JP2005176061A - Solid-state image pickup device, image input device and noise elimination method for solid-state image pickup device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that the detection accuracy of fixed pattern noise components is low since measuring signals of fixed pattern noise reflecting the dispersion in a column direction of a transistor for read of pixels can not be outputted without being affected by a dark current. <P>SOLUTION: Each of a valid pixel 210 and an invalid pixel 220 inside a pixel part 2 is provided with a first transistor 211 or 221 for read and a second transistor 212 or 222 for reset. A diode PD is provided in each valid pixel 210 and the diode PD is omitted in the invalid pixel 220. The second transistor 222 is controlled in the invalid pixel 220 to bias the gate of the first transistor 221 and signals of each vertical signal line 23 at the time are outputted from the pixel part 2 as the measuring signals of the fixed pattern noise components due to the dispersion of the electrical characteristics of the route of the signals. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a solid-state imaging device that photoelectrically converts input light by a diode in a pixel to generate pixel charge, and amplifies and reads the pixel charge by a transistor in the pixel, a noise removal method thereof, and the solid-state imaging The present invention relates to an image input device including the device.

  A CMOS image sensor is known as a typical solid-state imaging device having an active pixel structure that amplifies and reads out pixel charges in the pixel. A CMOS image sensor is generally considered to be less sensitive and less susceptible to noise than a CCD image sensor. However, along with recent remarkable progress in miniaturization technology of MOS transistors and wiring in CMOS integrated circuits, pixel miniaturization and high sensitivity have progressed, and because of the amplification function, a signal output (pixel signal) that is strong against noise can be obtained. It came to be able to. Further, as a technique for reducing variations in pixel signals, the progress of a CDS (correlated double sampling) circuit, among other signal processing circuits, has attracted attention as a device that can replace a CCD image sensor. It is a factor.

  Noise superimposed on the pixel signal of a CMOS image sensor is random noise that occurs randomly in time and space (that is, in a two-dimensional output screen), and vertical stripes, horizontal stripes, or screen unevenness in the same place on the output screen. It can be roughly classified into fixed pattern noise generated as a fixed pattern. Random noise appears as uneven noise when passing through ground glass across the entire screen, and mainly occurs in pixel diodes, amplification points, other amplification points, such as light shot noise and thermal noise, and the environment. It changes easily with temperature. On the other hand, fixed pattern noise is caused by an imbalance in electrical characteristics due to a difference in threshold voltage of the MOS transistor of the pixel, other circuits (for example, a capacitor of a CDS circuit provided for each vertical signal line) or wiring. Many things are produced.

  When light of a certain intensity is input to the entire pixel portion, a signal obtained from a vertical signal line that is provided for each pixel column and outputs a pixel signal to the outside of the pixel portion should have a certain level. However, since the fixed pattern noise component described above is superimposed on this signal, a phenomenon occurs in which the signal level varies for each pixel column, that is, for each vertical signal line.

  In order to remove fixed pattern noise components that differ for each pixel column, a pixel signal (hereinafter referred to as a measurement signal) is output in a state where light is blocked or photoelectric conversion itself cannot be performed, and is fixed from the measurement signal. The pattern noise component is detected and held, and when the pixel signal at the time of imaging (hereinafter, effective pixel signal) is output, the detected fixed pattern noise component is subtracted from the effective pixel signal to obtain the noise component. A technique for canceling is known (see, for example, Patent Document 1).

  In the technique described in Patent Document 1, a dummy line to which a photosensitive pixel capable of photoelectric conversion is not connected is provided in the pixel unit, and a signal output from the dummy line is used as noise component data, and a fixed pattern noise component Detection is in progress. The problem that this technology is trying to solve is that if the signal from the optical black part where the pixels of the same configuration are shielded from light is used as noise component data, there will be light leakage, and even if it is completely shielded from dark current Is superimposed on the signal, and it is difficult to accurately detect the fixed pattern noise component from such a signal. In the technique described in Patent Document 1, a dummy line is provided and a signal is read from the dummy line so as not to be affected by the dark current component.

However, although the output signal from the dummy line, what signals are output for the configuration is not clear it is unknown.
In addition, although a reading transistor exists in a pixel, the threshold value varies depending on the conditions of the manufacturing process and the position of the transistor. Therefore, the threshold difference is small in the near transistor, but the threshold difference is large in the position separated in the horizontal direction (row direction). Therefore, the threshold difference in the horizontal direction is one of the factors that determine fixed pattern noise. However, if a dummy line is provided as in Patent Document 1 and a signal is read out from the dummy line, the threshold difference of the pixel readout transistor cannot be reflected in the measurement signal of the fixed pattern noise.
Japanese Patent Laid-Open No. 06-189200

  The problem to be solved is that it is not affected by dark current, and because it cannot output a fixed pattern noise measurement signal that reflects variations in the row direction of the pixel readout transistors, the detection accuracy of fixed pattern noise components Is low.

  A solid-state imaging device according to the present invention is a solid-state imaging device in which a vertical signal line is connected to each pixel column in a pixel unit, and effective pixels and ineffective pixels are arranged in a row unit in the pixel unit. A first transistor for supplying a signal corresponding to the gate potential to a corresponding vertical signal line for each of the non-valid pixels; a second transistor for resetting the gate potential of the first transistor in the valid pixels; Each of the effective pixels is provided with a diode that generates a pixel charge corresponding to input light and supplies it to the gate of the first transistor, the diode is omitted in the ineffective pixel, and the second transistor in the ineffective pixel is provided. The gate of the first transistor is controlled to bias the signal of each vertical signal line at this time to a fixed pattern noise caused by variations in the electrical characteristics of the signal path. Outputted from the pixel section as a measurement signal components.

  An image input device according to the present invention is an image input device including a solid-state imaging device in which a vertical signal line is connected to each pixel column in the pixel unit, and the pixel unit of the solid-state imaging device includes an effective pixel and a non-effective pixel in a row unit. An effective pixel is arranged, and each of the effective pixel and the non-effective pixel has a first transistor for supplying a signal corresponding to the gate potential to the corresponding vertical signal line, and a gate potential of the first transistor in the effective pixel. A diode for generating a pixel charge corresponding to input light and supplying the pixel charge to the gate of the first transistor is provided for each effective pixel. The gate of the first transistor is biased by controlling the second transistor in the ineffective pixel, and the signal of each vertical signal line at this time is changed according to the electric characteristic of the path of the signal. Outputted from the pixel section as a measurement signal having a fixed pattern noise component caused by the month.

  In the noise removal method for a solid-state imaging device according to the present invention, a vertical signal line is connected to each pixel column in the pixel unit, and effective pixels and ineffective pixels are arranged in units of rows in the pixel unit. Each includes a first transistor for supplying a signal corresponding to the gate potential to the corresponding vertical signal line, and a second transistor for resetting the gate potential of the first transistor in the effective pixel, A diode for generating a pixel charge corresponding to input light and supplying the pixel charge to the gate of the first transistor is provided for each effective pixel, and in the solid-state imaging device in which the diode is omitted for the non-effective pixel, the signal path for each pixel column A noise removal method for removing a fixed pattern noise component caused by variation in electrical characteristics from an effective pixel signal, wherein a second transistor is connected to an ineffective pixel. From the signal output step of biasing the gate of the first transistor and outputting the signal of each vertical signal line at this time as a measurement signal of the fixed pattern noise component, and the measurement signal of the plurality of pixel columns, the fixed pattern A level detection step that detects the level of the noise component for each vertical signal line, and the fixed pattern noise component level is subtracted from the effective pixel signal output from the pixel unit according to the input light during imaging to remove the fixed pattern noise. And a noise removing step.

According to the present invention, when outputting the measurement signal of the fixed pattern noise, the second transistor of the ineffective pixel is turned on, and the gate of the first transistor is biased through the second transistor. The bias at this time is set according to the average value of the pixel charge amount generated by the effective pixel when, for example, images of various brightness are input. In a conductive state corresponding to the gate bias, the first transistor of the ineffective pixel is turned on, and the potential of the corresponding vertical signal line changes. This potential change passes through the vertical signal line as a fixed pattern noise measurement signal and is output to the outside of the pixel portion via a circuit or the like provided in the vertical signal line. The measurement signal is output simultaneously from all the vertical signal lines connected to each pixel column.
Thereafter, the level of the fixed pattern noise component is detected for each vertical signal line from the measurement signals output from the plurality of pixel columns, and the level of the fixed pattern noise component is output from the pixel unit according to the input light during imaging. Is subtracted from the effective pixel signal. As a result, fixed pattern noise is removed from the effective pixel signal.

  In the present invention, preferably, measurement signals sequentially output from the vertical signal lines are averaged for each vertical signal line. At this time, the amount of ineffective pixel data to be averaged for each vertical signal line (that is, the number of rows of ineffective pixels) is determined depending on how much random noise component is to be within the total noise component in the measurement signal. It has been. That is, when it is desired to suppress the random noise component uniformly superimposed on the measurement signal to 1 / M times or less of the total noise component, the number of rows of ineffective pixels in the pixel unit is set to M square or more.

According to the present invention, in the non-effective pixel that outputs the measurement signal, the diode that is connected to the gate of the first transistor in the normal effective pixel is omitted. Are excluded from the measurement signal. Further, the gate bias state of the first transistor is converted into the potential of the vertical signal line through the first transistor, and this becomes a measurement signal. Therefore, the measurement signals output from all the vertical signal lines connected to each pixel column change reflecting the variation in the pixel row of the threshold value of the first transistor. The detection accuracy of the fixed pattern noise component using the signal can be improved.
Further, it is possible to effectively reduce random noise with respect to all noise components of the measurement signal, and accordingly, it is possible to improve the detection accuracy of the fixed pattern noise component level.

Hereinafter, an embodiment of the present invention will be described using a CMOS image sensor which is a kind of solid-state imaging device as an example.
As for the CMOS image sensor according to the embodiment, FIG. 1 shows an overall configuration diagram, and FIG. 2 shows a circuit diagram of a pixel portion.
A CMOS image sensor 1 shown in FIG. 1 includes a pixel portion 2 in which pixels are arranged in a matrix, a column processing circuit 3, a vertical driving circuit 4, a horizontal driving circuit 5, a timing control circuit 6, and a signal processing circuit 7. Yes.
Although not shown, the column processing circuit 3 includes a CDS (Correlated Double Sampling) circuit, a current source transistor, a sampling circuit driven in a dot sequential manner by the horizontal drive circuit 5, and the like.

  The pixel unit 2 includes an effective pixel unit 21 and a non-effective pixel unit 22. The effective pixel unit 21 is an arrangement region of effective pixels having a diode that photoelectrically converts input light to generate charges, and the non-effective pixel unit 22 is an arrangement region of non-effective pixels that the diode does not have.

  A number of effective pixels 210 corresponding to the imaging resolution of the CMOS image sensor are arranged in the effective pixel unit 21 in the row direction (horizontal direction) and the column direction (vertical direction). In FIG. 2, only three effective pixels 210 belonging to the same row are shown for simplification, and only the first row and the last row are shown. Each effective pixel 210 constituting the effective pixel unit 21 is a three-transistor type in the example of FIG. 2, and includes a photodiode PD that photoelectrically converts input light and three transistors 211 to 213. The three transistors are a read transistor (first transistor) 211 that amplifies a pixel signal corresponding to the accumulated charge transferred to the node ND and outputs the amplified signal to the vertical signal line 23, and a reset voltage supply line from the floating state of the node ND. The reset transistor (second transistor) 212 that charges the node ND with a reset voltage and resets the amount of charge, and the node ND that is again in a floating state after reset are connected to the node ND. It consists of a transfer transistor 213 that transfers stored charges (usually electrons).

A power supply voltage supply line 24 is connected to the drain of the read transistor 211.
A reset control line 25 common to the pixels in the same row is connected to the gate of the reset transistor 212, and a reset voltage supply line 26 is connected to the drain thereof. Further, a transfer control line 27 common to the pixels in the same row is connected to the gate of the transfer transistor 213.
A vertical drive circuit 4 that supplies various signals or voltages to the power supply voltage supply line 24 and the reset voltage supply ship 26 and to the reset control line 25 and the transfer control line 27 is connected.

  A column processing circuit 3 and a horizontal drive circuit 5 are provided for parallelly processing the pixel signals read to the vertical signal line 23 to remove noise and converting the pixel signals into time-series signals. The vertical drive circuit 4, the column processing circuit 3, and the horizontal drive circuit 5 operate under the control of the timing control circuit 6 shown in FIG.

  The column processing circuit 3 includes a current source, a CDS circuit, and a sampling switch provided for each vertical signal line 23, although not particularly shown. The current source is provided to supply a constant current to the read transistor 211. In addition, the CDS circuit takes the difference between the voltage sampled and held at the black level of the pixel signal of the effective pixel unit 21 and the voltage sampled and held at the pixel signal level corresponding to the accumulated charge. This circuit cancels the noise component superimposed on the. The pixel signal level after this noise removal is held at the output of the CDS circuit, and further sampled in a dot-sequential manner by a sampling switch that is sequentially turned on by a pulse supplied from the horizontal drive circuit 5 and discharged to the internal bus. The The signal processing circuit 7 performs predetermined processing on the dot sequential pixel signal discharged to the internal bus.

  As shown in FIG. 2, a large number of ineffective pixels 220 are arranged in the ineffective pixel portion 22 in the horizontal direction and the vertical direction. The number of pixels in the same row is the same in the effective pixel portion 21 and the ineffective pixel portion 22. In addition, the number of pixel rows (number of lines) of the ineffective pixel portion 22 is the number of lines necessary to make the ratio of the random noise component to the total noise component equal to or less than a certain value, as will be described later. In FIG. 2, only three ineffective pixels 220 belonging to the same row are shown for simplification, and only the first row and the last row are shown.

  The non-effective pixel 220 has a configuration in which the photodiode PD and the transfer transistor 213 are omitted from the effective pixel 210. The non-effective pixel 220 has the same size and pattern as the readout transistor 211 of the effective pixel 210 and the same size and pattern as the first transistor 221 formed in a lump under the same conditions and the reset transistor 212 of the effective pixel 210. And the second transistor 222 formed collectively under the same conditions. In the ineffective pixel 220, no transfer transistor or photodiode is connected to the connection midpoint between the first and second transistors 221 and 222. Note that a transfer transistor or a photodiode may be formed, but even in that case, the pixel readout operation is not performed because these elements are not electrically connected to the connection midpoint of the first and second transistors 221 and 222. It is made so that no contribution can be made.

The drain of the first transistor 221 is connected to the power supply voltage supply line 24 and the source thereof is connected to the corresponding vertical signal line 23. The source of the second transistor 222 is connected to the gate of the first transistor 221, the drain thereof is connected to the voltage supply line 28, and the gate thereof is connected to the control line 29.
The voltage supply line 28 is a wiring layer having the same pattern as the reset voltage supply line 26 of the effective pixel unit 21 and formed in a lump under the same conditions. However, the supply voltage is the same as the reset voltage. Alternatively, an arbitrarily changed voltage may be used. Further, the control line 29 is a wiring layer having the same pattern as the reset control line 25 of the effective pixel portion 21 and formed in a lump under the same conditions, but the supply voltage is the same as the reset control voltage. The voltage may be arbitrarily changed.
In FIG. 2, the non-effective pixel unit 22 is disposed only on one side in the vertical direction of the effective pixel unit 21, but the non-effective pixel unit 22 may be disposed on both sides of the effective pixel unit 21 in the vertical direction.
The non-effective pixel unit 22 is used when measuring the level of fixed pattern noise. The measuring method will be described later.

  FIG. 3 shows a part of the circuit provided in the signal processing circuit 7. The signal processing circuit 7 includes two amplifiers 71 and 72, an AD converter 73, a noise component extraction unit 74, a subtraction unit 75 that subtracts the level of the fixed pattern noise detected by the noise component extraction unit 74 from the input pixel signal, and Have The noise component extraction unit 74 stores a memory 741 that stores a signal (measurement signal) read from the ineffective pixel 220 for each vertical signal line, and random noise that averages the measurement signal for each vertical signal line and suppresses random noise components. The suppression unit 743 includes a level detection unit 742 that detects the level of the fixed pattern noise component in units of a predetermined number of frames for each vertical signal line from the measurement signal in which random noise is suppressed.

  An analog signal processing circuit (not shown) is provided in the signal processing circuit 7. The analog signal processing circuit has functions of inversion amplification, clamp level control, and filtering to remove high frequency components, and the first-stage amplifier 71 shown in FIG. 3 represents an amplifier in the analog signal processing circuit.

Next, the operation of the CMOS image sensor configured as described above will be described.
This CMOS image sensor has a noise measurement signal readout mode in which only a non-valid pixel portion 22 is accessed to output measurement signals for a predetermined number of frames. This noise measurement signal readout mode is executed prior to the normal pixel signal output mode. The reading of the measurement signal may be performed in one frame, but it is preferable to perform the measurement signal in several frames in order to further improve the measurement accuracy.

  In the noise measurement signal readout mode, the vertical drive circuit 4 operates only the ineffective pixel unit 22. That is, a predetermined voltage is supplied to the voltage supply line 28 in a state where the power supply voltage is applied to the power supply voltage supply line 24 of the non-effective pixel unit 22, and then the predetermined voltage is supplied to the control line 29.

With respect to each voltage supplied to the voltage supply line 28 and the control line 29, the second transistor 222 is turned on, and at this time, the amount of charge supplied to the gate of the first transistor 221 is generated in the effective pixel unit 21. It is desirable to control to a value corresponding to the average charge amount of the signal charge. If the gate bias of the first transistor 221 is set regardless of the charge of the effective pixel signal, all elements that determine fixed pattern noise (such as the electrical characteristics of the circuit elements for each column) have linear characteristics with respect to the signal level. If it is, there is no problem, but in reality it is not. In the present embodiment, the measurement signal level can be set near the average level of the actual effective image signal. In this case, a more accurate measurement signal is output from the non-effective pixel unit 22.
This measurement signal readout operation is sequentially performed for each pixel row (line) of the ineffective pixel unit 22.

  The measurement signal sent through each vertical signal line 23 is processed for each vertical signal line by the column processing unit 3, and is discharged as a serial measurement signal to the internal bus under the control of the horizontal drive circuit 5. 7 is sent. At this time, fixed pattern noise that changes for each column is superimposed on the measurement signal due to the difference in electrical characteristics of the vertical signal line path including the column processing unit 3. Random noise is also superimposed on the measurement signal.

  In the signal processing unit 7, it is converted into a digital measurement signal by the AD converter 73 after predetermined analog signal processing. Further, after being amplified by the amplifier 72 as necessary, it is input to the noise component extraction unit 74. In the noise component extraction unit 74, first, random noise component suppression processing is executed. In this processing, when the measurement signals for a predetermined number of frames are accumulated in the memory 741, the random noise components are suppressed by reading the measurement signals from the memory 741 and averaging them for each vertical signal line.

Here, the number of rows (number of lines) of the ineffective pixel unit 22 used for the random noise component suppression processing will be described.
When the random noise suppression unit 743 described above suppresses the random noise component uniformly superimposed on the measurement signal to 1 / M times or less of the total noise component, the number of lines of the ineffective pixel unit 22 is 2 of M. It is set to more than the power. As a rule of thumb, random noise can be suppressed to a level that cannot be recognized on the display screen if the random noise component is ¼ or less of the total noise component. Hereinafter, the case where a random noise component is suppressed to 1/4 or less will be described.

As is known from probability statistics, let us consider a case where a sample is arbitrarily extracted from a distribution of a certain population (mean value: μ, variance: σ ² ). The variance σ _x ² of the distribution of the average value μ (x) of some samples can be obtained from the following equation with respect to the number of samples n.

σ _x ² [μ (x _i )] = σ _x ² [(1 / n) [x ₁ + x ₂ +... + x _n ]]
= (1 / n ² ) {σ _x ² (x ₁ ) + σ _x ² (x ₂ ) +... + Σ _x ² (x _n )}
= (1 / n ² ) {σ ₁ ² + σ ₂ ² +... + Σ _n ² }
= (1 / n ² ) · nσ ²
= Σ ² / n

Therefore, the standard deviation σ _x is σ / (n) ^1/2 . Since the number of samples n is the number of lines of the ineffective pixel portion 22, the number of lines n needs to be 16 or more in order to reduce the random noise component to ¼ or less.
When the number of lines n is less than 16 when the fixed pattern noise component is detected in a plurality of frames, the random noise can be reduced to ¼ or less. In other words, for example, when the measurement is performed in 4 frames, if there are 4 lines or more per frame, a total of 16 lines of measurement signals can be obtained. As a result, random noise can be reduced to 1/4 or less. However, when the gain of the camera using the CMOS image sensor is adjusted, for example, when a flash is applied, there is no time to capture a plurality of frames, so it is necessary to extract fixed pattern noise for one frame. is there. Even in such a case, in order to detect a fixed pattern noise component from data in which random noise is effectively suppressed, it is desirable to provide 16 lines or more of ineffective pixels 220 for each frame.

The measurement signal in which the random noise is suppressed in this way is stored in the memory 741 again.
Next, the level of the fixed pattern noise component is detected from the measurement signal in which the fixed pattern noise is suppressed. This detection is performed by the level detector 742 reading out the measurement signal from the memory. The method of level detection is arbitrary, but the simplest method can be performed by averaging the measurement signal for each vertical signal line. In the case of averaging, the processing contents are the same as random noise, the level detection unit 742 and the random noise suppression unit 743 are shared, and a measurement signal corresponding to 16 or more ineffective pixels for each vertical signal line. Thus, the ineffective pixel data is averaged only once, and the averaged ineffective pixel data becomes the level of the fixed pattern noise component in which random noise is suppressed.
The level of the detected fixed pattern noise component is stored and held in the memory.

  Note that the level detection method performed by the level detection unit 742 is not limited to averaging as described above, and for example, the median value can be detected as the level of the fixed pattern noise component. This is because when it is assumed that there is some bias in the extracted data distribution, the true value may be detected by using the median value rather than the average value. Note that the level detection unit 742 may be configured to be able to switch whether an average value or a median value is detected by external control.

When the process of the noise measurement signal readout mode is completed, the noise removal process can be performed on the effective pixel signal read out thereafter.
The vertical drive circuit 4 accesses the effective pixel unit 21, pixel signals are read out to all the vertical signal lines for each effective pixel line, and after the column processing, the effective pixel subjected to analog signal processing and AD conversion in the signal processing circuit 7 The signal is sequentially supplied to the “+” input terminal of the subtractor 75. At this time, in synchronization with the supply of the pixel signal, the level signal of the fixed pattern noise component of the corresponding vertical signal line is read from the memory 741 and sequentially supplied to the “−” input terminal of the subtractor 75. Therefore, fixed pattern noise for each vertical signal line is removed or suppressed in the effective pixel signal output from the subtracting unit 75. The effective pixel signal after this subtraction processing is sent to digital signal processing (not shown), and after being subjected to predetermined digital signal processing, it is used for image display on a display.

According to the present invention, the non-effective pixel 220 for measuring the level of the fixed pattern noise component is a pixel from which the photodiode is omitted, and thus is not affected by the defect.
In addition, the non-effective pixel 220 includes a first transistor 221 having the same size and pattern as the readout transistor of the effective pixel 211 and collectively formed under the same conditions. Since the measurement signal is read out to the vertical signal line 23, the variation in the line of the readout transistor is reflected in the measurement signal, and the detection accuracy of the fixed pattern noise is high accordingly.
In addition, the vertical drive circuit 4 controls the voltage supply line 28 and the control line 29, so that the level of the measurement signal output from the non-effective pixel 220 can be arbitrarily adjusted. Therefore, it is possible to detect the level of the fixed pattern noise component suitable for the pixel signal from.
Furthermore, since the number of lines of the ineffective pixel unit 22 is set to a number that can effectively suppress random noise, the influence of random noise is eliminated from the detection level of the fixed pattern noise component, and high-precision fixed pattern noise is reduced. Removal is possible.
In addition to the above, it is also possible to detect fixed pattern noise components from a plurality of frames, and in that case, it is possible to remove fixed pattern noise with higher accuracy.

  Note that the signal processing unit 7 shown in FIG. 3 can be realized as a signal processing IC provided outside the CMOS image sensor. Further, a configuration in which the noise component extraction unit 74 and the subtraction unit 75 are realized on a microcomputer program can be employed.

It is a block diagram which shows the block diagram of the whole CMOS image sensor concerning embodiment of this invention. It is a circuit diagram of the pixel part of the CMOS image sensor concerning embodiment of this invention. It is a figure which shows a part of circuit block provided in the signal processing circuit.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... CMOS image sensor, 2 ... Pixel part, 21 ... Effective pixel part, 210 ... Effective pixel, 211 ... 1st transistor, 212 ... 2nd transistor, 213 ... Transfer transistor, 22 ... Non-effective pixel part, 220 ... Ineffective pixels, 221... First transistor, 222... Second transistor, 23... Vertical signal line, 24... Power supply voltage supply line, 25... Reset control line, 26. 28 ... voltage supply line, 29 ... control line, 3 ... column processing circuit, 4 ... vertical drive circuit, 5 ... horizontal drive circuit, 6 ... timing control circuit, 7 ... signal processing circuit, 71, 72 ... amplifier, 73 ... AD Converter, 74 ... Noise component extraction unit, 741 ... Memory, 742 ... Level detection unit, 743 ... Random noise suppression unit, 75 ... Subtraction unit, PD ... Photodiode, ND Node

Claims (10)

A solid-state imaging device in which a vertical signal line is connected to each pixel column in the pixel unit,
Effective pixels and ineffective pixels are arranged in units of rows in the pixel portion,
A first transistor for supplying a signal corresponding to the gate potential to the corresponding vertical signal line for each of the effective pixel and the non-effective pixel, and a second transistor for resetting the gate potential of the first transistor in the effective pixel And having a transistor
A diode that generates pixel charge according to input light and supplies it to the gate of the first transistor is provided for each effective pixel,
For non-effective pixels, the diode is omitted,
The gate of the first transistor is biased by controlling the second transistor in the non-effective pixel, and the signal of each vertical signal line at this time is fixed pattern noise component caused by variation in the electrical characteristics of the signal path A solid-state imaging device that outputs from the pixel unit as a measurement signal. The solid-state imaging device according to claim 1, wherein a bias supplied to the gate of the first transistor when the measurement signal is output is set corresponding to an average charge amount of signal charges generated in the effective image. . A level detection unit that detects the level of the fixed pattern noise component for each vertical signal line from measurement signals of ineffective pixels in a plurality of rows;
A storage unit for storing the level of the fixed pattern noise component;
The subtractor for subtracting the level of the fixed pattern noise component stored in the storage unit from the effective pixel signal when an effective pixel signal is output from the pixel unit according to the input light during imaging. Solid-state imaging device. The solid-state imaging device according to claim 3, wherein the level detection unit detects the level of the pattern noise component for each vertical signal line from measurement signals of ineffective pixels for a plurality of frames. The solid-state imaging device according to claim 1, further comprising a noise measurement signal reading mode in which a measurement signal for a predetermined number of frames is output by accessing only the row of the invalid pixels. Random noise suppression that suppresses the random noise component uniformly superimposed on the measurement signal to 1 / M times or less of the total noise component by averaging the measurement signals sequentially output from each vertical signal line for each vertical signal line Further comprising
The solid-state imaging device according to claim 1, wherein the number of rows of ineffective pixels in the pixel unit is set to a square of M or more. An image input device including a solid-state imaging device to which a vertical signal line is connected for each pixel column in a pixel unit,
Effective pixels and ineffective pixels are arranged in units of rows in the pixel portion of the solid-state imaging device,
A first transistor for supplying a signal corresponding to the gate potential to the corresponding vertical signal line for each of the effective pixel and the non-effective pixel, and a second transistor for resetting the gate potential of the first transistor in the effective pixel And having a transistor
A diode that generates pixel charge according to input light and supplies it to the gate of the first transistor is provided for each effective pixel,
For non-effective pixels, the diode is omitted,
The gate of the first transistor is biased by controlling the second transistor in the non-effective pixel, and the signal of each vertical signal line at this time is fixed pattern noise component caused by variation in the electrical characteristics of the signal path An image input device that outputs from the pixel unit as a measurement signal. Random noise suppression that suppresses the random noise component uniformly superimposed on the measurement signal to 1 / M times or less of the total noise component by averaging the measurement signals sequentially output from each vertical signal line for each vertical signal line Further comprising
The image input device according to claim 7, wherein the number of rows of ineffective pixels in the pixel unit is set to a square of M or more. A vertical signal line is connected to each pixel column in the pixel unit, and effective pixels and ineffective pixels are arranged in units of rows in the pixel unit. A vertical signal corresponding to a gate potential is applied to each of the effective pixel and the ineffective pixel. A first transistor for supplying to the signal line; and a second transistor for resetting the gate potential of the first transistor in the effective pixel. In a solid-state imaging device in which a diode to be supplied to the gate of the transistor is provided for each effective pixel and the diode is omitted for the non-effective pixel, a fixed pattern noise component caused by variations in the electrical characteristics of the signal path for each pixel column A noise removal method for removing from an effective pixel signal,
A signal output step of biasing the gate of the first transistor by controlling the second transistor in the ineffective pixel, and outputting a signal of each vertical signal line at this time as a measurement signal of a fixed pattern noise component;
A level detection step for detecting a level of a fixed pattern noise component for each vertical signal line from measurement signals of a plurality of pixel columns;
A noise removing method for a solid-state imaging device, comprising: a noise removing step of subtracting a level of a fixed pattern noise component from an effective pixel signal output from a pixel unit according to input light during imaging to remove the fixed pattern noise. Random noise removal that suppresses the random noise component uniformly superimposed on the measurement signal to 1 / M times or less of the total noise component by averaging the measurement signal output sequentially from each vertical signal line for each vertical signal line And further comprising steps
10. The solid-state imaging device according to claim 9, wherein the number of rows of ineffective pixels that sequentially output the measurement signal to each vertical signal line in the signal output step is set to a square of M or more when removing the random noise. Noise removal method.

Images