Method for improving SNR in low illumination conditions in a CMOS video sensor system using a self-resetting digital pixel

• US 6,963,370 B2
• Filed: 09/24/2001
• Issued: 11/08/2005
• Est. Priority Date: 09/24/2001
• Status: Active Grant

• Current Assignee: Board of Trustees of the Leland Stanford Junior University (Stanford Management Co.)
• Original Assignee: Board of Trustees of the Leland Stanford Junior University (Stanford Management Co.)
• Inventors: DiCarlo, Jeffrey M., Wandell, Brian, El Gamal, Abbas, Liu, Xinqiao
• Primary Examiner(s): Ometz, David L.
• Assistant Examiner(s): JELINEK, BRIAN J
• Application Number:
  US09/962,847
• Publication Number:
  US 20030058356A1
• Time in Patent Office: 1,506 Days
• Field of Search: 348/297, 348/216.1, 348/308, 348/294, 348/302, 348/241, 348/296, 250/208.1, 250/214.A
• US Class Current: 348/297
• CPC Class Codes: H04N 25/581 acquired simultaneously
• CPC Class Codes: H04N 25/65 applied to reset noise, e.g. KTC noise related to CMOS structures by techniques other than CDS
• CPC Class Codes: H04N 25/772 comprising A/D, V/T, V/F, I/T or I/F converters

Abstract

A method for improving SNR without reducing dynamic range in a CMOS video sensor system under low illumination is described, wherein the CMOS video sensor system employing self-reset DPS architecture includes pixel level A/D conversion and wherein each DPS pixels is capable of resetting itself whenever a corresponding diode reaches saturation during integration time, the method comprising the steps of eliminating global frame reset for the self-reset DPS pixels, capturing and storing a plurality of frames, extending integration time beyond frame readout time, and generating frame images using one or more previously stored frames.

Claims

1. A method for improving signal-to-noise ratio (SNR) without reducing dynamic range in a complementary metal oxide semiconductor (CMOS) video sensor system under low illumination, wherein said system employing self-reset digital pixel sensor (DPS) architecture and wherein each DPS pixels capable of resetting itself whenever a corresponding diode reaches saturation during integration time, said method comprising steps of:

eliminating global frame reset for said pixels thereby extending integration time beyond frame readout time;

generating a frame image using one or more previously stored frames.

2. The method of claim 1, further comprising a step of:

terminating extended integration time when dramatic motion is detected thereby preventing severe motion blur.

3. The method of claim 2, wherein said self-reset pixels are digital pixels each capable of resetting itself during said integration time whenever a corresponding diode reaches saturation.

4. The method of claim 1, wherein said self-reset pixels are digital pixels each capable of resetting itself during said integration time whenever a corresponding diode reaches saturation.

5. A method for improving signal-to-noise ratio (SNR) without reducing dynamic range in a complementary metal oxide semiconductor (CMOS) video sensor system under low illumination, wherein said system employing self-reset digital pixel sensor (DPS) architecture and wherein each OPS pixels capable of resetting itself whenever a corresponding diode reaches saturation during integration time, said method comprising steps of:

eliminating global frame reset for said DPS pixels;

capturing and storing a plurality of frames;

extending integration time beyond frame readout time; and

generating frame images using one or more previously stored frames.

6. The method of claim 5, wherein said extending step further comprising steps of:

determining optimal duration of said extending step by analyzing captured frames; and

terminating said extending step when dramatic motion is detected thereby preventing severe motion blur.

7. The method of claim 6, wherein said capturing and storing step comprising steps of:

converting at pixel level captured analog pixel signals to digital pixel values; and

storing said digital pixel values.

8. The method of claim 5, wherein said capturing and storing step comprising steps of:

converting at pixel level captured analog pixel signals to digital pixel values; and

storing said digital pixel values in a memory.

9. The method of claim 8, wherein said memory is a pixel-level digital memory.

10. The method of claim 8, wherein said memory is a 3T dynamic memory.

11. The method of claim 8, wherein said memory is integrated at chip level.

12. The method of claim 5, further comprising steps of:

integrating fractional frame time; and

performing analog-to-digital conversion between frames.

13. The method of claim 12, further comprising steps of:

determining optimal duration of said extending step by analyzing captured frames; and

terminating said extending step when dramatic motion is detected by said analyzing step thereby preventing severe motion blur.

14. The method of claim 12, wherein said capturing and storing step further comprising a step of:

performing pixel level analog-to-digital conversion.

15. A method for improving signal-to-noise ratio (SNR) without reducing dynamic range in a complementary metal oxide semiconductor (CMOS) video sensor system under low illumination, wherein said CMOS video sensor system employing self-reset digital pixel sensor (DPS) architecture including pixel level analog-to-digital conversion and wherein each DPS pixels is capable of resetting itself whenever a corresponding diode reaches saturation during integration time, said method comprising steps of:

a) eliminating global frame reset;

b) capturing and storing a first frame;

c) determining photocurrent for said DPS pixels;

d) capturing and storing a next frame;

e) extending integration time beyond frame readout time;

f) analyzing captured frames to determine whether to terminate said extending step, wherein said extending step is terminated if motion blur is detected; and

g) generating frame image using number of self-resets occurred during said integration time and one or more previously stored frames.

16. The method of claim 15, further comprising repeating steps d) through g) until all frames are captured.