JP2008067060A - Imaging device - Google Patents

Abstract

[PROBLEMS] To correct horizontal pixel noise by taking into account that the influence on the pixel output due to power fluctuation or the like differs between an OB portion and an effective portion.
An imaging region having a light-shielded photoelectric conversion unit and a non-light-shielded photoelectric conversion unit;
Correction means for obtaining a correction value by correcting the signal from the light-shielded photoelectric conversion unit by using a correction coefficient, and correcting the signal from the corresponding non-light-shielded photoelectric conversion unit by the correction value An imaging device comprising:
[Selection] Figure 3

Description

  The present invention relates to correction of an imaging apparatus.

  In recent years, CMOS image sensors have been used in digital single-lens reflex cameras and video cameras. In the CMOS image sensor, as the CCD image sensor increases in number of pixels and one pixel becomes smaller, the optical signal also becomes smaller, and in order not to deteriorate the S / N, noise reduction is increasingly necessary.

  FIG. 6 shows a circuit of one pixel 60 of the CMOS image sensor. A photodiode 61 (hereinafter referred to as PD) receives a light image formed by a photographing lens (not shown) and generates and accumulates charges. Reference numeral 62 denotes a transfer switch (hereinafter referred to as TX), which is composed of a MOS transistor. Reference numeral 64 denotes a floating diffusion (hereinafter referred to as FD), which is a capacitor. The electric charge accumulated in the PD 61 is transferred to the FD 64 by the TX 62, and the electric charge is converted into a voltage, which is output from the amplifier 65 by the source follower. A row selection switch 66 outputs a pixel output to the vertical line 67. A reset switch 63 resets the potential of the FD 64.

  FIG. 7 shows the overall layout of the CMOS image sensor. Reference numeral 72 denotes a VOB which is composed of light-shielded pixels and is used to detect a black level, and is used to correct an output offset variation due to a dark current component or a temperature variation. Reference numeral 71 denotes an HOB, which is composed of light-shielded pixels like the VOB and is used for correcting the dark shading component in the vertical direction. As a cause of the dark shading component, there is dark current shading. As a characteristic of the CMOS image pickup device, there is one due to voltage shading due to the impedance of the power supply line. Normally, dark shading correction in the vertical direction is gentle, and if there is a scratched pixel, it causes a line scratch, so that the output of a plurality of rows is corrected by applying a low-pass filter.

Reference numeral 73 denotes an effective pixel area, which obtains an image formed by a photographing lens (not shown).
Reference numeral 75 denotes an SN circuit that makes a difference between a signal (S signal) from a PD and a noise signal (N signal). This removes noise peculiar to a CMOS image sensor, and thereby, an S / N equivalent to that of a CCD can be obtained.

  A pixel free from noise by holding a signal component S and a noise component N for the pixel output of the row selected by the row selection switch 66 and subtracting the noise component N from the signal component S for each pixel by the output amplifier 74. A signal is output.

  The noise component is obtained by holding the FD 64 reset with a pulse by the reset switch 63 while the TX 62 is turned off as the N signal of the SN circuit 75 via the amplifier 65, the row selection switch 66, and the vertical line 67. . Noise components include FD reset noise, pixel-to-pixel variations in the gate-source voltage of the amplifier 65, and the like.

  The signal component turns on the TX 62 with a pulse to convert the PD charge into a voltage with the FD 64 and holds it as an S signal in the SN circuit 75 as with the noise component. At this time, the noise is added to the reset noise.

The noise component is canceled by subtracting the N signal from the S signal when the amplifier 74 reads the signal.
Thereafter, correction is performed to match the black level of the signal from the amplifier 74 with the reference black level.
Thereafter, by subtracting the offset of the horizontal OB area (the difference between the level of the horizontal OB area and the reference black level), noise with a high frequency can be suppressed.
JP 2006-191449 A

  However, only subtracting the offset of the horizontal OB area has no effect on horizontal line noise.

  The reason why it cannot be canceled in the horizontal OB area is as follows.

  When the noise source is on the power supply, the sensor power supply wiring and the sensor base itself are distorted. The sensor base on the ground side of the FD 64 of the pixel 60 in FIG. 6 and the OB which is mounted on the signal output from the pixel by capacitive coupling from the surrounding wiring and the effective pixel which is not shielded by the wiring are affected by the OB. Because it is different.

  This will be described in more detail with reference to FIG. This is a cross-sectional view of the CMOS sensor, which includes pixel 1, pixel 2, pixel 3, and pixel 4. Pixel 1 and pixel 2 are OBs that are shielded from light by AL3, and pixels 3 and 4 are not shaded and are effective pixels. ML is a microlens that condenses light effectively on the PD. AL3, AL2, and AL1 are wiring layers, and AL3 is also used for light shielding. TX, FD, and PD are the same as those in FIG. Since AL3 is wide for the light-shielded pixel, the FD has an influence different from the effective pixel such as capacitive coupling.

  The degree of influence of noise such as a power source is experimentally that the effective pixel is about 0.4 to 0.8 times the OB pixel. This depends on the layout.

  That is, when the offset amount of the OB portion is subtracted as it is, overcorrection occurs and it seems that there is no effect on the horizontal stripe noise.

  This will be described with reference to FIGS. In each graph, the vertical axis indicates the vertical direction of the sensor, and the horizontal axis indicates the offset of the OB portion of the sensor from the black level. (B) shows the V direction dark shading of the output of the effective portion in the absence of light. The gradual change is due to the circuit impedance and the like, and the fine change shown in (C) and removed by the horizontal OB clamp of the AFE 2 is that the noise of the power supply is placed on the FD 64 of the pixel. (A) is V direction dark shading of the OB part, the solid line shows the same effective part as (B), and the broken line shows the behavior of OB. It can be seen that the gentle shading is the same as the effective portion, but the amplitude of the fine change is larger than that of the effective portion.

  In order to solve the above problems, an imaging region having a light-shielded photoelectric conversion unit and a non-light-shielded photoelectric conversion unit, and correcting a signal from the light-shielded photoelectric conversion unit using a correction coefficient The image pickup apparatus includes: a correction unit that obtains a correction value using the correction value and corrects a signal from the corresponding non-light-shielded photoelectric conversion unit based on the correction value.

  In addition, the imaging region having a light-shielded photoelectric conversion unit and a non-light-shielded photoelectric conversion unit, and a first correction for adjusting a black level of a signal from the non-light-shielded photoelectric conversion unit to a reference black level. Correction is performed by performing a difference process between the signal level of the signal from the light-shielded photoelectric conversion unit and the reference black level, and correcting the signal obtained by the difference process using a correction coefficient. There is provided an imaging apparatus comprising: a second correction unit that obtains a value and corrects a signal from the corresponding non-light-shielded photoelectric conversion unit based on the correction value.

  According to the present invention, it is possible to remove horizontal line noise that could not be removed so far by correcting that the influence on the pixel output due to the fluctuation of the power source is different between the OB portion and the effective portion. . In addition, dark shading can be removed by providing it independently of the dark shading clamp generated by circuit impedance in the sensor.

(Example 1)
FIG. 1 is an overall block diagram of the imaging apparatus.
FIG. 1 shows a CMOS image sensor. 5, 6, and 7, and an image formed by a photographing lens (not shown) is captured. Reference numeral 2 denotes a general analog front end (AFE), and FIG. 21 is a gain control amplifier used for sensitivity adjustment, and 24 is a horizontal OB clamp function, which gradually follows the offset between the OB output of each line and the black level. This is originally intended to correct minute and gentle dark shading, and is set so as not to be affected by noise in the OB portion. Further, the correction here is fed back to the AMP 21, and the correction amount is integrated as the line advances, and therefore follows only a loose change. Reference numeral 23 denotes analog-to-digital conversion (AD), which converts a pixel signal from the CMOS image sensor 1 to a 14-bit digital signal, for example, by applying a gain by adjusting the horizontal OB to the black level by the AMP 21 and the horizontal OB clamp 24. .

  A digital front end (DFE) 3 receives the digital output of each pixel and performs digital processing such as image signal correction and pixel rearrangement.

  Reference numeral 5 denotes an image processing apparatus that performs development processing and displays an image on the display circuit 8 or records it in the recording circuit 9 via the control circuit 6. The recording circuit includes a compact flash (registered trademark) memory. The memory means 4 is used as a working memory at the development stage of the image processing apparatus 5, and is also used as a buffer memory when the imaging means is continuously performed and the developing means is not in time. Reference numeral 7 denotes an operation member that electrically accepts the operation member in the digital camera. Reference numeral 10 denotes a timing generation circuit, which creates various timings for driving the CMOS image sensor 1.

FIG. 3 shows the internal configuration of the DFE 3. Here, only the parts constituting this embodiment are described.
31 averages the pixel output constituting the horizontal OB for each line. A general digital processing circuit is configured, and digital addition averaging is performed by setting the start position and end position of the OB pixel in a register (not shown). The averaging has the effect of reducing the influence of random noise for each pixel.

  32 performs the following operation. The reference black level determined by the system is subtracted from the average value obtained at 31. Thereby, the offset from the digital value which is the reference black level determined by the system is calculated with respect to the average value obtained in 31. Further, a correction amount is newly calculated by multiplying the obtained offset value by a coefficient (0.4 to 0.8) determined by the structure of the CMOS image sensor.

What is being processed here is a detailed change of the residue obtained by removing FIGS. 4A to 4C by the horizontal OB clamp 24. That is, the fine change of the broken line in (A) is determined by the average of 31, and the correction amount equivalent to the fine change of the effective part in (B) is obtained by multiplying the coefficient by 32. 33 is the correction amount obtained in 32 and the pixel output of the effective portion is corrected and subtracted and output. These are done line by line.
Although the processing by hardware has been described above, since it is digital processing, the processing performed by the DFE 3 may be performed by software.

  FIG. 8 shows an operation flowchart. As a circuit block, DFE3 is deleted from FIG. When the present invention is implemented with software, it is difficult to process in real time at the speed of image reading, and therefore software processing is performed on image data once stored in the memory circuit 4. The software indicates the operation of the control circuit 6, and reads / writes the image data in the memory circuit 4 via the image processing device 5.

  In step 1, the CMOS image sensor 1 is exposed with a shutter (not shown). In step 2, the shutter (not shown) is closed, and the signal from the CMOS image sensor is converted into a digital signal by AFE 2 and stored in the memory circuit 4 via the DFE 3 and the image processing device 5 in step 3. The correction process of this embodiment is performed from step 4. In step 5, n representing the line number is initialized to zero. In step 6, n is incremented by one. In step 7, the OB signals that are light-shielded pixels of the n-th row of image signals stored in the memory circuit 4 are averaged. The averaging effect is to reduce the influence of random noise and scratched pixels. In step 8, an offset from the black reference level is obtained. In step 9, calculation is performed by multiplying the offset obtained in step 8 by a coefficient. Here, the coefficient depends on the layout of the CMOS image sensor, and is 0.4 to 0.8. In step 10, the offset stopped in step 9 is subtracted from the effective pixel signals in the same row and written to the memory circuit 4. In step 11, it is determined whether or not processing has been performed on all image signals by looking at the row n. If not completed, the process returns to step 6; if completed, the process proceeds to step 12 and the next process is performed. However, development processing is performed.

  Also, the difference from the horizontal OB clamp 24 is that the horizontal OB clamp 24 is fed back to the input of the AD 23 by integrating about several percent with respect to the offset, and is the same as applying LPF in a plurality of rows. is there. Therefore, if the period is somewhat loose, the coefficient becomes 1 to follow.

  Since the present embodiment is performed independently for each line and no feedback is applied, the coefficient is always constant and does not become 1 even in a loose period.

It is a whole block diagram of an imaging device. It is a block diagram of an analog front end. It is a block diagram of a digital front end. It is a figure explaining the shading of the dark of a perpendicular direction. It is sectional drawing of a CMOS image sensor. It is a circuit diagram of 1 pixel of a CMOS image sensor. It is a whole layout of a CMOS image sensor. It is a flowchart which is a 2nd Example of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 CMOS image sensor 2 Analog front end 3 Digital front end 4 Memory 5 Image processing apparatus

Claims (2)

An imaging region having a light-shielded photoelectric conversion unit and a non-light-shielded photoelectric conversion unit;
Correction means for obtaining a correction value by correcting the signal from the light-shielded photoelectric conversion unit by using a correction coefficient, and correcting the signal from the corresponding non-light-shielded photoelectric conversion unit by the correction value When,
An imaging device comprising: An imaging region having a light-shielded photoelectric conversion unit and a non-light-shielded photoelectric conversion unit;
First correction means for performing correction to match a black level of a signal from the photoelectric conversion unit that is not shielded with a reference black level;
Perform a difference process between the signal level of the signal from the light-shielded photoelectric conversion unit and the reference black level, and obtain a correction value by correcting the signal obtained by the difference process using a correction coefficient, Second correction means for correcting a signal from the corresponding non-light-shielded photoelectric conversion unit by the correction value;
An imaging device comprising:

Images