SENSOR DARK PIXEL OFFSET ESTIMATION

• US 20130271628A1
• Filed: 11/20/2012
• Published: 10/17/2013
• Est. Priority Date: 04/17/2012
• Status: Abandoned Application

• Current Assignee: Planet Labs PBC
• Original Assignee: Skybox Imaging, Inc. (Planet Labs PBC)
• Inventors: Nishihara, H. Keith, Robinson, M. Dirk, Ku, Chwen-Yuan
• Application Number:
  US13/681,712
• Publication Number:
  US 20130271628A1
• Time in Patent Office: Days
• US Class Current: 348/243
• CPC Class Codes: H04N 25/63 applied to dark current

Abstract

Examples of systems and methods to provide estimates of dark current for pixels of a photosensor as a function of the temperature of the sensor and the gain applied to the photosensor are described. In various implementations, the dark current estimated for each pixel can depend at least partly on a global scale factor and a global bias that depend on temperature and gain and a temperature-independent and gain-independent offset value for each pixel. The scale, bias, and offsets may be determined from multiple dark field images taken by the sensor over a range of operating temperatures. In some cases, the scale and bias can be determined using a subset of less than all the image pixels. Scale and bias derived for a particular sensor can be used in the calibration of different sensors.

Claims

1. A method for correcting for dark current in a photosensor comprising a plurality of pixels, the method comprising:

under control of one or more physical computing devices configured with specific computer-executable instructions,

accessing a scale factor and a bias factor for the plurality of pixels of the photosensor, the scale factor and the bias factor depending at least partly on a temperature of the photosensor and a gain of the photosensor, the scale factor having the same value for each pixel in the plurality of pixels, and the bias factor having the same value for each pixel in the plurality of pixels;

for each pixel in the plurality of pixels of the photosensor;

accessing a pixel value that represents a response of the pixel to receipt of light;

accessing an offset value for the pixel, the offset value not dependent on the temperature or the gain of the photosensor;

determining a corrected pixel value that depends at least partly on the pixel value, the scale factor, the bias factor, and the offset; and

providing the corrected pixel value.

2. The method of claim 1, wherein the plurality of pixels includes all the pixels of the photosensor.

3. The method of claim 1, wherein the photosensor comprises a total number of pixels, the plurality of pixels comprises a first number of pixels, and the total number is greater than the first number.

4. The method of claim 3, wherein at least some of the pixels of the photosensor are divided into multiple groups of pixels, with each of the multiple groups including a plurality of pixels, and the method of claim 1 is separately applied to each of the multiple groups of pixels using a respective scale factor and bias factor that applies to the group.

5. The method of claim 1, wherein determining the corrected pixel value comprises subtracting an estimate of the dark current for the pixel from the pixel value, wherein the estimate of the dark current for the pixel is the sum of (i) the scale factor multiplied by the offset value and (ii) the bias factor.

6. The method of claim 1, wherein the pixel value is a reset-corrected pixel value from the photosensor.

7. The method of claim 1, wherein providing the corrected pixel value comprises storing the corrected pixel value in nontransitory computer storage or communicating the corrected pixel value via a transmission medium.

8. The method of claim 1, wherein accessing a scale factor and a bias factor or accessing an offset value for the pixel comprises one or more of:

accessing a lookup table or evaluating a mathematical function.

9. The method of claim 8, wherein the mathematical function comprises a spline function or a polynomial function.

10. A system for correcting for dark current in a sensor comprising a plurality of pixels, the system comprising:

physical data storage configured to store;

(i) a scale factor applicable to the plurality of pixels,

(ii) a bias factor applicable to the plurality of pixels, and

(iii) an offset value for each pixel in the plurality of pixels,

wherein the scale factor and the bias factor depend at least partly on a temperature of the sensor and a gain of the sensor, and the offset value does not depend on the temperature or the gain of the sensor; and

a physical computer processor in communication with the physical data storage, the physical computer processor configured to;

access the scale factor and the bias factor;

access a pixel value representing an output of a pixel in the plurality of pixels;

access the offset value for the pixel in the plurality of pixels; and

determine a corrected pixel value for the pixel, wherein the corrected pixel value depends at least partly on the pixel value, the scale factor, the bias factor, and the offset value.

11. The system of claim 10, wherein the computer processor is configured to determine the corrected pixel value by subtracting from the pixel value (i) the scale factor multiplied by the offset value and (ii) the bias value.

12. The system of claim 10, wherein to access the scale factor, the bias factor, or the offset value, the computer process is configured to access a lookup table or evaluate a mathematical function.

13. A sensor system comprising:

the system for correcting for dark current of claim 10; and

the sensor.

14. The sensor system of claim 13, further comprising a temperature sensor configured to measure the temperature of the sensor.

15. An imaging system comprising the sensor system of claim 13, wherein the sensor comprises a photosensor.

16. The imaging system of claim 15, further comprising a temperature sensor configured to measure the temperature of the photosensor.

17. A method for estimating dark current in a sensor comprising a plurality of pixels, the method comprising:

under control of one or more computing devices configured with specific computer-executable instructions,

accessing a dark current value for each of the plurality of pixels in the sensor;

fitting the dark current values for the plurality of pixels to a dark current model, wherein the dark current model is based at least partly on;

a scale factor;

a bias factor; and

an offset,

wherein the scale factor and the bias factor depend at least partly on a temperature of the sensor, the scale factor has a first same value for all the pixels in the plurality of pixels, the bias factor has a second same value for all the pixels in the plurality of pixels, the offset is independent of the temperature of the sensor, and the offset has values that can be different for different pixels in the plurality of pixels; and

providing the scale factor, the bias factor, and the offset.

18. The method of claim 17, wherein at least one of the scale factor or the bias factor further depends at least partly on a gain of the sensor.

19. The method of claim 17, wherein the fitting is performed by optimizing an objective function based at least partly on a sum of squared differences between the dark current value and the dark current model for each of the pixels in the plurality of pixels.

20. The method of claim 17, wherein the fitting determines the scale factor and the bias factor from fewer than all of the pixels in the sensor.

21. The method of claim 17, wherein the fitting first determines the scale factor and the bias factor, and then determines the offset value for a pixel in the plurality of pixels based at least in part on the scale factor, the bias factor, and the dark current value for the pixel.

22. The method of claim 17, wherein the dark current value for each of the pixels in the plurality of pixels is an average dark current value determined at least partly from a plurality of dark frame images taken by the sensor.

23. The method of claim 17, wherein fitting comprises determining a lookup table for at least one of the scale factor, the bias factor, or the offset.

24. The method of claim 17, wherein fitting comprises determining a mathematical function that represents at least one of the scale factor, the bias factor, or the offset.

25. The method of claim 17, further comprising measuring the dark current value for each of the plurality of pixels in the sensor.

26. The method of claim 17, wherein the plurality of pixels used for the fitting does not include pixels flagged as bad pixels.

27. The method of claim 26, further comprising identifying the bad pixels by determining pixels whose dark current is above a threshold.

28. The method of claim 26, further comprising:

excluding a first set of bad pixels for fitting the scale factor and the bias factor; and

excluding a second set of bad pixels for fitting the offset,

wherein the first set of bad pixels includes more bad pixels than the second set of bad pixels.

29. A system for estimating dark current in a sensor comprising a plurality of pixels, the system comprising:

physical data storage configured to store a dark current value for each of the plurality of pixels in the sensor; and

a physical computer processor in communication with the physical data storage, the physical computer processor configured to;

fit the dark current values for the plurality of pixels to a dark current model, wherein the dark current model is based at least partly on a scale factor, a bias factor, and an offset, the scale factor and the bias factor depending at least partly on a temperature of the sensor, the scale factor having a first same value for all the pixels in the plurality of pixels, the bias factor having a second same value for all the pixels in the plurality of pixels, the offset independent of the temperature of the sensor, and the offset having values that can be different for different pixels in the plurality of pixels; and

output the dark current model to the physical data storage.

30. The system of claim 29, wherein to fit the dark current values, the computer processor is configured to determine a lookup table or a functional fit to one or more of the scale factor, the bias factor, or the offset as at least a function of the temperature of the sensor.

31. A method for calibrating a plurality of pixels on a sensor, the method comprising:

under control of one or more physical computing devices configured with specific computer-executable instructions,

accessing a global factor that is the same for each of the plurality of pixels;

accessing a local factor for each of the plurality of pixels, the local factor capable of having values that are different for different pixels in the plurality of pixels;

accessing a pixel value for a pixel in the plurality of pixels;

determining a corrected pixel value for the pixel, wherein the corrected pixel value depends at least partly on the pixel value, the global factor, and the local factor for the pixel; and

providing the corrected pixel value.

32. The method of claim 31, wherein the global factor comprises a plurality of global factors or the local factor comprises a plurality of local factors.

33. The method of claim 31, wherein the global factor depends at least partly on a global calibration parameter associated with the sensor.

34. The method of claim 33, wherein the global calibration parameter comprises one or more of:

temperature of the plurality of pixels, gain of the plurality of pixels, quantum efficiency of the plurality of pixels, or light exposure level of the plurality of pixels.

35. The method of claim 31, further comprising using optimization techniques to determine a factorization of a calibration model for the sensor into the global factor and the local factor.

36. A method for determining dark current in a second sensor based at least in part on a dark current model for a first sensor that is different from the second sensor, the dark current model for the first sensor comprising a scale factor and a bias factor, the scale factor and the bias factor depending at least partly on temperature, the scale factor and the bias factor each having a respective global value for all the pixels in the first sensor, the method comprising:

accessing the scale factor and the bias factor for the first sensor;

accessing dark current values for pixels of the second sensor;

determining, by execution of instructions by a physical computer processor, offset values for each of the pixels of the second sensor, the offset values based at least in part on the dark current values for the second sensor and the scale factor and the bias factor for the first sensor; and

providing a dark current model for the second sensor based at least partly on the offset values for each of the pixels of the second sensor and the scale factor and the bias factor for the first sensor.

37. The method of claim 36, wherein at least one of the scale factor and the bias factor further depends at least partly on a gain of the first sensor.

38. The method of claim 36, wherein providing a dark current model for the second sensor comprises communicating the dark current model to the second sensor.

39. The method of claim 36, further comprising measuring the dark current values for pixels of the second sensor.