3. The video camera according to claim 2, wherein the frequency component mixing means includes an extreme value detecting means for detecting a maximum point and a minimum point of the output of the second filter, and the extreme value detecting means. A video camera, comprising: a mixing ratio holding means for holding the mixing ratio of the frequency component mixing means so as not to change for a predetermined period when a maximum point or a minimum point is detected. 4. The video camera according to claim 1, further comprising a non-linear compression means for non-linearly compressing the signal level of the video signal, wherein the detail signal generation means is not compressed by the non-linear compression means. A video camera, wherein the detail signal is generated from a video signal. 5. The video camera according to claim 1, wherein the video signal is a digital video signal, and a detail signal of the digital signal is generated by digital signal processing. 6. A signal level detecting means for detecting a signal level of a direct current component of a video signal, and a part of frequency components of the video signal is extracted from the video signal to be represented by the video signal. Detail signal generating means for generating a detail signal for correcting the contour of the image, and a detail signal generated by the detail signal generating means based on the output of the signal level detecting means, as the signal level of the DC component is lower. A video signal contour correction device comprising: control means for controlling the detail signal generating means so that the frequency component of the video signal becomes low; and mixing means for mixing the detail signal with the video signal. 7. The contour correction device according to claim 6, wherein the detail signal generating means compares a first filter for extracting a relatively high first frequency component from the video signal with the video signal. A second filter for extracting a second frequency component having a relatively low frequency, an output of the first filter, and an output of the second filter at a mixing ratio determined based on a control signal provided from the control means. And a frequency component mixing means for controlling the contour correction device. 8. The contour correction device according to claim 7, wherein the frequency component mixing means includes an extreme value detecting means for detecting a maximum point and a minimum point of the output of the second filter, and the extreme value detecting means. A contour correction device comprising: a mixing ratio holding means for holding a mixing ratio of the frequency component mixing means so as not to change for a predetermined period when a maximum point or a minimum point is detected. 9. The contour correction apparatus according to claim 6, further comprising a non-linear compression means for non-linearly compressing the signal level of the video signal, wherein the detail signal generation means is compressed by the non-linear compression means. A contour correction device, wherein the detail signal is generated from a non-existing video signal. 10. The contour correction apparatus according to claim 6, wherein the video signal is a digital video signal, and a detail signal of the digital signal is generated by digital signal processing. 11. An image pickup means for picking up an image of a subject and outputting a video signal, a signal level detection means for detecting a signal level of a DC component of the video signal, and a detail signal generation for generating a detail signal from the video signal. Means for controlling the detail signal generating means so that the boost frequency of the detail signal generated by the detail signal generating means becomes lower as the signal level of the DC component is lower, based on the output of the signal level detecting means. And a mixing means for mixing the detail signal with the video signal.