JPS5832553B2 - color television camera - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a color television camera that obtains a plurality of color signals using a single image sensor.

A phase shift method is used to form a subject image on a photoelectric conversion surface via a color stripe filter, obtain a composite signal modulated by the stripe filter by scanning the beam, and obtain a color television signal by processing the signal. There are separation methods and frequency separation methods, and the present invention relates to the latter, and proposes a new method that solves the problems of several conventionally proposed methods.

Conventionally proposed frequency separation methods can be roughly divided into the following three methods.

A two-frequency separation method in which two color components are spatially modulated at two frequencies that are more than twice the frequency band of the color signal, and each modulated signal is separated through a bandpass filter and extracted by amplitude detection.

(2) Two color components are spatially modulated at the same frequency using a color stripe filter that has a different phase relationship for each scanning line in the vertical direction, and the modulated signal is delayed by one horizontal period and this modulated signal is 2 by calculating the signal
A single frequency separation method that separates the two color components and extracts them by amplitude detection.

(3) Two color components at two different frequencies, and two
Spatial modulation is performed using color stripe filters with different phases for each scanning line in the vertical direction so that the frequency difference between the two is within twice the frequency band of the color signal, and the two color signals are combined using the method (2) above. Two frequency phase separation method.

However, each method has the following problems.

That is, in the case of the two-frequency separation method (ke), the modulation degree of the image pickup tube worsens as the modulation frequency increases, and the noise energy increases approximately in proportion to the square of the frequency, resulting in poor S/N. (For example, in a 1-inch vidicon, when the modulation frequency is 4ME (z), the modulation degree is 73%, 5MHz
In the latter case, the S/N is 59% compared to the former.
(approximately 4 dB deterioration) Also, there are drawbacks such as deterioration of color shading due to the difference in modulation frequency.

In the case of the single frequency separation method (2), since the modulation frequency is the same, it is superior to the method (1) in terms of S/N, but since it uses vertical correlation, When capturing an image of an object that changes in direction (horizontal scanning direction and perpendicular direction), there is a drawback that color errors occur at the changing vertical edges of the object.

For example, if the subject changes from white to black in the vertical direction, a magenta color error will occur at the edges.

In addition, (2) Since signals are separated using a delay line in the horizontal period, color interference components with a frequency of 1/2 of the horizontal frequency are likely to occur.

The two-frequency phase separation method (3) has the same drawbacks as the method (2) described above, and the S/N ratio is better than the method (1) but inferior to the method (2).

Now, the cause of a color error occurring at a changing vertical edge portion of an object when color component signals are separated using vertical correlation will be explained using the case (2) of the single frequency separation method as an example.

FIG. 1 shows the configuration of the single frequency separation method described above.

Figure 2 shows a color stripe filter placed in front of the photoelectric conversion surface of the image pickup tube, which modulates the red component at an angular frequency ω with ON and cyan (C), and modulates the red component with white W and ? Jff), the blue component is modulated at the angular frequency ω, and the phases differ by 900 and 900, respectively, between successive scanning lines.

In Fig. 1, 1 is an image pickup tube with a built-in live filter shown in Fig. 2, and the modulated signal is transmitted to the preamplifier 2.
through low-pass filters 3, 4 and band increase FiJ
(BPA) 5.

Here, the low-pass filter 3 is used to attenuate the modulated frequency component to obtain the luminance signal Y(t), and the low-pass filter 4 is used to attenuate the modulated frequency component to obtain the luminance signal Y(t). It is for taking out.

The output signal of the band amplifier 5 is operated by a delay line (IHDL) 6.900 phase shifter 7 that delays one horizontal period, a subtraction circuit 8, and an addition circuit 9, and is passed through detectors 10 and 11 to produce a color component signal R t), B(t) are extracted and applied to the encoder 12 together with the aforementioned Y(t)) yl(t) to obtain an NTSC signal.

Now, the output of the band amplifier 5 in one horizontal scan is represented by , and the output in the next scan is represented by .

Here, θ is a constant representing the phase.

The signal obtained by passing the signal shown in equation (2) through the 900 phase shifter 7 is as follows.

At this time, the output of the subtracting circuit 8 is (1) and (3): + from this formula, 2 Eqt) cos (ω is 10 θ), while the output of the adding circuit 9 is 2 B(t) cos (ω is 1 θ).
).

Detectors 10 and 11 separate color component signals R(t) and B(t).

Here, we will consider a case where an image with no correlation in the vertical direction is captured, for example, when the subject changes from white to black.

The output signal of the image pickup tube 1 produced by scanning the white part is as follows, and the output produced by scanning the black part becomes zero.

At this time, the outputs of the subtraction circuit 8 and addition circuit 9 are both B(t
) cos (ω one θ) + R(t) cos (ωt+θ
).

If the blue component, red component, and output level of the image pickup tube in the white part are a, then R(t) = B(t) = a, detector 1
The maximum output level at 0°110 is 2a.

At this time, since the output of the low-pass filter 4 is zero, this portion becomes magenta, resulting in a color error.

Next, consider the case where the subject changes from red to black in the vertical direction.

This is the output of the image pickup tube 1 generated by scanning the red part,
The output produced by scanning the black part will be zero.

At this time, the outputs of the subtraction circuit 8 and addition circuit 9 are both R(t
) cos (ωt+θ).

When R(t)-a, the output level of the detectors 10 and 11 +ia.

At this time, since the output of the low pulse filter 4 is zero, this portion becomes magenta, resulting in a color error.

Generally, color television cameras are designed to produce the best images at a color temperature of around 3000'K, and when photographing objects with other color temperatures, a color temperature conversion filter is installed in front of the image pickup tube.

Therefore, in the output of the image pickup tube, the blue component is generally smaller than the red component.

Now, assume that the blue component in the output of the image pickup tube 1 is 1/3 of the red component.

When imaging an achromatic color, the gain of the sound component is 3 times the gain of the red component in order to match the output levels of the detectors 10 and 11.
Need to double it.

Therefore, the color error in the vertical edge portion of the image pattern has a large blue component.

Now, consider a case where the subject changes from white to black in the vertical direction.

When first scanning the white part and then scanning the black part, the outputs of the subtraction circuit 8 and addition circuit 9 are both B(t)cos
(ω is 10) 1R(t) cos (ωt+θ).

When θ-〇, drop *(B(t)+R(t)) c
os ωt.

The pressure of the detector 11 is (B(t)+R(t)),
Since the gain of the detector 1o is three times that of the detector 10, the output is 3 (B(t)+R,(t)).

The output level of the blue component and red component of the image pickup tube in the white part is R(
t)=a, B(t)-1, the output of the detector 11 for the red component is 1-a, and the output of the detector 100 for the blue component is 4.
It becomes a.

This results in a color error with a strong blue color.

Similarly, when the color changes from red to black in the vertical direction, the output of the detector 10 is 3 R(t), and the output of the detector 11 is R(t).
), resulting in a strong blue color error.

Since the two-frequency phase separation method (3) also uses vertical correlation, a similar vertical color error occurs.

As a signal processing method to reduce such color errors, we have already proposed in another application a signal separation method that uses the horizontal correlation of the output signals instead of the vertical correlation of the output signals. .

FIG. 3 is a block diagram showing an example of the configuration of a signal processing circuit of a color television camera according to the invention, and blocks having the same configuration as in FIG. 1 are designated by the same numbers.

The filter configuration is the same as in FIG. 2, but the repetition frequency is different.

The angular frequency of the red component is ω1, and the angular frequency of the blue component is ω2.

In the figure, 13 and 14 are delay circuits that delay the blue component and red component by 1/2 of the repetition period, respectively.
6 is an addition circuit or a subtraction circuit, and 17 and 1B are a band amplifier and a wave detector.

To explain the operation of this embodiment, the output S of the band amplifier 5
Consider the case where (t) is given by the following equation.

At this time, The output of the delay circuit 13 is the delay time Then, The output of the delay circuit 14 has a delay time τ1 becomes.

Here, when n=2m (m: positive integer), equations (6) and (7) are expressed by the following equations (8) and (9).

Therefore, the outputs of adders 15 and 16 are obtained from equation (5) + equation (8) and equation (5) + equation (9), respectively, and are (1,0
) formula and old) * main formula.

Therefore, it can be seen that the red and blue component signals are extracted by passing through the band amplifier/detector 17 and 18, respectively.

In this embodiment E, when rn=1, that is, the delay circuit 13 delays the blue component by 1/2 of the repetition period, and the delay circuit 14 delays the red component by 1/2 of the repetition period. is expressed in the form of equation (12) and equation (■3).

Figure 4a shows the difference ω1−ω in the modulation angular frequency of the red and blue components.
Bandwidth amplifier 17 when 2 is within the band of the color signal
The figure shows the frequency characteristics of the output B(t) of the band amplifier 18, and it can be seen that the two color component signals can be separated.

As is clear from general formulas (10) and (13), in the case of this embodiment, the delay time of delay circuits 13 and 14 is set not to 1/2 of the repetition period of each color component but to an odd multiple thereof. Even if the color component signal separation power Z'ij]
In addition, if subtracter circuits are used in place of adder circuits 15 and 16, each delay time can be multiplied by an even number, as can be seen from the same analysis with n=2m in equation (7). Can be separated even if set.

Comparing such a signal processing method with the two-frequency separation method described above, for example, if the modulation frequencies of the two-frequency separation are 4MHz and 5MHz, in this method, the 4MHz
z and 4.5 MHz, this method has a higher S/N ratio of about 1 to 2 dB, and is also advantageous in terms of color shading because the modulation frequencies are close.

On the other hand, when compared to the single frequency separation method, it is somewhat disadvantageous in terms of S/N and color shading, but it is said that color errors do not occur at vertical edges even when capturing images of subjects with no correlation in the vertical direction. There are big advantages.

This processing method uses horizontal correlation to separate colors, so if a subject is imaged discontinuously in the horizontal direction, color errors will occur at the horizontal edges, but the color signal band Since the area is narrow, the color error at the vertical edge of the conventional single frequency separation method does not appear as noticeable.

Horizontal color error II' caused by this method! , for example, can be reduced by the following method.

Now, if we consider red and blue vertical stripes as shown in Figure 5a as objects, the corresponding output signals will be as shown in Figure 5A and Figure 5C, and the 100 and 2
The portion indicated by 00 appears as a color error.

Assuming that the red to blue image pickup tube output ratio is 11, experiments have shown that the amplitude ratio of each actual signal component to the leakage signal component is about 20 dB to 30 dB.

Generally, the red component is larger than the blue component in the image pickup tube output, so the influence of the leakage signal component on the red component can be almost ignored.

However, since leakage to the blue component is a problem, a correction circuit is required.

An example of the correction circuit is shown in FIG.

In the figure, 19 is an IH delay line that delays the output of the adder circuit 16 for one horizontal period, and 20 is an IH delay line that delays the output of the adder circuit 16 by 900.
21 is a subtraction circuit.

The block for the correction circuit is designated 300.

Now, the output of the band amplifier 5 is R(t) cosω1t + B(t) cosω2t
Assuming that the signal of the output of the adder circuit 16 generated when the object shown in FIG. 5a is imaged and scanned is:
......-(14).

Here, J R(t) is the leakage component due to crosstalk,
B'(t) represents a constant term in equation (13).

Since the phases of the red and blue components differ by 900° and 190° between successive scanning lines, the output signal of the adder circuit 16 resulting from the next scan will be A, and the output of the 900 phase shifter 20 will be A. R(t) cos (c)1 t B(t) cosω2t . . . t16).

Therefore, the output of the subtraction circuit 21 is 2 B(t) cos
ω2t is obtained and crosstalk is removed.

Since crosstalk is removed using vertical correlation after modulation signals of other color components are removed, color errors at vertical edges are greatly reduced compared to the above.

Now, let us consider as an example the case where an image of a subject that changes from white to black in the vertical direction is captured.

When the previous scan was white and the next scan is a black part, the output of the adder 15 is zero, and the output of the subtraction circuit 21 is B'(t
)cosω2t.

Assuming that the blue component of the image pickup tube output is 1/3 of the red component, and the gain of the blue component is three times the gain of the red component, the output of the detector 18 is 3B(t), which is R(t)-a, B( t)-1, the output of the detector 18 will be a.

Compared to the color separation using only vertical correlation, the blue component is improved from 4a to a, and the red component is improved from -a to 0.

Next, consider the case where an image that changes from red to black in the vertical direction is captured.

The output of the adder 15 is zero when the previous scan was red and the next black part is scanned, and the output of the adder 16 is zero when the previous red part is scanned. Next, the output of the subtraction circuit 21 becomes 0 when scanning the black part.

In this case, there will be no color error at the vertical edges.

At this time, crosstalk appears, but since it is only a part of the image, color errors are greatly reduced.

This comparison is shown in Figure 7a. I will explain in b and b.

When an object as shown in FIG. 7A is imaged, the reproduced image is obtained by color separation using vertical correlation, and the portion indicated by 400 is the color error at the vertical edge portion.

In the reproduced image shown in FIG. 7C, the portion numbered 500 is a color error due to crosstalk, and this color error has been greatly reduced.

When the red and blue components of the image pickup tube output are approximately equal in ratio, if they are separated using the method shown in Figure 3, the crosstalk will be -20~
There is -30dB.

A method to further reduce this will be explained.

FIG. 8 shows an example of its configuration.

Blocks 1-6 and 8-12 move in the same way as blocks 1-6 and 8-12 of the same symbol in FIG.

The configuration of the stripe filter is as shown in FIG. 2, and the red component is modulated at an angular frequency ω1, and the blue component is modulated at an angular frequency ω2.

22 and 23 are phase shifters, and the phase shifter 22 shifts the phase by 90° at the angular frequency ω2, and the phase shifter 23 shifts the phase at the angular frequency ω□.
900 translocations.

Reference numeral 24 designates a circuit that removes only the angular frequency ω2, which may be configured as blocks 13 and 15 in FIG. 6 and may be removed by dividing by pitch correlation, or may be removed by an l-wrap circuit.

Similarly, 25 is a circuit that removes only the angular frequency ω1.

Now, the output of the band tank 11'' unit 5 is R(t) cosω1 t + B(t) cosω2
Let it be t.

In the next scan, the phase shift is 900° and -90°, respectively, so the output of the band amplifier 5 is R(t) cos (ωt
t+90) +B(t) cos (ω, t-90)
becomes.

The output of the phase shifter 22 is R(t) cos (ω1t+90+θ'1) + B(t) cos ω
2 t, and the output of the phase shifter 23 is R(t) cos ω1 t + B(t) cos (ω
2 t90-1-〇12).

The output of the subtractor 8 is R(t) (cosω1 t −cos (ω1t+90+θ11)
), and the output of the adder 9 is B(t) (cos ω2t + cos(ω2t 90 + θ'2)).

Here, θ'1 is the angular frequency ω1 when passing through the phase shifter 22.
When the phase shifter θ passes through the phase shifter 23, a phase shift of the angular frequency ω2 occurs.

For simplicity, if θ'1 - θ'2 - 900, the subtracter 8
The output of is 2 R(t) cos ω1t, and the output of the adder 9 is 2 B(t) cos ω2t.

The signal 2 R(t) cosω1t passes through a circuit 24 for removing the angular frequency ω2 and a detector 11 to obtain a red component, and similarly obtains a blue component.

Next, the color error of the vertical edge of the object will be explained.

When capturing an image of a subject that changes from white to black in the vertical direction,
The output of the band amplifier 5 when scanning the white part is R(t) cosω1 t + B(t) cosω2t
Therefore, the output of the band amplifier 5 that scans the black portion becomes zero.

Therefore, when a white part was scanned in the previous scan and a black part is currently scanned, the outputs of the subtracter 8 and adder 9 are R(t) cosω1t + B(t) cosω2t, respectively.
becomes.

The output of the circuit 24 that removes the angular frequency ω2 is R(t) c
The output of the circuit 25 that removes osω1t and angular frequency ω1 becomes B(t) cosω2t.

The outputs of the detectors 11 and 10 are R(t) and B(t). The output level of the blue component and red component of the image pickup tube in the white part is a
shall be.

R(t) - B(t) = a, and the detector is 10°11
The 0 output level is a.

The vertical edge color error of the present invention is improved by 6 dB compared to the case where separation is performed only by vertical correlation.

Next, when imaging a subject that changes from red to black in the vertical direction, the output of the band amplifier 5 when scanning the red part is R(t)cosω,t, and when scanning the black part Therefore, when the red part was scanned in the previous scan and the black part is currently being scanned, the outputs of the subtractor 8 and adder 9 are R. (t)
cosω1t.

At this time, the output of the wave detector 11 becomes R(t), and the output of the wave detector 10 becomes zero.

Compared to the case where separation is performed only by vertical correlation, the red component of the color error of the vertical edge of the present invention is not improved, but the blue component is improved from a to 0.

In this case, since a red image is captured and a red color error occurs at the vertical edge, the reproduced image does not look unsightly, and the improvement effect when viewed visually is significant.

Although the example of the improvement has been described when the ratio of the blue component and the red component of the image pickup tube is approximately equal, the present configuration can be applied even when the ratio of the blue component and the red component is different.

For example, consider a case where the red component is three times the blue component.

When capturing an image of a subject that changes from white to black in the vertical direction, the image pickup tube output for the white part is R(t) - B(t)
-a, the red component of the color error of the vertical etched portion is a, and the blue component is a.

Compared to the color error when color separation is performed only by vertical correlation, the red component is improved to 1-a-+a, and the blue component is improved to 4a-+a.

When comparing the color error of the vertical etch of an object that changes from red to black with the color separation based on vertical correlation only, the red component a-+a
, blue component 3a→0 (R(t)=O), which is greatly improved.

Furthermore, in the configuration of FIG. 6, the delay circuit 13 and the adder 15
may be replaced with a circuit 24 that removes the angular frequency ω2, and the delay circuit 14 and adder 16 may be replaced with a circuit that removes the angular frequency ω1.

According to the above-described method for improving color errors in vertical etching, errors in hue and color saturation can be reduced, but the problem of hue change remains unsolved.

In the present invention, the color of the vertical etch is hue Q compared to the color of the previous scan.
A color television camera is proposed that allows correction of color errors that do not change, but only change color saturation.

FIG. 9 shows the configuration of an embodiment of a color television camera according to the present invention.

In this figure, blocks that are the same as those shown in FIG. 6 are designated by the same numbers.

A block 600 indicated by a broken line is a portion added to reduce the color error of the vertical etch of the object.

26 is a delay circuit that delays the output of the low-pass filter 4 by one horizontal period, and the output of this output and the output signal of the low-pass filter 4 are added by equal amounts in an adder 27, and the luminance component for color demodulation Y ( t).

A detailed configuration of block 600 is shown in FIG.

A modulator 28 amplitude-modulates the output of the low-pass filter 4, and is supplied with a signal for modulation from an oscillator 31.

Note that 3 and 5 generated in the encoder 12 instead of the oscillator 310
It is also possible to add an 8 MHz color carrier to the modulator.

29 is an ultrasonic delay line for one horizontal period; 30 is an amplification detector that amplifies and detects the output of the delay line 29; 32 is a modulator 2;
8. A delay circuit having the same delay time as the signal delay time caused by the amplified wave detector 30.

Adder 27 adds the outputs of delay circuit 32 and amplification detector 30.

With this configuration, the luminance signal that is not modulated by the stripe filter can be delayed by one horizontal period using an inexpensive ultrasonic delay line.

In this way, the luminance component y(t) for color demodulation is obtained from the signal obtained by delaying the luminance signal that is not modulated by the stripe filter by one horizontal period and the original signal, and the red component and blue component are determined using vertical correlation. However, since the crosstalk between the red and blue components in the vertical etched area of the subject is removed, the brightness for color demodulation obtained when scanning the area in front of the vertical etched area is component Yi(t), red component R1(t
), the ratio of the blue component B1(t), the luminance component Y2(t) for color demodulation, and the red component R2(t) when scanning the next part.
) The ratio of the blue component B 2 (t) is equal, and (Yt(t)
: Bx(t)-Yz(t) : R2(t) :
B2(t) ) NTS using signals of different magnitudes
Since it is converted to a C signal, the color change in the vertical etched area does not change the hue, but only the color saturation.
Color errors at vertical edges are less noticeable.

Also, the vertical edges of an achromatic subject are not colored.

This will be explained specifically.

Consider the case where an achromatic image as shown in FIG. 11A is captured that changes from black to white and from white to black in the vertical direction.

When color separation is performed using the apparatus shown in FIG. 8, the luminance signal for color demodulation y(t), the red component signal R(t), and the blue component signal B(t)
The output waveforms (manual input signals of the encoder) are 口, Ａ, ２.

However, in the embodiment shown in FIG. 9, the color demodulation luminance signal y(
t) is obtained by adding the signal delayed by one horizontal period and the original signal, so the respective signals Y(t) and R(t)
.

B(t) becomes E, H, G in Figure 11.

As can be seen from this waveform, in the configuration of FIG. 8, the vertical edge that changes from black to white becomes green, and the vertical edge that changes from white to black becomes magenta, but in the embodiment of FIG. 9, each vertical edge Achromatic, meaning it has no color.

When an achromatic object is imaged in this way, the vertical edges are not colored at all.

Next, let us consider the case where an image of a subject whose color changes from black to yellow and from yellow to black as shown in FIG. 12A is taken.

Luminance y for color demodulation when color separation is performed using the device shown in Figure 8 (
t), the red component signal R(t) and the blue component signal B(t) are the 12th
The waveform shown in Fig. 2 shows the waveform.

Even in the part where there is no vertical correlation, there is no crosstalk between the red component signal R(t) and the blue component signal B(t), so B(t)=O
becomes.

Similarly, the waveforms of the signals Y(t), R(t), and B(t) when color separation is performed in the embodiment shown in FIG. 9 are shown in FIGS.

As you will see, in the device shown in Figure 8, the vertical edge part where the color changes from black to yellow becomes greenish-yellow.
The vertical etch that changes from yellow to black becomes red.

However, in embodiments of the present invention, the vertical edges are also yellow, only varying in saturation.

In this embodiment, as shown in FIG. 2, stripe filters are used in which adjacent horizontal scanning lines have phases different by -900 and 90 degrees, but the present invention can be similarly applied to cases with other phases.

As described above, the color television camera according to the present invention performs color separation using vertical correlation, but the imaging signal obtained by modulating the input light with a stripe filter and the signal delayed by one horizontal period and separates the signal modulated by the 10th hull component through means for calculating and removing the second frequency component, and calculating the imaging signal and a signal obtained by delaying this by one horizontal period The signal modulated by the 20th envelope component is separated through a means for removing the frequency component of
A luminance signal for color demodulation is obtained by adding the signal delayed by a horizontal period, and the signal modulated by the first color component, the signal modulated by the 20th component, and the luminance signal for color demodulation are added. Since the vertical edges of an achromatic image are not colored and the vertical edges of a colorful image do not change in hue, only the degree of saturation differs. , the image quality has improved significantly.

[Brief explanation of drawings]

Figure 1 is a block diagram of a conventional color television camera, Figure 2 is a configuration diagram of its stripe filter, Figure 3 is a block diagram of another color television camera, and Figure 4 is the frequency characteristics of its color signal components. Figure 5 is a diagram showing crosstalk, Figure 6 is a block diagram of another color television camera, Figure 1 is an explanatory diagram showing color error, and Figure 8 is a block diagram of yet another color television camera. 9 is a block diagram of a color television camera according to an embodiment of the present invention, FIG. 10 is a block diagram showing the detailed configuration of its main parts, and FIGS. 11 and 12 respectively explain color errors. This is a diagram for 1... Image pickup tube, 6... Delay line, 8...
...Subtraction circuit, 9...Addition circuit, 24,2
5... Specific angular frequency signal removal circuit, 26...
...Delay circuit.

Claims (1)

[Claims] 1 The first color component has a first phase and a first repetition frequency in the horizontal scanning direction, and the second color component has a second phase in the horizontal scanning direction that is different from the first phase. The first and second color components of the input light of the subject image are modulated by a stripe filter configured to have a phase of means for photoelectrically converting the input light of the subject image obtained by photoelectric conversion to obtain an imaging signal; and means for calculating the imaging signal and a signal delayed by one horizontal period and removing the second frequency component. a means for separating a signal modulated by the first color component through a means for calculating the image signal and a signal obtained by delaying the image signal by one horizontal period;
means for separating a signal modulated by a second color component through means for removing a frequency component of the second color component; means for obtaining a luminance signal for color demodulation by adding the signals, and a means for obtaining a luminance signal for color demodulation using the signal modulated by the first color component, the signal modulated by the 20th envelope component, and the luminance signal for color tone. and means for obtaining a color imaging output of an image.