JP2710317B2 - Color imaging device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial applications]   The present invention relates to a logarithmic color imaging device, and more particularly to a dynamic color imaging device.
It has means such as expansion of the work range. [Conventional technology]   Generally, the dynamic range of the image sensor is about 50 dB
Is said to be. However, the brightness of general subjects
The difference, or dynamic range, is larger
There are many, sometimes more than 100dB. Hand to solve this point
As described in JP-A-60-52171
Technical means, but only for black and white images
And not for color images.
No. [Problems to be solved by the invention]   JP-A-60-52171 (Japanese Patent Application No. 58-160916)
The dynamic range improvement means listed is for black and white images
And for color images
is not. When targeting color images, color images
It is necessary to perform special processing according to the unique characteristics of the image
However, there is no concrete means at present. For this reason,
When performing color imaging, there are various
It was difficult to obtain color images.   Therefore, the present invention uses a conventional image sensor (for example, 50 dB).
Color dynamic image
Significantly improve color image quality.
It is an object of the present invention to provide a color imaging device that can perform the operation. [Means to solve the problem]   The present invention solves the above problems and achieves the object.
The color that captures the color image for the observer to observe
-In imaging devices,   A color video signal that inputs the subject as a color video signal
Signal input means and the brightness from the input color video signal
A luminance signal extracting means for extracting a signal component to be represented;
The signal component representing the brightness extracted by the signal extracting means is
A non-linear compression means for performing non-linear compression, and
The signal component representing the brightness of the color video signal is nonlinearly compressed
Set by the non-linear compression means
Saturation and compression of color video signal at input using compression degree
A new color saturation has been added so that the later
Color signal synthesizing means for synthesizing color signals.
It is characterized by the following. [Action]   The following actions are taken by taking such measures.
The color space represented by the RGB three primary color vectors is
By the matrix conversion, as shown in FIG.
And a hue plane M (R-Y and
(Displayed in the rectangular coordinates of BY)
Can be. In FIG. 1, θ indicates the hue, γ indicates the saturation,
CB indicates a color vector. By the way, the color image
To extend the dynamic range, only the luminance signal Y
It is sufficient to compress, but the visual saturation γ is
Standardized (the brightness is doubled even if the absolute value of saturation is the same)
, The saturation is halved).
The reduction must be multiplied by the saturation γ. That is, the luminance Y
For logarithmic compression, multiply logY / Y by RY and BY
Good. By taking the above measures,
The dynamic range without affecting color saturation and hue
It becomes possible. [Example]   FIGS. 2 and 3 show the overall structure of the first embodiment of the present invention.
It is a block diagram showing composition. As shown in FIG.
The light image passing through the lens 1 and the half mirror 2 is
It is incident on a and 3b. The image pickup devices 3a and 3b are driven by the drive circuit 4.
To convert the light image into an electrical signal. This signal is RGB
The signals are sent to the separation circuits 5 and 6, and are separated into two sets of RGB signals.   Now, the dynamic range of the imaging devices 3a and 3b is set to 50 dB.
Then, [the amount of transmitted light] / [the amount of reflected light] of the half mirror 2 is 50
When set to dB, the imaging characteristics of the imaging elements 3a and 3b are as shown in FIG.
Become like The input coordinates in Fig. 4 are on a logarithmic scale.
You. In FIG. 4, a is an imaging characteristic of the image sensor 3a.
B is the imaging characteristic of the image sensor 3b. Above image sensor 3
When the outputs of a and 3b are simply added, as shown in Fig. 5,
A logarithmic imaging characteristic of 100 dB expressed in a similar manner is obtained. Fifth
The broken line in the figure indicates an ideal logarithmic characteristic. Of the above broken line
As close as possible to the ideal logarithmic characteristic indicated by the broken line
Logarithmic amplifiers 7 to 12 as shown in FIG.
Have been.   FIG. 6 is a diagram showing the internal configuration of logarithmic amplifiers 7 to 12, and FIG.
The range of the broken line (50dB) in Fig. 5 is further logarithmically
This is an example in which the characteristics are approximated. 13 to 15 in FIG.
Each amplifier has an amplification of 12.5 dB. Also 16 ~
19 is a window that performs a limiter with an amplitude equivalent to 12.5 dB
C. 20 is an adder.   FIG. 7 is a diagram showing the operation of the logarithmic amplifier shown in FIG.
21 is the output of the window circuit 16, 22 is the output of the window circuit 17.
Output, 23 is the output of the window circuit 18, 24 is the window number
The output of the road 19 is shown. These four outputs are
The addition by the adder 20 in FIG.
Ideal pair with 50dB dynamic range
It can approach numerical characteristics.   In the example of FIG. 6, a case where approximation is performed in four stages is shown.
However, it can be said that a better approximation can be obtained by increasing the number of stages.
Needless to say. Actually used for window circuits 16-17
Can be implemented with a small number of stages using the logarithmic characteristics of semiconductor devices
Noh.   As can be seen from the above description, two sets of RGB outputs
Ideal logarithmic characteristics by passing through logarithmic amplifiers 7-12
As shown in FIG. 2, this is added to the adders 26, 27, 2 for each RGB.
By adding each at 8, more accurate 100dB
Obtain signals logR, logG, logB with dynamic range
Can be.   By the way, regarding color images,
The numbers logR, logG, logB are converted to logY, logY (RY) / Y, logY (B
−Y) must be converted to the form / Y. Then the means
Will be described.   The basic principle is that the inverse transformation is performed by an antilogarithmic amplifier.
After extracting color signals R, G, B, linear matrix conversion circuit
To Y, R−Y, BY, and then logY, logY
(RY) / Y, logY (BY) / Y
You. However, with a logarithmic amplifier, a dynamic range of 100 dB
Signal processing after converting to R, G, B signal with edge
Is actually the dynamic range of the electrical circuit itself
(S / N) is not possible. However, the relative ratio of R, G, B
According to the visual characteristics, the rate is not required to be more than 40 dB. One
In other words, even at 40 dB or more, differences in luminance, saturation, and hue are
I do not feel it. Using this visual characteristic
Floating antilogarithmic conversion circuit for signals logR, logG, logB
It can be realized by the point method. The following
Specific means will be described.   First, the signals logR, logG, logB are converted to amplifiers 29, 30,
At 31, amplification is doubled as shown at 32 in FIG. And
The real-time average of these outputs is always the window in Figure 3.
Half of the limiter value W of the circuits 36, 37, 38, that is, W / 2
Subtractors 33, 34, 35 perform subtraction dynamically as
The characteristics are as shown in FIG. 8 to 50 are 50 dB.
Indicates the window width. The value logM to be reduced in the above subtraction
Is logM = {(logR + logG + logB) / 3}-W / 2 It is represented by the following formula. The above value logM is the average shown in FIG.
It is created by the value operation circuit 40. Window circuit 36,3
The limiter value W of 7,38 is 50 dB of the output of the amplifiers 29,30,31.
Of the dynamic range. That is, the signal logR, l
A signal in the range of ± 25 dB around the average value of ogG and logB
The signals are output from the dough circuits 36, 37, and 38. Here
The window width was selected to be 50 dB, because the window width was 10 dB.
This is because 50 dB was set in consideration of the margin of dB. 50dB die
Antilogarithmic amplifier with good S / N if in the dynamic range
(Exponential amplifier) can be composed of a matrix after
Conversion processing is also facilitated.   FIG. 9 is a diagram showing the internal configuration of the antilog amplifiers 41, 42 and 43.
is there. In FIG. 9, 61 to 63 are subtractors, and 64 to 67 are
Eggplant gain amplifier, 68-71 clips below O (V)
A clip circuit 72 is an adder.   Each output of the window circuits 36, 37 and 38 shown in FIG.
Antilog amplifiers 41, 42, 43 with a dynamic range of dB
Input, where it was converted to a linear value for the input.
That is, it is input to the matrix conversion circuit 44. Matrix conversion
Expressing the output of circuit 44 as an equation: Where log^-1Is in the range of 50dB.   I mean   * M is the aforementioned (R, G, B)^1/3 Becomes The output of this matrix conversion circuit 44 is again 50 dB
Logarithmic amplifiers 45, 46, 47
Thereafter, the average value logM is added by adders 48, 49, and 50.
It is. This gives a dynamic range of 100 dB
ogY, log (RY), log (BY) can be synthesized.
Wear. After this, logY is adjusted by the adder 51 for gain adjustment and
A multiplier 52 for which automatic gain control is performed and a coefficient S is given.
Thus, the dynamic range is adjusted. this
The point is disclosed in the above-mentioned JP-A-60-52171.
The technical means used may be used.   On the other hand, log (RY) and log (BY) are subtractors 53,5
The logY, the output of logarithmic amplifier 45, is reduced by 4 and l
og {(RY) / Y}, log {(BY) / Y}
It is. And antilog with 100dB dynamic range
Amplifiers 55 and 56 form (RY) / Y, (BY) / Y
After the conversion, the output of the multiplier 52 is output by the multipliers 57 and 58.
Multiplied by logY, logY / Y ・ (RY) logY / Y ・ (BY) Is converted to the form After this, the inverse matrix circuit 59
Convert to R ', G', B 'signal or NTSC signal
Thus, it can be handled as a normal color TV signal.   According to the first embodiment, the dynamic range is narrow.
Although it uses an image sensor, the color image die
Greatly improved natural range, extremely good color
An image can be captured.   The second embodiment of the present invention will be described below with reference to FIG.
I do. Note that the same circuits as those in the first embodiment have the same numbers.
wear.   Output of amplifiers 29, 30, 31 shown in FIG. 3 of the first embodiment
Is input to the gamma correction circuits 72, 73, and 74 in FIG.
Positive is given. In other words, the outputs of amplifiers 29, 30, 31 are logR, log
Since logarithmic compression is applied to G, logB, R, G,
Applying γ correction to B means this as shown in the following equation.
It suffices to multiply these signals by γ. logR^γ= Γ · logR logG^γ= Γ · logG logB^γ= Γ · logB   By the way, since the γ correction circuit usually has γ <1, the tenth
Use resistors R1 and R2 as shown at 72, 73 and 74 in the figure.
It can be easily configured with a subtractor by resistance division
You. However, as shown in Fig. 12, the maximum amplitude value after gamma correction
(Saturation value) to make it equal to that before correction.
Reference voltage V_RIs set equal to the maximum amplitude value (saturation value) before correction
Have been. Next, the signal logR that has undergone gamma correction^γ, logG^γ, lo
gB^γIs input to an average calculation circuit 40 having three inputs.
And subtractors 33, 34, 35 and adders 75, 76, respectively.
Entered in 77. Average value calculation circuit 40 and subtractors 33, 34, 35
Window circuits 36, 37, 38, antilog amplifiers 41, 42, 43 and mat
The operations of the Rix circuit 78, the logarithmic amplifier 45, and the adder 48 are as described above.
This is the same as the first embodiment. The matrix circuit 78
It is for making only. LogY which is the output of adder 48
(It has a dynamic range of 100dB.)
9 for gain adjustment or automatic gain adjustment.
Dynamic range adjustment or automatic die
The dynamic range adjustment is performed. Applied to adder 79
The gain adjustment voltage is log b and the dyna
Assuming that the mic range adjustment voltage is a, the output of the multiplier 80 is
a log bY. The changeover switches 81 and 82 adjust the gain and
Switching between automatic control and manual control of dynamic range adjustment
It is for choosing.   Next, the source of automatic gain adjustment and automatic dynamic range adjustment
Will be described. Automatic gain adjustment is one screen of luminance signal
The average value of minutes, or the average value of a part of the screen (root mean square)
Or a weighted flat portion of the screen
The feedback control is performed so that the average value is constant.
It is. Automatic dynamic range adjustment is one screen of luminance signal
Minute standard deviation, or the standard deviation of a portion of the screen,
Or the weighted standard deviation of one part of the screen
This is performed by performing feedback control so as to be constant. Fig. 10
Of which, 83 are rows for calculating the average value of the aforementioned luminance signal.
It is a pass filter. Compare the output of low-pass filter 83
Gain (set by variable resistor VR1) by amplifier 84
After the reference voltage 86 and the comparison amplifier (differential amplification),
The signal is input to the adder 79 via the switch 81 and added to the luminance signal.
It is. In this way, a feedback loop is established, and the luminance signal
The average value of the signal (the output of the low-pass filter 83) is
The pressure is automatically controlled to be equal to 86. Like this
Then, the luminance signal subjected to the automatic gain control is input to the multiplier 80 next.
Is forced. The output of the multiplier 80 is the standard deviation
It is input to the difference generation circuit 100. First, the subtractor 100-1
Average value of the luminance signal (output of the low-pass filter 83)
After being subtracted, detection is performed by the square detector 100-2.
It is. And, by the low-pass filter 100-3,
After the average value is obtained, it is input to the square root circuit 100-4.
Then, a standard deviation value of the luminance signal is obtained. This value is compared
The amplifier (85)
Dynamic range reference voltage 87 and comparative amplification (differential amplification)
After being input to the multiplier 80 via the changeover switch 82,
Multiplied by the signal. This creates a feedback loop.
And the standard deviation of the luminance signal is based on the dynamic range.
It is automatically controlled to be equal to the voltage 87. In addition,
The standard deviation generation circuit 100 for obtaining the standard deviation value shown in FIG.
Simplified to the standard deviation generation circuit 100 'shown in FIG.
You may. That is, the average value is subtracted by the subtraction circuit 100-1.
After that, absolute value detection is performed by the detection circuit 100-5,
Calculate the average of the variance of the luminance signal with the pass filter 100-6
May be. In this way, the gain and dynamic range
The controlled luminance signal is then input to logarithmic amplifier 92.
You.   In FIG. 10, the changeover switches 81 and 82 are shown.
When the contact Sa is switched on,
Gain setting voltage and dynamics set by resistors VR3 and VR4
A clean range setting voltage is applied, and gain control and
And dynamic range control.   Note that, for example, the gain is manually controlled and the dynamic range is automatically controlled.
Can be selected like dynamic control, and vice versa.
it can.   By the way, in the first embodiment, the color signals RY, B
By multiplying −Y by the degree of compression logY / Y of the luminance signal Y,
Only the luminance signal Y is compressed without affecting hue and saturation.
Shrinking. For this, the degree of compression logY of the luminance signal Y
/ Y is multiplied by R, G, B respectively, and logY / YR, logY / YG, log
The same result is obtained for Y / Y · B. This is logY, logY
/ Y (RY), logY / Y (BY) to inverse matrix circuit
When passed, logY / Y ・ R, logY / Y ・ G, logY / Y ・ B
I understand. Therefore, in this embodiment, the output of the multiplier 80,
In other words, a log bY is temporarily changed by a logarithmic amplifier 92 of 100 dB.
Compress and log (a log bY). Then this signal
And subtract logY, which is the output of adder 48, by subtractor 88.
As a result, a signal of log (a log bY / Y) is obtained. This reduction
The output of the arithmetic unit 88 is logR, logG, logB by the adders 75, 76, 77.
Log (a log bY / Y ・
R), log (a log bY / Y • G), log (a log bY / Y • B)
Obtainable. Then 100dB anti-log amplifier 89,90,9
1 gives a log bY / Y ・ R, a log bY / Y ・ G, a log bY / Y ・ B
That can affect hue and saturation
And only the luminance signal can be compressed. Where the above
The features of the second embodiment compared with the first embodiment will be described.
In one embodiment, the color signals RY and BY have positive and negative amplitudes.
Logarithmic amplifiers 46 and 47, adders 49 and 50, and subtractors 53 and 5
4 and antilog amplifiers 55 and 56 separate the signal into positive and negative
The calculations have to be done and the circuit becomes somewhat complicated.
This is because the logarithm of the logarithm can only be mathematically positive.
is there. However, in the second embodiment, only positive values
The circuit is simple because it processes only R, G, B signals that do not
become.   Next, the floating port used in each of the above embodiments
Another method of the int method will be described with reference to FIGS. 13 and 14.
You.   To perform antilogarithmic conversion of the above floating method
For example, in FIG. 10, the input signal is logR^γ, logG^γ,
logB^γ(In Fig. 3, they are logR, logG, and logB.)
Set the average of these signals to the center of the window width
I have. In contrast, the floating type
Force signal logR^γ, logG^γ, logB^γ(Or even logR, logG, logB
good. ) Is detected and the largest signal is detected.
Signal is set to the upper limit of the window, and -50
The width of the window up to the level range below by dB
It is. The signal smaller than the width of the window,
Signals below -50 dB below the signal level are treated as having the same amplitude.
However, as described above, the human visual characteristics of color images
Since it has no effect, it is truncated once in the window circuit
Would. The actual circuit is logR as shown in Fig. 14.^γ, log
G^γ, logB^γSignal is logR^γAnd logG^γ, logG^γAnd logB^γ, lo
gB^γAnd logR^γTo comparators 93, 94, and 95, respectively.
Is entered. And each comparison output is to determine the maximum value
Read-only memory ROM (lookup table) 96
Is input to The judgment signal from the read only memory 96 is
Selects the maximum signal input to the high-speed multiplexer 97
Therefore, as a switching signal for performing a dynamic switching operation
Function. By this series of operations, as shown in FIG.
To logR^γ, logG^γ, logB^γThe largest signal is
It will always be selected. The low-pass filter 98
This is for removing the switching noise. And this maximum
To obtain a window width of -50 dB from
The output of the filter 98 is equivalent to a window width of 50 dB by the adder circuit 99
Is added to the subtraction circuits 33, 34, 35
Are subtracted and added by an adder circuit 48 (48, 49, 50 in FIG. 3).
And the floating point method
Rix operation is performed.   By the way, the above embodiment is, of course, a digital circuit.
It can be configured as it is. The analyzer shown in FIGS. 3 and 10
A / D conversion of floating log arithmetic circuit for log amount
Lookup using read-only memory
Executed by non-linear matrix operation by table method
Or a floating-point arithmetic digital signal.
Even if implemented by a
Good and practical when digital technology advances in the future
It is a target.   Next, an example in which another function is added to the present invention will be described.
Now. The output of the adder 48 is obtained by logarithmically compressing the luminance signal.
Therefore, a two-dimensional high-pass filter is inserted after the adder 48.
Eliminates multiplicative noise such as uneven lighting.
Thus, structure emphasis can be efficiently performed.
The detailed principle is described in Japanese Patent Application No. 60-272885.
Therefore, I will not explain.   In each of the above-described embodiments, the R, G, and B primary color signals are output from the image sensor.
It was obtained. Complementary color for color image sensor
System: yellow (Ye), cyan (Cy), magenta (Ma)
There is also an example of using a color filter such as. Like this below
An example in which a simple color image sensor is used will be described.
You.   FIG. 15 shows an example of a complementary color filter. Mg, G, Cy, Ye four colors
You are using FIG. 16 shows a color signal in the third embodiment.
FIG. 15 shows a block diagram of a signal processing circuit.
The color filter is obtained from an image sensor attached to the light receiving surface.
Signals logMa, logG, logCy, logYe with logarithmic characteristics are subtracters
Input to 102 to 105 respectively. The value logM to be subtracted is logM = {(logMa + logG + logCy + logYe) / 4} −W / 2 This value is generated by the average value calculation circuit 101.
Is done. Real time of signals output from subtractors 102 to 105
The average value is always the limit value W of the window circuits 106 to 109.
Halved and converted to a linear signal by antilog amplifiers 110-113
Can be reproduced. (Note that using the method shown in Fig. 14
To a window width of 50 dB below the maximum level.
No. ) Linear signals output from antilog amplifiers 110 to 113
Ma, G, Cy, and Ye are converted into luminance signals Y and
The signals are converted into color difference signals RY and BY. Where the matrix
The coefficients of the matrix of the operation performed in the circuit 114 are
Needless to say, it is set according to the characteristics of the complementary color filter.
It is. In the case of the complementary color filter shown in FIG.
If the transmittance is set appropriately, the luminance signal and color difference signal
Is represented by Y = Ma + Cy + G + Ye RY = (G + Cy)-(Mg + Ye) BY = (Mg + Cy)-(G + Ye) That is, the coefficient of the matrix is 1, and the matrix circuit
Reference numeral 114 denotes a simple configuration including only an adder or a subtractor.   Now, the matrix circuit 114 is calculated by the subtracters 102 to 105.
The output signal is divided by the average value M.
You. Therefore, these signals are converted by logarithmic amplifiers 115-117.
After logarithmic compression, logM is added by adders 118 to 120.
LogY, log (RY), log (BY)
You. The logY signal is obtained by the adder 121 and the multiplier 122 in FIG.
Control of gain and dynamic range as shown in
Is performed. On the other hand, the RY and BY signals are subtracted by a subtractor 12.
After the logY signal is subtracted by 3,124, the adder 126,127
The log (logY) output from the logarithmic amplifier 125 is added.
It is. And a linear signal by antilog amplifiers 128 and 129
Converted and visually corrected signal (RY) '=
(RY) logY / Y, (BY) '= (BY) logY / Y
can get. When using complementary color signals, the normal gun
Correction cannot be performed, but logarithmically compressed luminance signal
Subtractor 130 divided by R3 and R4 to give bias voltage V
Gamma correction and color according to the luminance signal
Same as gamma correction by adjusting the level of the difference signal
It is possible to obtain the effect of   Reference numeral 131 denotes a gamma correction function for correcting the color difference signal level.
It is a number conversion circuit. The luminance signal Y 'and the color obtained above
The difference signals (RY) 'and (BY)' are converted to NTSC signals.
Or converted to R ', G', B 'signals by the inverse matrix circuit.
You.   FIG. 17 shows color signal processing in the fourth embodiment of the present invention.
This shows a circuit, in which a complementary color filter is used.
Perform color signal processing. logMa, logG, logCy, logYe signals
Is connected by a floating point anti-log amplifier.
It is shown in Fig. 16 that the signal is converted into a shape signal Mg, G, Cy, Ye.
This is the same as the embodiment. These signals are input to a 4-input adder 13
2 to obtain a luminance signal Y. Further logarithmic increase
After being log-compressed by the width unit 115, the coefficient is added by the adder 118.
M is added, resulting in a logY signal. Adder 121 and multiplier 1
Adjustment of gain and dynamic range by 22
As a result, a compressed luminance signal Y 'is obtained.   Y 'signal passed through logarithmic amplifier 125 and logY signal are subtracted
The log (Y '/ Y) is obtained by the subtractor 133. this
The signal and logMg, logG, logCy, logYe are added by adders 134 to 137.
And each complementary color signal is multiplied by Y '/ Y.
Signal. Thereafter, the antilogarithmic amplifiers 138 to 141 output 4
Two linear signals Mg-Y '/ Y, G-Y' / Y, Cy-Y '/ Y, Ye
Y ′ / Y are obtained, and these are added to the 4-input
The color difference signal (R-
Y) ′, (BY) ′ are obtained.   As explained above, when using the complementary color filter
Also performs color logarithmic imaging in the same way as for the R, G, B primary color signals.
Can be.   By the way, in each embodiment described above, as shown in FIG.
A wide dynamic lens using two image sensors 3a and 3b
Signal (for example, 100 dB)
The output characteristics (photoelectric conversion characteristics) are converted to logarithmic characteristics.
In this case, the number of image sensors can be reduced to one.
There are advantages such as   Hereinafter, the second embodiment of the present invention in which the output characteristics of the imaging side are logarithmic characteristics
Five embodiments will be described.   As a method of converting the output characteristics to logarithmic characteristics, for example,
(Imaging) Logarithmic compression within the element is mentioned. This tool
As an example, a horizontal OFD (overflow
IL-CCD (interline transfer type CC) with Rain 144
D) A description will be given of realization using 145.   The IL-CCD 145 has a light receiving element row 146 in the vertical direction.
The vertical shift registers 147 are alternately arranged, and
Between the element row 146 and the vertical shift register 147.
Transfer gate 148_TGApply
By doing so, each vertical shift register 147 is adjacent
Signal charges accumulated in the light receiving element row 146
Is done. Then, vertical transfer to vertical shift register 147
Clock φ_VThe signal charge by applying
Transfer in the (vertical) direction and transfer to the horizontal shift register 149
The horizontal (horizontal) image is stored in the horizontal shift register 149.
Horizontal shift-like clock φ for prime number_HBy applying
Thus, a CCD output signal can be output via the output amplifier 150.
In addition, an aperture formed adjacent to each light receiving element row 146
The bar flow drain 144 normally has an appropriate value and a positive voltage.
Applied (in the case of n-channel),
The excess accumulated charges are caused to overflow.
In this embodiment, the overflow drain 144
It controls the applied voltage to make the output characteristics logarithmic.
You. The drain is grounded via a resistor R.   The method of logarithmic compression in the element is basically the
The depth of the potential well of the optical element is determined by the exposure time t.
As time passes, it changes according to a function V (t) represented by the following equation.
To Here, the time t is equal to or less than the maximum exposure time (period) T,
That is, O ≦ t ≦ T. A is the degree of logarithmic compression (da
B is a constant that represents the dynamic range).
It is a constant to pass.   In order to derive the above equation (1), the function V shown in FIG.
Consider (t).   In FIG. 18, the horizontal axis represents time t, and the vertical axis represents potential.
Represents the depth of the well, and this function V (t) is the potential
It is a curve showing the temporal change of the depth of a well.   That is, during one exposure period, a small distance from the exposure start point
After a short time, the depth of the potential well is
Because of the shallow depth, all the light signal charges of extremely low brightness are charged.
However, optical signal charges of higher intensity are
Saturates to the depth of the well, and extra than this saturation
The charge is discarded in the OFD 144.   Thus, the depth of the potential well increases with time.
The magnitude is represented by V (t). In this case, the points at each time t
The amount of charge at the depth of the tensile well is considered
To the slope dV (t) / dt of the tangent of V (t) at time (t)
The accumulation of the corresponding optical signal charges is repeated, and this function V
It increases according to (t). Function V (t) at time t
The slope dV (t) / dt of the tangent increases with time t.
You. Therefore, the lower the luminance component, the more the charge accumulation is performed.
The exposure time becomes longer, and the signal level increases accordingly. one
On the other hand, the higher the luminance component, the shorter the actual exposure time becomes,
The signal increase at the high luminance level is suppressed accordingly.
You.   The function V (t) is set such that the above-mentioned suppression ratio becomes logarithmic compression.
Can be obtained as follows.   Some time t₁Tangent P (t) of the function V (t) at₁) Is next
It is expressed by an equation. This slope dV (t₁) Photo charge of luminance with / dt is the exposure period
During T And the tangent P (T) at the maximum exposure time T is
From equation (2) Only the potential well is charged.   Therefore, to make the output characteristics of the element have a logarithmic characteristic
Is P (T) = A log {Q (T)} (5) The following relationship must be established.   By the way, when the incident light amount is 0 (that is, when the charge amount is 0).
Time), P (T) = − ∞ in the above equation (5)
However, in actual photoelectric conversion characteristics, although the photocharge starts from 0,
Correspondingly, P (T) must also start from 0. Follow
Therefore, the above equation (5) takes this into account, P (T) = A log {Q (T) +1} (5 ') Should be. (That is, the photoelectric conversion characteristics of the device
Equivalent to shifting the vertical axis to the right by one. ) Also gain
Changing T is to be able to change T
For example, it suffices to multiply T by B, and T in equations (3) and (4)
May be replaced with BT.   Therefore, from equations (3), (4) and (5 '), Becomes   Where time t₁Is an arbitrary time t from 0 to T,
Since the equation needs to hold, t₁With tAnd rearranging this equation (7) Equation (1) is obtained.   FIG. 20 shows a graph of the function V (t) satisfying the expression (1).
You.   In an actual circuit, for example, a dynamic range of 100 dB
When the compression of the current is performed, for example, the dark current at t = T
Example where the amount of accumulated noise charge is the maximum saturation level Emax of the device
For example, 1/10^FiveDetermine the constant A so that
The gain B is determined so that the signal becomes Emax.   For reference, when A = 1 and B = 1, V (t) is V (t) = log {T / (T-t)}-t / T Becomes Here, log represents a natural logarithm.   Fig. 21 shows the photoelectric conversion of the device when logarithmic compression is performed.
Show characteristics. However, this characteristic has a dynamic range of 50
100dB imaging dynamic range using dB element
An example in the case of enlargement is shown.   In the above IL-CCD145, the depth of the potential well is changed
One way to do this is to increase the OFD gate voltage by the OFD gate barrier.
All charges accumulated in the light receiving area become the lowest and flow to the OFD.
Level V_TwoThe height of the OFD gate barrier
Highest level V₀Up to continuous according to the above formula (1)
It is performed by changing to.   However, the voltage applied to the OFD gate is actually negative.
By reversing the polarity of the above equation (1),_TwoOr
The function shown in broken lines in Fig. 22
Becomes Note that the horizontal axis is time t, and T is, for example, 1/30 sec or 1
/ 60sec.   The above OFD gate control signal S_ThreeIs, for example,
Generated.   FIG. 23 shows a block diagram of an OFD control signal generation circuit.   First, a basic system with fixed gain and dynamic range.
The system is explained according to the flow indicated by the broken line in FIG.
I will tell.   First, the timing output from the system controller 151
Phase-locked sawtooth wave generation
The path 152 has a sawtooth wave S shown in FIG.₁Is output
You. This sawtooth wave S₁Has a frequency of 60Hz (field readout
At 30 Hz (when reading out frames).
Voltage is, for example, V₁It is. However, this peak voltage V₁Than
Low voltage level V_TwoAnd the sawtooth wave S₁Raw when limit
So that the length of the upper side of the screen corresponds to vertical blanking
Is set. This sawtooth wave S₁Is a function generation circuit 153
To the function curve V (t) according to the above equation (1).
Inverted output S_ThreeIs generated. This output S_ThreeIs shown in Fig. 24 (b).
Signal S before inversion shown_TwoThe waveform shown in FIG.
Become. That is, this signal S_ThreeIs the voltage level when t = 0
V_TwoAnd V₀The waveform is clamped by. still,
Consider the γ characteristics of the image sensor in the characteristics of the function generation circuit 153.
In consideration of this, the characteristic obtained by correcting the γ characteristic is represented by the value of A in equation (1).
Γ correction can be performed in the device. An example
For example, the control signal S indicated by a broken line in FIG._ThreeFrom the solid line
Control signal S_Three'To output a gamma-corrected signal
It can also be done. Perform gamma correction in the element in this way
And the gamma correction circuit provided in the video signal processing unit becomes unnecessary,
The circuit can be simplified. (In this case, see Fig. 25 and Fig. 43
Is unnecessary. )   By the way, depending on the subject, a bright subject
Or vice versa for dark subjects. Also,
Some subjects have a wide dynamic range, some are dynamic
Some subjects have a narrow range.   Therefore, the information of every subject always has the same compression characteristics.
If it is not necessary to take images at
Gain control (automatic gain control: AGC or manual
Gain control) or dynamic range control (automatic
Dynamic range control: ADC or manual dynamic range control
Is effective.
become.   The above gain control can be performed by controlling the exposure time.
It is sufficient if B in equation (1) can be changed.
A control circuit for performing this gain control has a sawtooth as shown in FIG.
Output signal S of tooth wave generation circuit 152₁Into the limiter 154
And the voltage level V_TwoAnd then input to the subtractor 155
And the voltage V_TwoSubtract from This subtraction output is output to multiplier 156.
Input and gain control signal S_Four, And again to the limiter 157
Input and voltage V_TwoAnd input to the subtractor 158,
Pressure V_TwoSubtract from The output signal S of the subtractor 158₁′
It is input to the number generation circuit 153. At this time, the gain control signal S
_FourIs a signal processing unit for color logarithmic imaging at the subsequent stage (see FIG. 25).
This logY signal is passed through LPF161.
1 field (or 1 frame) integration
Value, and set the voltage to the appropriate level with the variable resistor 162.
Pressure is applied to one input, the other
Output signal S input to the input terminal and compared and amplified_FourIs switching
The signal is input to the multiplier 156 via the switch 164, multiplied, and controlled by the AGC.
Is controlled. Also, the photographer switches this changeover switch 164.
By setting the variable resistor 165 to an arbitrary value
Manual gain control to multiply voltage is also available
You.   The signal S output through the changeover switch 164_FourIs
This corresponds to B in equation (1).   The dynamic range control is performed by the function generation circuit 15.
3 is the dynamic range control signal S_FiveControl its characteristics with
It is done by doing.   At this time, the dynamic range control signal S_FiveIs as follows
Is generated.   First, the gain control signal S_FourLogY signal through LPF161
And the logY signal before passing through this LPF 161 are denoted by reference numeral 10 in FIG.
Input to the standard deviation generation circuit indicated by 0 (or 100 'in FIG. 11)
Then, a signal output from the circuit 100 is obtained.   This signal is provided (independently of the gain control,
Variable resistor 166 that can be set to level is applied to one input terminal
To the other input of the comparison amplifier 167
With the comparison output through the comparison amplifier 167, the changeover switch 168 is set.
Dynamic range control signal S passed_FiveThe function generator 15
3 to control the dynamic range by the ADC.
U. Also, instead of using the ADC, the photographer sets the switch 168
To the manual side, and the value set arbitrarily with the variable resistor 169.
Can also be controlled manually. Where the signal S_FiveIs
This corresponds to A in equation (1).   In the embodiment shown in FIG. 23, manual and automatic control
Can be performed by the changeover switches 164 and 168.   Video signal processing for color logarithmic imaging in the fifth embodiment
The control section is shown in FIG.   In this embodiment, the logarithmic characteristic can be automatically or manually set.
Change.   By the way, the image processing unit for color logarithmic imaging shown in FIG.
When performing a floating matrix operation,
After converting to near, perform matrix conversion and logarithmic conversion again.
Exchange.   Antilog amplifiers 41 to 43, 89 to 91 used here, logarithmic amplification
If the logarithmic characteristic of the output from the element is fixed,
It may be fixed corresponding to this.   However, the logarithmic characteristic of the output from the element is inside the element.
Or, it can be changed before the video processing unit for color logarithmic imaging
In this case, antilogarithmic amplifiers 41 to 4 are used depending on the logarithmic characteristics at that time.
It is necessary to change the characteristics of 3,89-91 and logarithmic amplifiers 45,92
is there.   Therefore, the dynamic range control signal S_FiveUsing
For the antilog amplifiers 41 to 43, 89 to 91, the divider 171
By 1 / S_FiveAfter making this signal 1 / S_FiveAntilogarithmic increase using
By controlling the characteristics of the width gauges 41-43 and 89-91,
Do the right thing.   For the logarithmic amplifiers 45 and 92, the signal S_FiveAs it is
By controlling the characteristics of the logarithmic amplifiers 45 and 92.
Correction.   The color signals MR, MG, MB passed through the antilog amplifiers 41 to 43 are
The luminance signal MY is generated by the matrix circuit 78, and the logarithm is increased.
It is input to the width unit 45. Note that the broken line B indicates that gamma correction is performed in the device.
The signal S_ThreeInstead of S_Three′
It becomes unnecessary.   In the fifth embodiment, the control is performed by controlling the OFD gate.
Although the number compression is performed, in the sixth embodiment of the present invention,
Uses FT-CCD (frame transfer type CCD) 173 shown in Fig. 26
For storage potential control (control of potential well depth)
More similar functions are realized. (However, with IL-CCD
The light receiving part is not a diffused photodiode, but a MOS
The same principle can be used for a photodiode. )   In FIG. 26, the FT-CCD 173 is adjacent to the light receiving section 174.
A transfer unit 175 is provided, and signal charges accumulated in the light receiving unit 174
Is the fast vertical transfer signal CK in normal usage₁To the application of
Thus, the data is transferred to the transfer unit 175. After transferring to this transfer unit 175
Is the vertical transfer clock φV shown in FIG. 27 (c)._TwoVertical
To the horizontal shift register 176.
Clock φH for horizontal shift of the number of pixels in the direction (FIG. 27 (d)
Repeat) to apply CC through output amplifier
D signal is output.   By the way, in this embodiment, the vertical transfer clock CK is used.₁
Before applying the voltage, as shown in FIG.
Control signal S₆To make a charge signal with logarithmic compression characteristics
Then, as described above, the high-speed transfer clock CK₁Apply and roll
Transfer to the sending unit 175. That is, in this embodiment, the light receiving unit 174
Has a storage potential control signal S₆And vertical transfer clock CK₁And seal
Control signals φV₁so
Express.   As shown in FIG. 26, the control signal φV₁Is the subtractor 17
Voltage V by 7_TwoFrom the signal S_Three(See Fig. 27 (a))
Signal S₆(See Fig. 27 (b))
The vertical transfer signal CK for driving the CCD₁Is subtracted (however, IL-CCD
When applying, the CCD drive signal CK₁Need to subtract
No) signal φV₁Is applied to the input end of the light receiving section 174
I do. In this case the signal S_ThreeThe voltage V_TwoInstead of subtracting
The signal before inversion in the function generation circuit 153 shown in FIG.
(Shown in (b))_TwoMay be added.   ΦV₁To an appropriate level when output from the subtractor 177
Has been converted.   FIG. 28 shows the control signal φV₁Is generated.
That is, the voltage V_TwoFrom the signal S_ThreeSubtract
The generated signal S shown in FIG. 28 (b)₆Is shown in FIG.
Indicating signal CK₁Is further subtracted to obtain the signal shown in FIG.
φV₁Is generated.   By the way, in the fifth embodiment, in the case of IL-CCD, OFD
The logarithmic compression characteristic was realized by controlling the
Controlling the transfer gate as in the seventh embodiment
Inner log compression can also be performed.   FIG. 29 shows the transfer gate control in the seventh embodiment.
2 shows a logarithm-compressed IL-CCD178 in the device.   The logarithmic compression IL-CCD178 in the device uses the OFD gate of the IL-CCD.
Positive voltage V₀Is applied and the signal accumulated in the light receiving
When the signal charge becomes excessive, it flows to the OFD gate side. (Den
Pressure V_TwoLevel or higher. ) However, the signal accumulated in the light receiving section
The charge is the transfer gate clock φ in normal usage.
_TGIs transferred to the vertical shift register and transferred.
Other than when the signal charge leaks to the vertical shift register side.
However, in this embodiment, during charge accumulation (after transfer)
Outside), so that some of it leaks to the vertical shift register side
A simple control signal is applied to achieve logarithmic compression characteristics.   Therefore, the control shown in FIG.
Signal S_ThreeAnd the transfer gate signal φ shown in FIG.
_TGAnd logarithmic compression control as shown in FIG.
Signal φ_TG'And apply it to the transfer gate terminal.
I am trying to. Note that the transfer gate signal φ
_TGIs output from the system controller.   C is set to an appropriate level when output from the adder 179.
Has been converted.   The control signal S shown in FIG._ThreeIndicates the time of charge accumulation (exposure
The control signal S_ThreeVertical shift
The charge leaked to the transistor side is the vertical rotation shown in Fig. 30 (d).
Transmission clock φV₁To sweep it out. this
Therefore, this vertical transfer clock φV₁Is the license of the image sensor.
It is desirable to have the maximum clock speed within the range.
On the other hand, the transfer gate signal output after one exposure period
No.φ_TGSignal charges transferred to the vertical shift register by
Is the vertical transfer clock φV shown in FIGS. 30 (d) and (e)._Two
And output is canceled by the horizontal shift register clock φH.
After that, the logarithmically compressed CCD signal is output. This place
Clock φV_TwoAnd clock φH are output in synchronization with each other.
You. (However, φV_TwoAnd φH are applied with a phase shift of 1/2.
It is. )   In the seventh embodiment, signal charges are stored in a vertical shift register.
In addition to the vertical transfer function of
To improve the overflow charge absorption capacity
Can be done. Also reduce the load on the OFD gate
Can be   Further, in the above embodiment, the logarithmic compression characteristic is
A control signal is generated for
Generating control signals with digital circuits as shown in
Can also.   The control signal generation circuit in the eighth embodiment is shown in FIG.
The output signal of the system controller 151 is
At the address end of the backup table (ROM table, etc.) 180
The applied and read data is analyzed by the D / A converter 181.
It is converted to a log signal and has an appropriate cutoff characteristic.
Control signal S by smoothing through the_ThreeGenerate a
doing. The signal input to the D / A converter 181 is
The signal becomes a fine step-like signal as shown by the solid line in Fig. 32, and the D / A
The signal is converted to an analog signal by the converter 181 and passed through the LPF 182.
The control signal S shown by the broken line in FIG._Threebecome.   In the eighth embodiment, the gain control is performed by the system control.
LA 151 to the above signal S_FourClock speed
And dynamic range control is also performed by the above signal
S_FiveAddress to lookup table 180 according to
Control by reading the information of the corresponding curve.
Will be   Another embodiment of each embodiment for obtaining the above-described logarithmic compression characteristics in the element.
As an example, the method shown in FIG. 34 may be used.   In the ninth embodiment of the present invention, as shown in FIG.
Expose N times and transfer to vertical shift register during light period
And add the charge amount for N times on the vertical shift register.
And performs logarithmic compression within the element.   However, at this time, the exposure time of once, twice,.
As shown in Fig. 33, nonlinearly decreasing according to equation (1),
And the OFD gate voltage is such that the gate height is 1 / N of normal
Is set to N reads (to vertical shift register)
After that, signal charges are read out in the same manner as in normal readout.
Will be   As shown in FIG. 34 (a), each exposure time t₁, T_Two, ...
Later, the transfer gate clock φ_TGAnd apply each dew
The signal charge accumulated during the light time is transferred to the vertical shift register.
And then add it by this vertical shift register. (Fig. 18
The description applies to the IL-CCD. ) But N times
Immediately after the addition of the signal charges at the time of exposure is completed, FIG.
As shown in (b) and (c), the vertical transfer clock φV₁as well as
Apply the horizontal shift register clock φH to output terminal
To output the CCD signal.   By the way, N transfer gate clocks φ_TGmark
After the CCD signal is output after normal
In addition, unnecessary signal charges accumulated in the light receiving
Transfer gate clock (φ in FIG. 34 (a)).
) And then apply one as shown in FIG. 34 (b).
Vertical transfer clock φV_TwoIs applied for the number of pixels in the vertical direction.
(After that, the horizontal shift register has the number of horizontal pixels (not shown).
(With the clock applied). In addition, OFD gate
As shown in FIG. 34 (d), when the gate height becomes 1
Voltage V₀In contrast, the gate height is 1 / N (here, for example, 1/5)
Voltage V₁Is applied.   Thus, in this embodiment, the CCD photoelectric conversion output
The polygonal line characteristic approximates the characteristic to the logarithmic characteristic.   The method of logarithmically compressing and outputting signals in the image sensor is called CC.
The present invention applied to XY address type image sensor instead of D
The tenth embodiment will be described below. XY address type
Since the image sensor sequentially scans the pixels in the screen, the pixels
The timing of the start and end of the exposure differs every time.
Therefore, care must be taken when performing signal compression. Machine
Exposure start / end timing using mechanical shutter
Or when using a field sequential endoscope
Is common to all pixels, similar to the method described in the previous CCD.
Just add a pulse to compress the signal at the right time
No. However, in the general case, the exposure timing
Signal compression pulse is also different
Need to be applied with This method is applied to the XY address
SIT (Static Induction Transistor)
The case where the present invention is applied to the image sensor will be described below.   Fig. 35 shows the structure of one pixel 183 of the SIT image sensor.
(A) shows the structure, and (b) shows an equivalent circuit. No.
When light enters the pixel 183 shown in FIG. 35, holes are generated in the gate 184.
Accumulates, the gate potential rises, and the source 185 and drain 1
The current flowing between 86 increases. Reference numeral 187 is a gate 184
Is connected to the capacitor. Source 185 and de
Detecting the current between the rain 186 from each pixel and obtaining the video signal
You. How to compress the output signal in such an image sensor
The law is described in Japanese Patent Application No. 58-213488.
Therefore, using the principles presented here,
An imaging element that can obtain the characteristics will be described. Figure 36 shows imaging
14 shows an overall circuit configuration of the element device 188.   In the image sensor device 188, the pixel 183 shown in FIG.
It is arranged like a liquor. The gate 184 of each pixel is
, 189-n connected to lines 189-1, 189-2,.
5 is connected to column lines 190-1, 190-2, ..., 190-m respectively
Have been. The drain 186 of each pixel 183 is common to all pixels.
Connected. Row line 189-i (i = 1, 2,...)
n) is connected to the vertical scanning circuit 191 and the column lines 190-j
(J = 1, 2,... M) are the horizontal scanning circuit 192 and the reset circuit 1
Connected to 93.   The vertical scanning circuit 191 includes a vertical shift register 194, an analog
A shift register 195 and a signal mixing circuit 196.
You. The horizontal scanning circuit 192 includes a horizontal shift register 197,
A selection switch 198 and a video line 199 are provided. Re
The set circuit 193 includes a reset switch 200. the above
A signal waveform indicating the operation timing of the image pickup device
Figure 7 shows.   In FIG. 37, φs is output from the horizontal scanning circuit 192.
Pulse. Also, φ_G1, Φ_G2, Φ_GnIs the vertical scanning circuit 19
This pulse is output from 1 and φ_RSIs the reset circuit 193
Is a pulse applied to.   The operation of the image pickup device 188 shown in FIG.
This is described below with reference to FIG.   Each pulse φ in Fig. 37_G1, Φ_G2, ... φ_GnAt the voltage
V_RDIs a voltage for reading the corresponding row line 189-i.
This voltage V_RDIs applied vertically shift
Provided by register 194. Also, voltage V_OFIs a horizontal bra
Voltage that is applied every
Provided by the shift register 195. The signal mixing circuit 196
Vertical shift register 194, analog shift register 195
The output is mixed at an appropriate timing, and the pulse φ_G1, Φ_G2,
... is generated. In the horizontal scanning circuit 192, the horizontal scanning
The horizontal shift register 197 operates every period.
, The horizontal selection switches 198,…, 198 are sequentially opened,
190-1,190-2,190-n signals are sequentially sent to video line 199
Read. In the reset circuit 193, the reset switch
Switch 200 is pulse φ every horizontal blanking period_RSTo
Open synchronously. Pixel 1 now connected to row line 189-1
Think about 83. Pulse φ_G1Is V_RDWhen the water
The signal of each pixel 183 is read sequentially by the operation of the flat scan circuit 192.
Will be issued. In the subsequent horizontal blanking period, V_OFIs each
Applied at a high value to completely reset pixel 183, and
When the reset switch 200 is opened, the reset of the pixel 183 is performed.
The gate of pixel 183 has a low value at the start of integration.
Becomes Next horizontal blanking period after one horizontal scanning cycle
At V_OFIs a slightly lower value. On the other hand,
The gate potential of the element 183 depends on the amount of light incident on the pixel 183.
It is rising. For pixels 183 with a large amount of light,
The clip potential is clipped, and the potential of pixels with a small amount of light is
It is kept without being ripped. Therefore, the amount of light
Only the signal of the pixel 183 is compressed. Next blanket
This signal compression operation is performed again during the
You. Compression level is voltage V_OFIs determined by Hereinafter the same
As described above, signal compression is performed every horizontal blanking period.
Φ after vertical scanning cycle_G1But again V_RDPressure
The compressed signal is read. Where V_OFThe voltage of the 38th
Figure φ_AIs changed according to equation (1) as shown by
For example, a logarithmically compressed signal output is obtained.   Image connected to another row line 189-i '(i' ≠ 1)
For the element 183, the pixel is operated by the operation of the vertical scanning circuit 191.
Signal φ delayed by 93 read timings_GBut
Added. (This delay is equal to the read voltage V_RDabout
In the vertical shift register 194, the voltage V_OFAbout analog
It is provided by a shift register 195. ) Therefore, each pixel 183
Is exactly the same as row line 189-1,
A compressed signal is obtained from all pixels 183.   The degree of signal compression is input to the analog shift register 195
Pulse φ as a control signal for logarithmic compression_Achange
Can be set freely. This pulse φ_AIs always O
Bolts provide linear output characteristics that are not compressed
You. Also, this pulse φ_ASwitch between high voltage and low voltage
In this case, the characteristic is represented by two broken lines. Also (1)
Adjust the amplitude, slope, etc. even when matching the function of the formula
Change the dynamic range of logarithmic compression
Can be done. For this reason, AGC and ADC
Pulse control φ_ACan be performed.   The analog shift register used for the vertical scanning circuit 191
For example, a BBD (Bucket Brigade Device)
Can be used. This is shown in FIG. 39 (a)
MOS transistor Q1 and capacitor C1
Structure as shown in FIG. 39 (b)
With. Analog shift with simple circuit using BBD
Register 195 can be configured.   The method of logarithmically compressing the signal and outputting it using the method described above
The formula can be applied to other XY address type image sensors.
You.   Next, use the CMD (Charge Modulation Device)
An eleventh embodiment will be described.   FIG. 40 shows the configuration of one pixel of CMD, and FIG.
FIG. 2B shows an equivalent circuit. Normally on gate 201
Has a negative voltage applied. Gate when light enters
Holes accumulate under 201 and the potential increases. To gate 201
Higher voltage (negative voltage) than during light accumulation due to signal reading
When applied, the current between source 202 and drain 203
It flows according to the amount, and the signal of the pixel is read. Gate 20
When a positive voltage is applied to 1, holes below the gate 201 disappear, and
Set. The overall configuration of the image sensor is shown in FIG.
Pixel 183 is replaced with the CMD pixel shown in Fig. 40.
Become. Further, a reset circuit is unnecessary. Operation timing
The signal waveform indicating the signal is the same as in FIG.
Loose φ_G1, Φ_G2, ... φ_GnVoltage to suit CMD
Just change it.   In such an image sensor, the horizontal
A positive voltage is applied to the gate 201 during the blanking period.
Thus, the pixel is reset. The next blankin
During the reset period, a voltage lower than that at the time of reset is applied. incident
In a pixel having a large amount of light, the gate potential becomes a positive potential,
The signal is clipped. On the other hand, pixels with low light
No. is kept as it is. For this reason, images with large light
Only raw signals are compressed. Below horizontal blanking period
This compression operation is performed every
By changing the voltage applied to the
Thus, a logarithmically compressed signal output is obtained.   The present invention applied to still another XY address type image sensor
As the twelfth embodiment of the Ming, AMI (Amplified Mos Imager)
An example using is shown below.   FIG. 41 shows an equivalent circuit of one pixel 204 of the AMI. Photoda
Ide 205 is connected to the gate of MOS transistor 206.
The MOS transistor 206 has a drain connected to a MOS transistor.
207 is connected to the source. Photodiode 205
Is reset by a positive voltage via the diode 208
You. When light enters, the cathode voltage of the photodiode 205
The rank drops. Therefore, the current of the MOS transistor 206 is
The intensity decreases as the light intensity increases, and this is obtained as an output signal.   The configuration of the imaging device is as shown in FIG. Basic
Is the same as FIG. 36, and the same elements are denoted by the same reference numerals.
doing. A row line 209 for reading is connected to one pixel.
Two row lines 210 are provided for the set.
You. The row line 209 for reading is the vertical shift register 194
Row line 210 for reset is connected to the analog shift
Connected to the register 195.   Voltage V applied to reset row line 210_OFIs a signal reading
At the time of protruding, it is set to OV. High after signal read
When a different voltage is applied, each pixel passes through a diode 208.
Then, the photodiode 205 is reset. Then V_OF
As the pixel is reduced, the light intensity is reduced
The signal is kept as it is, but in pixels with a large amount of light
The signal is compressed because the voltage is clipped. This
V_OFIs changed according to the equation (1) to obtain the logarithm
A compressed signal can be obtained.   As described above for the various image sensors, XY
Even in an address type image sensor,
Change the voltage at which the signal is clipped appropriately and
The voltage changes for each pixel according to the signal readout timing.
To compress the signal logarithmically in the image sensor
be able to.   The fifth to twelfth embodiments after FIG.
Explains an example of inner logarithmic compression.
Has a mosaic-like color filter array on the front of the device.
By mounting or using a three-plate structure,
Can appear.   Using elements that realize these logarithmic compression characteristics, logarithmic pressure
A reduced color imaging device can be realized as shown in FIG. on the other hand,
Using the elements of these embodiments, the signal for automatic gain control in the element is used.
No.S_FourAnd signal S for automatic dynamic range control_FiveTo make
Automatic gain control circuit, automatic die in series with the logY signal
Combination of the dynamic range control circuit and display of its output
Conversely, by using it as a logY signal for
Extended dynamic range for narrow subjects
Can be performed.   In the video signal processing unit of FIG.
The part performs this operation. In FIG. 43, parts other than A
It is the same as FIG. 25, and the part A is shown in FIG. 10 and FIG.
, The description is omitted.   Each embodiment of the logarithmic compression within the device up to now
Devices, especially CCDs.
There is also an imaging tube such as a vidicon. Vidiko out of the tube
The photoelectric conversion characteristic is originally a logarithmic characteristic.
ing.   Therefore, when using a vidicon, output
Using logR, logG, logB, the image for color logarithmic imaging
This object can be achieved by inputting the signal to the signal processing unit.   However, in this case, the logarithmic characteristics are fixed, and ADC and
The manual adjustment is performed electrically at a later stage.   Logarithmic pressure in the device in the fifth to twelfth embodiments described above
The following advantages are obtained as the effect of the reduction.   By performing logarithmic compression within the element, (1) The output from the element itself has a logarithmic characteristic
Therefore, the signal processing at the subsequent stage can be simplified and the size can be reduced.
You. (2) Logarithmic compression with one or one set (for three-plate type) of elements
And no memory is required for compression. (3) During the output of one element, the entire dynamic range of the subject
Can be inserted, and the subject can be
Extract and display image information of any brightness level
Can also. (4) Since logarithmic compression is performed inside the element, outside the element
(Use multiple devices or memory after multiple imaging
S / N compared to performing logarithmic compression
Will be better.   The signal is compressed in the CCD image sensor, and the logarithm
The embodiment having the property has been described above. Well, Con
When capturing an image of a subject with low trust, etc.
It is more convenient to give the force an antilogarithmic characteristic (exponential characteristic)
No. This involves multiplying and subtracting positive values after logarithmic transformation of the signal.
Calculation to reduce the displayed dynamic range.
And Rather than performing this operation with a signal processing circuit
Performing in an image sensor is advantageous in terms of S / N. Less than
Below, the CCD image sensor output is set to have exponential characteristics.
An example will be described.   FIG. 44 shows a CCD device used in a thirteenth embodiment of the present invention.
The configuration is shown. 221 is a light receiving unit, and 222 is a light-shielded storage unit.
You. Photodiodes 223 are arranged in rows and columns in the light receiving section 221.
Are lined up. Vertically shaded photodiode rows
Next to shift register 224 and overflow drain 225
It is arranged in contact. 226 is perpendicular to photodiode 223
Transformer controlling charge transfer between direct shift registers 224
227 is the saturation level of the photodiode.
This is an OFD gate that controls the signal. In the storage section 222 that is shielded from light
Is provided with a vertical shift register 228 connected to the light receiving unit 221.
The horizontal shift register 229 has a vertical shift register.
It is connected to the register 224. Horizontal shift register 229
An output amplifier 230 is connected to one end. That is, this C
CD is a so-called frame interline transfer CCD
(FITCCD).   In such an image sensor, the accumulated signal charge is
The light is immediately transferred to the light-blocking storage unit 222, and is received during light storage.
The vertical shift register 224 of the optical unit 221 is empty. So
Here, during light accumulation, the potential of the transfer gate 226 is properly adjusted.
Control the vertical shift register 2 from the photodiode 223
Output by extracting the charge overflowing as 24 as a signal
Can have an exponential characteristic.   Figure 45 shows the photodiode, vertical shift register,
Structure and potential distribution of bar flow drain
FIG. FIG. 46 shows the case where voltage is applied to the transfer gate.
Pulse φ_TGIs shown. Pulse φ_TGIs the photodiode
A positive signal such that all loads transfer to the vertical shift register
This is a pulse that decreases from the bell Va in a convex curve. this
When the curve is represented by V (t), V (t) is a perfect exponential characteristic.
Satisfies the following equation with respect to time t.
As given. Where a is gain, b is dynamic range, c is OFD
It is a constant for adjusting the DC component of the gate voltage.   The curve V (t) shown in FIG.
Loose φ_TGIs applied to the transfer gate 226 during the light accumulation period.
The output of the exponential characteristic
You.   In practice, instead of the continuous curve shown in FIG.
Line approximation or other approximation functions may be used.   The pulse φ of the curve shown in FIG._TGWhile applying
When photocharges are accumulated, the
Potential barrier between the logic and vertical shift registers is low
Therefore, the charge is divided and accumulated in both (Fig. 47
(A)), the barrier increases with time and the charge is
It becomes accumulated only in the iod (Fig. 47
(B)). In this case, the weaker the amount of incident light, the more the charge
The time cannot be exceeded soon. Therefore, the vertical shift
Regarding the electric charge stored in the
The accumulation time is short, and the accumulation time is longer as the incident light intensity is stronger.
Will be. As a result, the exponential function shown in FIG.
An incident intensity versus output characteristic is obtained. In this case the photo
Mode and the vertical shift register have the same saturation charge.
Then the slope of the tangent near the saturation of the output signal doubles
The logarithmically converted signal has twice the gain
Is equivalent to To further increase this magnification, a photo die
It is necessary to increase the saturation charge of the ode. That is, the magnification
In order to make it N times, the saturation charge of the photodiode must be
It is necessary to multiply (N-1) times. This is a photodio
It is necessary to increase the size of the card part or increase the applied voltage
However, in the former, the element size increases, and in the latter, there is a restriction on the withstand voltage.
Difficult from.   Therefore, the charge of the photodiode is pulsed during light accumulation.
About the method of increasing the magnification by discharging the waste
It writes in.   Fig. 49 shows the pulse φ applied to the transfer gate._TGPassing
And the pulse φ applied to the OFD gate_OFDGIs shown. Pulse φ
_TGDecreases from Va according to equation (8), and when OV is reached, the voltage
Is increased by Va, and continuously changes according to equation (8).
Become On the other hand, φ_OFDGIs φ_TGJust before reaching OV
Ruth is added. This change is repeated an appropriate number of times
You.   This pulse φ_OFDGIs added, from the start of light accumulation
Is the same as in the previous embodiment. And photodio
At the moment when the charge of the_OFDGPulse
As a result, as shown in FIG.
Iodine charge is drained to the overflow drain
You. Immediately after this, the photodiode and vertical shift register
The potential barrier between the two becomes lower, as shown in FIG.
Light accumulation is resumed as before. Thus, the potential
As the barrier rises over time,
As shown in FIG.
Will be accumulated in the star. This operation is repeated
Thus, the input-output characteristics shown in FIG. 51 are obtained.
When charge discharge during accumulation is performed n times,
The slope becomes (n + 2) times, and as shown in FIG.
In the case of two discharges, the gain is quadrupled.
You.   In a CCD image sensor as shown in FIG.
The normal operation of storing photocharges in the diode is performed.
I can make it. Also, during photo accumulation,
Barrier between the drain and overflow drain
By appropriately increasing the
As described above, logarithmic compression can be performed. So
Switching between this operation and the operation of the exponential characteristic
By using the same element, logarithmic output and exponential
Sexual output. Therefore, like this
The output of the various elements is connected to the logarithmic imaging processing circuit as described above.
By continuing, the dynamics observed according to the subject
Capable of imaging while reducing or expanding cleansing
It works.   FIG. 52 shows a configuration example of the logarithmic imaging signal processing unit. Basic
Is the same as FIG. 3, and the same reference numerals denote the same components.
It is attached. Logarithmic amplifiers 231-233 are provided in the circuit.
The signals input to the antilog amplifiers 41 to 43 are
4 to 239 can be switched. This
Device outputs have logarithmic characteristics due to the operation of switches 234 to 239
In this case, the antilogarithmic amplifiers 41 to 43 and the element output
Input signal to logarithmic amplifiers 231-233
Thus, conversion to a linear signal is performed. Also, in the element
The dynamic range of the image is continuously changed
If set, log and antilog accordingly
It is only necessary to control the gain of the amplifier and perform appropriate conversion.
Is similar to that described in FIG.   The luminance signal output from the matrix circuit 44 is logarithmically amplified.
To switch 45 or exponential amplifier 240 by switches 241,242.
Input and return to logarithmic or exponential characteristics at the time of imaging.
You. If the element output is exponential,
In this case, there is no need to use
The output of the average value calculation circuit 40 is set to 0 level by the switch 243.
Is switched to. As a result, the subtracters 33, 34, 35 and the adder
Addition and subtraction in the arithmetic units 48, 49, and 50 are not performed, and
Correct visual color difference signal correction for output
It is.   Other processing is the same as that already described.
Therefore, the description is omitted.   In the above embodiment, the photodiode is formed by the diffusion layer.
That was the case. Photo diode to MOS gate
When formed by MOS photodiodes using
A fourteenth embodiment of the present invention will be described below. Figure 53 is a photo-diode
FIG. 3 is a diagram showing a cross-sectional structure and a potential distribution near a node.
You. 244 is the gate on the photodiode (PD gate), 2
45 is vertical shift register gate, 246 is overflow
It is a drain. Fig. 54 shows the pulse φ applied to the PD gate.
_PDGIs shown. φ_PDGIncreases from level 0 according to equation (8)
Pulse.   Accumulate photocharges while applying such pulses
In the beginning of the accumulation, the potential well under the PD gate
Due to the shallow depth, the charge shifts vertically as shown in Fig. 55 (a).
It is also accumulated in the register, but potential over time
Since the well becomes deeper, the electric charge thereafter becomes as shown in FIG.
Is accumulated only in the photodiode. Soshi
The output has an exponential characteristic according to the same principle as the previous embodiment.
Is obtained. The exponential gain is
The larger the saturation charge, the larger the charge.   In this embodiment, an electrode exists above the photodiode.
Therefore, light sensitivity is slightly reduced, but transfer gate
Or an overflow control gate is no longer necessary.
And the structure is simple.   In FIG. 44, the configuration of the CCD image sensor is shown as a FIT-CCD.
The reason for this is that images are continuously captured.
After imaging one screen, stop imaging for the period necessary for signal readout.
Stop and start the light accumulation of the next screen after signal reading is completed.
In this case, an IL-CCD can be used.   The overflow drain is connected to the photodiode.
Adjacent, but this is below the photodiode
So-called vertical over using n substrate through P layer
A flow drain may be used. Apply to n substrate
Voltage and the overflow drain
To change the potential barrier between
This allows the same operation as in the previous embodiment to be performed.
Wear.   In the above embodiment, the apparatus and method for compressing the logarithm within the element are described.
However, for example, as in an electronic endoscope,
If there are steps, the light intensity of the light source forming the illumination means is changed.
Illuminance can be logarithmically compressed.
You.   In this case, the image pickup device forming the image pickup means is a color filter.
If equipped with a filter array, color
-Image pickup is performed, and the prior application example (Japanese Patent Application Laid-Open No. 58-160917, U.S.A.)
SP-4584606).   On the other hand, the image sensor does not have a color filter array
For monochrome (RGB etc.) plane sequential imaging
A fifteenth embodiment of the present invention will be described. Sequential shooting
In the case of the image method, for example, an exposure amount (illumination)
The irradiance (that is, (light intensity) x (time)) is 1: 300 (element
(When the dynamic range of the body is 50dB)
Next, the illumination light is sequentially changed.   After each illumination period, the shading period for the vertical transfer period of the element
Provide a gap. In this case, the lighting means is an RGB rotation filter
1:30 for the ratio of the (aperture) window of each color filter for each color
Can be set to 0, or for RGB strobe lighting
The ratio of the ON time for each color may be 1: 300.   In this way, for example, the exposure amount is changed twice for the same color R,
The output data is stored in the R frame memory for the first time.
Second, the first data is stored in the R frame memo.
Addition to the second output data while sequentially reading from the memory
Then, the contents of the frame memory are written into the added data.
Change. By performing this for each of the R, G, B colors,
Image data with logarithmic characteristics of line approximation
It accumulates in moly. After the RGB color data is complete,
For example, the color shown in Fig. 10 via the D / A converter
After input to the video processing unit for logarithmic imaging, RGB output or NTS
The color is displayed on the monitor as C composite output.   (In this embodiment, the frame memory is stored for each color.
For each exposure level (2 stages in this example), and for each exposure level
Is stored in the frame memory at each
May be added. )   Fig. 56 shows an electronic scope focusing on the frame memory section
1 shows a configuration of an apparatus.   In the figure, an electronic scope device 301 is inserted into a body cavity or the like.
The electronic scope 302 is elongated so that it can be inserted
The electronic scope 302 is connected, and the light source unit 303 and the signal processing unit
Scope control unit 304 with built-in step and this scope
A monitor for display connected to the video output terminal of the controller 304
305.   The tip 306 of the electronic scope 302 has an objective for imaging.
Lens 307 and a fixed lens disposed on the focal plane of the objective lens 307.
A body image sensor (SID) 308 is housed. Also, this electron
A light guide 309 for transmitting illumination light is provided in the scope 302.
Is inserted, and the incident end face side is connected to the light source section 303.
The white light from the xenon lamp 310 etc.
Is the rotary filter 311 or the rotary filter 311 shown in FIG. 57 (a) or (b).
Is R (red), G (green), B (blue) through 312
Light is irradiated in a sequential manner.   Then, from the exit end face of the light guide 309, the illumination lens
The light is emitted toward the subject through a noise 313.   The xenon lamp 310 emits light with the power of the power supply 314.
You. The RGB rotation filter 311 or 312 is rotated by the motor 315.
This motor 315 is driven by a motor driver 316
Driven by signals.   An SID drive (not shown) is installed in the scope control unit 304.
Drive signal from this SID driver
The signal read from SID 308 by applying
A / D converter 31 in frame memory unit 317 via cable
8 and converted to a digital signal. This A / D
The signal passing through the inverter 318 passes through a multiplexer 319.
Input to the three multiplexers 321, 322, 323.   These three multiplexers 321,322,323
One pair of adders 324a, 324b; 325a, 325b; 326a, 326b and 1
Paired multiplexers 327a, 327b; 328a, 328b; 329a, 329b and
Is connected.   Further, the above-mentioned pair of adders 324a, 324b; 325a, 325b; 326
a, 326b and the multiplexers 321,322,323
A pair of multiplexers 327a and 327 for switching between one of the contacts
b; 328a, 328b; 329a, 329b output is a pair of buffers 333a,
333b; 334a, 334b; 335a, 335b, a pair of R frames
Memory 336a, 336b, G frame memory 337a, 337b, B
Input to the frame memories 338a and 338b. These one pair
Frame memories 336a, 336b; 337a, 337b; 338a, 338b
Via latches 339a, 339b; 340a, 340b; 341a, 341b, respectively.
Connected to the adders 324a, 324b; 325a, 325b; 326a, 326b.
Read from frame memories 336a, 336b; 337a, 337b; 338a, 338b
Data and data from multiplexers 321, 322, 323
327a, 327b; 328a, 328b; 32
9a, 329b and buffers 333a, 333b; 334a, 334b; 335a, 335b
Via frame memory 336a, 336b; 337a, 337b; 338a, 3 again
38b can be written.   Also, a pair of frame memories 336a, 336b; 337a, 337b; 33
8a and 338b go through multiplexers 342, 343 and 344, respectively.
Connected to the video D / A converter 345. This bidet
The analog signal converted by the D / A converter 345
For example, input to the video processing unit 346 for logarithmic color imaging shown in FIG.
The output is displayed on the monitor 305.   The pair of frame memories 306a, 306b; 307a, 307b; 308
a, 308b, one frame memory 306a, 3
07a and 308a are, for example, read and write in even field
And the other frame memories 306b, 3
07b and 308b perform reading and writing in odd fields
Used for   By the way, the RGB rotation filter 311 forming the light source unit 303
Or 312 is shown in FIG.   FIG. 57 (a) shows the electronic scope built into the tip 306.
Line transfer method and frame transfer method
Or XY address system.
FIG. 2B shows the case where the interline transfer method is used.
This is an example.   In FIG. 57 (a), the R dichroic filter 35
1a, 351b, G dichroic filters 352a, 352b, B die
Croic filters 353a and 353b have a filter area of 1: 300
The ratio is set. (For example, 1R, 300R, etc. have an area ratio of 1: 3
Means 00. )   In FIG. 57 (a), each filter (for example, 351a, 3
52b), a light-shielding part is provided.
In the RGB rotation filter 312 in the case of the transmission method SID,
As shown in (b), R, G, B
Dichroic filters 354a, 354b; 355a, 355b; 356a, 356
b are each formed continuously, and each is 354, 35
Shown at 5,356.   The color of each filter is determined by using the start marker 357 (No. 57).
(In the case of Fig. (A)) and 358 (in the case of Fig. (B))
And read timing after each exposure.
361a, 361b; 362a, 362b; 363a, 363b (Fig. 57 (a)
Case), 364a, 364b; 365a, 365b; 366a, 366b (Fig. 57
In the case of (b)), the position detection sensor (or reader) 367
(See Fig. 56). Hereinafter, FIG. 57 (a)
The case where the filter 311 is used will be described.   For example, 1R (1G), R (G) indicated by ((1B))
((B)) Dichroic filters 351a (352a) ((35
3a) The image data in the even field exposed in
A / D converter 318, multiplexer 319, multiplex
321 (322) ((323)), 327a (328a) ((329
a)), the latch 333a (334a) ((335a))
Frame memory 336a (G frame memory 337a) ((B
It is written to the frame memory 338a)) (composite same order).
Next, R (G) indicated by 300R (300G) ((300B))
((B)) After exposure with filter 351b (352b) ((353b))
The A / D converter 318
Kusa 319, multiplexer 321 (322) ((323)), addition
324a (325a) ((326a)), multiplexer 327a (32
8a) ((329a)) and buffers 333a (334a) ((335
a)) through the R frame memory 336a (337a) ((338
a)). In addition, each frame memory 336a-338b
Indicates the data writing speed and reading speed are SID308
Depending on the transfer capacity of the
It is decided to stand up. Also, as shown in FIG.
At the time of embedding, an address is specified as shown in FIG.
(B), the speed is twice as fast as
Read / write is performed and read as shown in FIG.
Sometimes the data read from the frame memory is latched.
Can be held and output at the time of a write signal.
I'm trying.   That is, latches 339a-341b are synchronized with the read timing
Data from each frame memory 336a-338b
Touch. It is imaged under the 1R filter 351a and the latch 33
The data latched in 9a is captured by the 300R filter 351b
When the input data is input, it is added by the adder 324a.
Later (for even fields),
And written into the R frame memory 336a.   That is, the total exposure of R using the R filters 351a and 351b is
Upon completion, (1R + 300) is stored in the R frame memory 336a.
Data under the lighting of R) is stored.   The same applies to G and B.   This is repeated for R, G, B, and RGB data for one field is
Data is stored. When this operation is completed, the multiplex
321 to 323 and 342 to 344 are switched to perform the same operation.
The other frame memory 336b, 337b, 338b for R, G, B
Write data to the frame memories 336b, 337b, 338b sequentially
At the same time, the R, G, B frame
Mori 336a, 337a, and 338a are in read mode,
Data is simultaneously read from the flash memory 336a, 337a, 338a.
After passing through the 3-channel video D / A converter 345,
A video signal processing unit for color logarithmic imaging of a stage (for example, see FIG. 10)
346), the output of which is displayed on the monitor 305.
It is.   In the light source unit 303 shown in FIG. 56, the motor 315
Rotation speed detector 37 such as rotary encoder, tachometer, etc.
Motor driver 316 to which the rotation speed detected in 1 is input
The rotation is controlled at a constant speed.   The frame memory 317, motor driver 316, etc.
Is controlled by the system control unit 151.   In the above description, the rotary filter 311 shown in FIG.
As described above, when the IL-CCD is used as SID308,
In this case, the rotation filter 312 shown in FIG.
Vertical shift register at the timing of the data marks 364a to 364b
The same color on the image pickup surface.
U.   The operation after data acquisition is the same as described above.
You.   Utilizes the element shutter function when using the above IL-CCD
The compression ratio setting can be changed,
Exposure can be performed continuously, minimizing light-shielding parts
Of the exposure period (at one round of the rotating filter)
It has the advantage that the ratio can be increased.   In the fifteenth embodiment, the illumination period is controlled to
Performs color imaging with several compressions, but the logarithm
Applying to an electronic scope using a compressed image sensor
Is the logarithmic pressure inside the element described above,
By replacing with the color imaging device of the reduced embodiment,
realizable.   In addition, the imaging method of the electronic scope is changed to the RGB plane sequential imaging method.
, The element without the color filter array
Use a system of internal logarithmic compression and, for example,
Light source, multiplexer, A /
D converter for 3 frames (actually the 15th implementation
(Similar to the example, capacity is preferably 6 frames.)
By incorporating frame memory, video D / A converter, etc.
It is feasible.   An electronic scope is formed with the above-mentioned color logarithmic compression characteristics.
When it is formed, there is a disadvantage that the latitude is narrow
The characteristics of the electronic scope can be improved.   For example, for deep subjects such as the esophagus and intestines,
In a conventional electronic scope image, the back of the subject
There is a problem that it is crushed black and the front side is saturated with white
However, according to the electronic scope of this logarithmic compression characteristic,
Such a problem can be resolved, and
Not only can observation and measurement become possible,
Image information of the dynamic range of
Can be recorded on a disk, etc.
Is very advantageous in image processing.   An example in which logarithmic color imaging is applied to an electronic camera
This will be described below.   Electronic cameras replace cameras that expose images on silver halide film.
And record the electrical signal output from the image sensor on a recording medium.
By doing so, a still image is recorded and reproduced. No.
Figure 59 shows this conceptual diagram.   Reference numeral 401 denotes an electronic camera imaging device, and 402 denotes a playback device. Imaging machine
At 401, an image of a subject is captured by an image sensor 404 using a lens 403.
Project. The output of the image sensor 404 passes through the signal processing circuit 405
The recording medium 406 is recorded on, for example, a magnetic disk. This magnet
The disc is loaded on the player 402 and read by the recording / reproducing circuit.
The output signal is sent to a TV monitor 408 through a signal processing circuit 407.
And play back the image. Such an electronic camera system
When performing logarithmic color imaging in the system 400,
Functions such as logarithmic compression and color signal processing
Various methods depending on how to incorporate it into the image machine and playback device
Can be considered.   The sixteenth embodiment shown in FIG.
This is an example that incorporates all the principles. In the imager 401
The video output of the image sensor 404 is input to the signal processing circuit 405.
You. In the signal processing circuit 405, for example, as shown in FIG.
There is such a logarithmic imaging processing circuit 409. Means as described above
With a dynamic range of, for example, 100 dB
Here, the image signal is a logarithmic compression process of the luminance signal and a color difference signal.
Is subjected to visual correction processing, and is recorded on the recording medium 406.
You. The dynamic range of the signal after the above processing is
Since it is compressed to about 50 dB like a normal signal,
It is possible to record with the same recording capacity as usual, for example,
In the case of the still video floppy system standard, one fee
The field signal is recorded on one track of the magnetic disk.   The recording medium on which the signal has been recorded is reproduced by the reproducer 402.
The signal is input to the signal processing circuit 407. This place
In this case, the signal processing circuit
It may be exactly the same as the corresponding one. Video signal is a signal processing circuit
Output from 407 and transmitted to TV monitor.   As described above, all logarithmic color imaging
In the case of performing on the image machine 401 side, signal gain and dynamics
It is necessary to perform all control of the
is there. These controls are performed by the gain dynamic
Provision of a switch and a variable resistor for setting
Should be set. Also, the image sensor 404 is continuously
Operate to capture multiple images continuously (continuous shooting
Mode), it can be set automatically. This
This includes an automatic gain adjustment circuit as shown in FIG.
Using the dynamic dynamic range adjustment circuit,
Automatic adjustment is performed using the average value of the luminance signal.   When capturing only one image, the front
Automatic control using frame information is not possible. in this case
Is equipped with a photometric element in the
Gain and dynamic range are set using the
To take an image.   FIG. 61 shows a 17th embodiment in which logarithmic imaging processing is performed by a playback device.
Show. In the imaging device 401, the exposure amount differs by 50 dB 2
Takes an image (dynamic range 50dB)
Record on the body. As the means, for example, the imaging device 404
The light accumulation time is reduced by the mechanical shutter or the image sensor itself.
An image is taken at a shutter speed of 16.7 ms. this
The video signal at the time to the recording medium 406 via the signal processing circuit 405.
Record. Next, set the light accumulation time to 0.05 ms to perform imaging.
You. This video signal is recorded again on the recording medium. like this
For example, a signal with an exposure amount different by 50 dB
It is recorded on two tracks of the disk. That is, in this case
Requires the recording capacity of the recording medium to be twice that of a normal recording medium. one
On the other hand, in the reproducing device 402, the two sheets recorded on the recording medium
The image is subjected to logarithmic imaging processing into one image. For example, magnetic
Two tracks of the disc are played by the double head,
Two images are read simultaneously. This image signal is a signal
The RGB separation circuit in the processing circuit 407 generates a pair of primary color signals.
After that, logarithmic amplification is performed in the same manner as shown in FIG.
Signal for each primary color after logarithmic conversion
Is added, and the primary color signal has a dynamic range of 100 dB.
No. Hereinafter, logarithmic imaging processing in the signal processing circuit 407 will be described.
The circuit 409 compresses the luminance signal and corrects the color difference signal.
Done. The processed video signal is output from the player.
Is shown on the TV monitor.   As in this embodiment, the logarithmic imaging processing circuit is
, The image gain and dynamic range
Has the characteristic that it can be set freely in the player 402.
The image already recorded on the recording medium.
It can be observed while changing the effect of the image.   The above method requires twice the capacity of the recording medium
However, by compressing the data,
The amount can be reduced.   FIG. 62 shows an 18th embodiment of the present invention in which the capacity of the recording medium is small.
This is an example.   The signal output from the image sensor 404 in the playback device 401
Compression in the data compression circuit 410. This signal is
The information is recorded on the recording medium via the width device 411. Playback machine 402
The signal read from the recording medium 406
After being restored to the signal before compression by the circuit 412, signal processing
The signal is sent to the circuit 413. Here, the logarithmic imaging processing circuit 409
The compression of the luminance signal and the correction of the color difference signal are performed,
No. is obtained.   In this embodiment as well, as in the case of the seventeenth embodiment,
The gain of the image and the
The dynamic range can be set freely. Ma
A function conversion circuit is provided in the last stage of the signal processing circuit 407.
In addition, the luminance signal can be changed. This shines
The best effect on image reproduction as well as logarithmic compression of the degree signal
Can be set to have.   Next, a circuit for performing logarithmic imaging processing is provided by an imaging device and a playback device.
A nineteenth embodiment, which is divided and provided, will be described.
Although various specific examples can be considered for such a case,
Fig. 63 shows an example in which only logarithmic compression of a signal is performed by an image pickup device.
Is shown.   A logarithmic amplifier can be used to perform logarithmic compression.
However, logarithmic compression is performed by the image sensor itself using the method described above.
Performing is advantageous in terms of S / N. Taken in Fig. 63
The image sensor 401 has an image sensor 404 that performs logarithmic compression within the device.
Then, the output is input to the signal processing circuit 405.   Reference numeral 414 denotes a drive circuit for driving the image sensor. Explained first
As described above, the drive circuit 414 overflows the image sensor 404.
It can change the shape of the pulse applied to the low drain.
Changes the gain and dynamic range.   When performing this control automatically, as described above,
A logY signal is required. In the configuration of Fig. 63, the logY signal
Is output to the playback device side, so the
Automatically gains practically enough by using logG signal
Control and automatic dynamic range control
You.   The logarithmically compressed signal output from the image sensor is
For example, a magnetic disk is used as a recording medium.
When using, the FM modulator 415 in the signal processing circuit 405
FM modulated, amplified by the recording amplifier 416, and
Written to disk 417. Note that the dynamic
If the range is changed, its dynamic level
The signal indicating the image is also recorded on the magnetic disk.   In the reproducing device 402, the data read from the magnetic disk 417 is read.
The amplified signal is amplified by the regenerative amplifier 418 and the FM modulator 419
To return to the original signal. Further, the color separation circuit 420
It is separated into ogR, logG, and logB signals. These signals are inverted
The signal is converted to a linear signal by the logarithmic amplifier 421.
Refer to the dynamic range recorded in
Inverse conversion needs to be performed. As an antilog amplifier,
For example, the configuration of the floating point method as described above
It can be. The R, G, B signals that have become linear signals are
The luminance signal Y and the color difference signals R-Y, B-
Converted to Y. Thereafter, the luminance signal is output by the logarithmic amplifier 423.
Again logarithmically compressed. The compression degree at this time is
May be the same as the value of the compression degree at the time, or the effect of image reproduction
It may be set appropriately while watching.   The color difference signals RY and BY are multiplied by logY / Y by the correction circuit 424.
And a visual correction is made. Received these processes
The signals Y ', (RY)' and (BY) 'are
The signal is converted into an NTSC signal by 5 and output to the monitor.   FIG. 64 shows a twentieth embodiment of the present invention. In this example
An IC memory is used as a recording medium. In the imager 401
The signal output from the image sensor 404 is an A / D converter
It is converted to a digital value by 426. Where
Similarly, a signal compression pulse is sent from the drive circuit 414 to the image sensor 404.
It is preferable that the output has a logarithmic characteristic.
No. In this case, the quantization bit number of the A / D converter 426 is used.
With 8 bits, a signal of 48 dB can be recorded.
When observing only images in the narrower light range on the greige machine
In order to minimize the quantization error in
It is desirable to have about 10 to 12 bits. Digital
The video signal converted to the value is converted to a data compression circuit 427 and a code.
After the data amount is reduced by the conversion circuit 428, the data
Be recorded.   In the reproducing device 402, the signal output from the IC memory 429 is output.
After the signal is restored to the original signal by the data decompression circuit 430,
The signal is separated into logR, logG, and logB signals by the color separation circuit 431.
You. Thereafter, the signal is converted into a linear signal in the signal processing circuit 432.
After the conversion, it is converted into a luminance / color difference signal by matrix operation.
Is replaced. In addition, compression of luminance signal and correction of color difference signal
Is output as a video signal. The above
The processing in the player is performed as digital values from IC memory.
Since the signal is obtained, it is processed by digital operation as it is
, But even in that case,
The input video signal is an analog signal by a D / A converter
It is said.   For example, in the apparatus shown in FIG.
1: 300 alternately captures the imaged signal one frame at a time
It can also be displayed on a monitor.   When logarithmic compression is performed, logY / Y is converted to the color difference signal log (R-
Y), log (BY) is multiplied to prevent color shift
However, color emphasis must be performed so that this ratio can be changed.
Can also be. [The invention's effect]   As described above, according to the present invention, logarithmic compression and the like are performed.
Means, the desired dynamic lens
You can get a color image.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 to FIG. 9 relate to a first embodiment of the present invention, and FIG.
FIGS. 2 and 3 are block diagrams showing the overall configuration, FIG. 4 is a diagram showing the characteristics of the image sensor, and FIG. 5 is a simplified diagram showing the outputs of the two image sensors. 6 shows the internal configuration of the logarithmic amplifier, FIG. 7 shows the operation of the logarithmic amplifier shown in FIG. 6, and FIG. 8 shows the operation of the antilogarithmic conversion circuit. FIG. 9 is a diagram showing the internal configuration of an antilog amplifier, FIG. 10 is a block diagram of a color signal processing circuit in a second embodiment of the present invention, and FIG.
FIG. 11 is a block diagram showing a standard deviation generating circuit having a configuration different from that shown in FIG. 10, FIG. 12 is an explanatory diagram of the operation of the gamma correction circuit in the second embodiment, and FIG. FIG. 14 is a block diagram showing a configuration of a portion for setting a window width with reference to a maximum level of an input signal, and FIG.
FIG. 15 is an explanatory diagram showing a complementary color filter according to the third embodiment of the present invention, FIG. 16 is a configuration diagram of a color signal processing circuit according to the third embodiment of the present invention, and FIG. 17 is a fourth embodiment of the present invention. FIG. 18 is a configuration diagram of a color signal processing circuit. FIG. 18 is an explanatory diagram showing a configuration of a solid-state imaging device according to a fifth embodiment of the present invention.
19 is an explanatory diagram for changing the depth of the potential well in the light receiving area of the solid-state imaging device of FIG. 18 with the exposure time to obtain a logarithmic compression characteristic, and FIG. 20 shows the depth of the potential well of FIG. FIG. 21 is a graph showing characteristics when the logarithmic compression characteristic is set, FIG. 21 is a characteristic diagram showing input / output of the solid-state imaging device when the logarithmic compression characteristic is set, and FIG. 22 is a waveform of a control signal applied to the overflow drain gate. FIG. 23 is a block diagram showing a configuration of a control signal generation circuit applied to an overflow drain gate, FIG. 24 is an explanatory diagram of an operation for generating a control signal applied to an overflow drain gate, and FIG. 25 is a fifth embodiment. FIG. 26 is a configuration diagram of a color video processing unit according to the sixth embodiment of the present invention, FIG. 27 is a timing chart for explaining the operation of the sixth embodiment, and FIG. 29 is a timing chart for explaining an operation for generating a control signal of an accumulation potential in the sixth embodiment, FIG. 29 is an explanatory diagram showing a solid-state imaging device in the seventh embodiment of the present invention, and FIG. 30 is a solid-state image sensor shown in FIG. FIG. 31 is an explanatory diagram of the operation of compressing the logarithm in the element by the image pickup element, FIG. 31 is a block diagram showing the configuration of a digital control signal generation circuit in the eighth embodiment of the present invention, and FIG. 32 is the control signal generation circuit of FIG. FIG. 33 is a waveform diagram of a control signal generated by the circuit, FIG. 33 is an explanatory diagram of a method for compressing the logarithm in the element in the ninth embodiment of the present invention, FIG. 34 is an operational explanatory diagram of the ninth embodiment, and FIG. FIG. 36 is a diagram showing a structure and an equivalent circuit of one pixel in a tenth embodiment of the present invention. FIG. 36 is a configuration diagram showing a solid-state image pickup device which performs logarithmic compression in the device in the tenth embodiment, and FIG. 37 is a tenth embodiment. 38 is a timing chart for explaining the operation of the example. Waveform diagram of the control signal pulse applied to Torejisuta, the
FIG. 39 is a diagram showing a BBD constituting an analog shift register in the tenth embodiment, FIG. 40 is a diagram showing a structure and an equivalent circuit of one pixel of the CMD in the eleventh embodiment of the present invention, and FIG. FIG. 42 is an equivalent circuit diagram of one pixel of the AMI in the twelfth embodiment, FIG. 42 is a configuration diagram of an intra-element logarithmic compression imaging device using the AMI, and FIG. FIG. 44 is a configuration diagram of a processing unit, FIG. 44 is a configuration diagram of a frame interline transfer type solid-state imaging device used in the thirteenth embodiment of the present invention, and FIG. 45 is a cross-sectional structure and a potential distribution of one pixel portion of the thirteenth embodiment. FIG. 46 is a waveform diagram of a control signal applied to the transfer gate, and FIG.
FIG. 48 is an explanatory diagram of the operation in the thirteenth embodiment.
FIG. 49 is a waveform diagram when a control signal is applied in a plurality of times in order to increase the saturation charge amount, FIG. 50 is an operation explanatory diagram when the control signal of FIG. 49 is used, and FIG. 51 is a saturation charge diagram. FIG. 52 is a diagram showing output characteristics with respect to input light intensity when the amount is increased.
FIG. 53 is a configuration diagram of an image processing unit capable of responding to both logarithmic and exponential characteristic imaging signals. FIG. 53 is a diagram showing a cross-sectional structure and potential distribution near a MOS photodiode in a fourteenth embodiment of the present invention. FIG. 55 is a waveform diagram of a control signal applied to the MOS photodiode, FIG. 55 is an operation explanatory diagram, and FIG.
FIG. 56 is a configuration diagram of a field sequential type electronic scope apparatus according to the fifteenth embodiment of the present invention, FIG. 57 is an explanatory view showing an RGB rotation filter when performing field sequential illumination, and FIG. 58 is a frame memory in a light mode. FIG. 59 is a schematic diagram when the present invention is applied to an electronic camera, and FIG. 60 is a sixteenth embodiment of the present invention in which a logarithmic imaging processing function is built in the imaging device.
FIG. 61 is a block diagram of a seventeenth embodiment of the present invention having a built-in logarithmic imaging processing function on the reproducing apparatus side, FIG. 61 is a block diagram of an eighteenth embodiment of the present invention, FIG. Is a block diagram of a nineteenth embodiment of the present invention in which only an logarithmic compression is performed by an image pickup device, and FIG. 64 is a block diagram of a twentieth embodiment of the present invention using an IC memory as a recording medium. DESCRIPTION OF SYMBOLS 1 ... Image forming lens, 2 ... Half mirror, 3a, 3b ... Image sensor, 7-12, 45-47 ... Logarithmic amplifier, 13-15, 29-31
…… Amplifier, 16-19,36-38 …… Window circuit, 20,26
~ 28,48 ~ 50,51,72 ... Adder, 33 ~ 35,53,54,61 ~ 63 ...
... Subtractor, 40 ... Average calculation circuit, 41-43,55,56 ... Anti-log amplifier, 44 ... Matrix conversion circuit, 52,57,58 ...
Multiplier, 59: inverse matrix conversion circuit, 60: NTSC conversion section, 68 to 71: clip circuit.

Claims (1)

(57) [Claims] A color image pickup device for picking up a color image for an observer to observe, comprising: a color image signal input means for inputting a subject as a color image signal; and extracting a signal component representing brightness from the input color image signal. Luminance signal extracting means; non-linear compressing means for non-linearly compressing the signal component representing the brightness extracted by the luminance signal extracting means; and a signal component representing the brightness of the input color video signal being non-linearly compressed. In this state, the saturation of the color video signal at the time of input and the saturation after compression are newly set so as to be substantially constant on the observer's eyes by using the compression degree set by the non-linear compression means. A color imaging apparatus comprising: a color signal synthesizing unit that synthesizes a color signal.