High dynamic range imaging device
The present invention relates to improved imaging devices having high dynamic range and to monitoring and automatic control systems incorporating the improved imaging devices.
1. An imaging device comprising:
a two dimensional array of pixels configured to independently select one of a plurality of available integration periods;
wherein each pixel of said two dimensional array of pixels is sampled to select one of said plurality of available integration periods;
wherein a first sample of said each pixel sampling is for a longest of said plurality of available integration periods;
circuitry for conversion of a first value corresponding to a charge level on any given pixel directly to a floating point numerical format; and
circuitry for detecting and flagging readings which are abnormally low or abnormally high for an integration range selected and used to acquire said reading of a light induced charge for said pixel.
2. An imaging device as in claim 1 wherein said circuitry is configured to normalize said first value corresponding to said charge level.
3. An imaging device as in claim 1 further comprising a set of selectable integration periods for said pixels and a second value corresponding to a selected integration period for a given pixel, said circuitry is configured for conversion of said first value and said second value directly to a floating point numerical format.
4. An imaging device as in claim 3 wherein said circuitry is configured to normalize said first value corresponding to said charge level and said second value combined.
5. An imaging device comprising:
a two dimensional array of pixels configured to independently select one of a plurality of available integration periods;
wherein each pixel of said two dimensional array of pixels is sampled to select one of said plurality of available integration periods;
wherein a first sample of said each pixel sampling is for a longest of said plurality of available integration periods;
one of said set of available integration periods is selected and recorded in memory for substantially each pixel;
corresponding reading of an integrated charge level on each pixel is digitized;
a representation of a selected integration period along with said corresponding reading of an integrated charge level are mapped directly into a floating point representation; and
circuitry for detecting and flagging readings which are abnormally low or abnormally high for an integration range selected and used to acquire said reading of a light induced charge for said pixel, such that a least significant bit of a mantissa value is not utilized for said converted value, but is used to add information about a selected pixel value without significantly changing an actual value of said pixel.
6. An imaging device as in claim 5 further comprising circuitry to normalize said representation of a selected integration period along with said corresponding reading of an integrated charge level.
7. An imaging device as in claim 5 wherein a first selected integration period for any given pixel is independent of a second selected integration period for a second pixel.
8. An imaging device as in claim 5 wherein one of said set of available integration periods is selected and recorded in memory for substantially every active pixel; and at the end of an integration period, a reading of an integrated charge level on each pixel is digitized and output in a form that may be mapped directly into a mantissa of a floating point representation along with a corresponding indication of said integration period used to read said pixel is output in a form which may be mapped into an exponent of a floating point representation of a pixel value.
9. An imaging device as in claim 8 wherein a ratio in sensitivity for a pixel using a first available integration period relative to a second sensitivity using another of said available integration periods is approximately an integral power of 2.
10. A device as in claim 8 wherein the indication of the lower than expected reading is included in the floating point pixel reading using an encoding which does not result in substantial alteration of the numerical pixel value from that which would be provided if the indication was not included.
11. An imaging device, comprising:
a two dimensional array of pixels configured to independently select one of a plurality of available integration periods;
wherein each pixel of said two dimensional array of pixels is sampled to select one of said plurality of available integration periods;
wherein a first sample of said each pixel sampling is for a longest of said plurality of available integration periods;
wherein values of indices identifying said selected integration period for a given pixel along with a corresponding digitized reading of a given pixel value are mapped directly into an IEEE 754-based floating point format; and
circuitry for detecting and flagging readings which are abnormally low or abnormally high for an integration range selected and used to acquire said reading of a light induced charge for said pixel, such that a least significant bit of a mantissa value is not utilized for a converted value of said pixel, but is used to add information about a selected pixel value without significantly changing an actual value of said pixel.
12. An imaging device as in claim 11 where a range of said floating point number has a minimum value of 0 and a maximum value of 1.
13. An imaging device as in claim 11 where a range of said floating point number is set directly from mapping of integration period index values and pixel values.
14. An imaging device as in claim 11 wherein at an end of an integration period, a reading of an integrated charge level on each pixel is digitized and output in a form that is mapped directly into a mantissa of a floating point representation along with a corresponding indication of said integration period used to read said pixel is output in a form which is mapped into an exponent of a floating point representation of a pixel value.
15. An imaging device as in claim 14 wherein a ratio in sensitivity for a pixel using a first available integration period relative to a second sensitivity using another of available integration periods is approximately an integral power of 2.
16. An imaging device comprising:
a two dimensional array of pixels configured to independently select one of a plurality of available integration periods;
wherein each pixel of said two dimensional array of pixels is sampled to select one of said plurality of available integration periods;
wherein a first sample of said each pixel sampling is for a longest of said plurality of available integration periods;
one of said set of available integration periods is selected and recorded in memory for substantially each pixel;
a corresponding reading of an integrated charge level on each pixel is digitized;
a representation of a selected integration period along with said corresponding reading of an integrated charge level are mapped directly into a floating point representation; and
circuitry for conversion of a first value corresponding to a charge level on any given pixel directly to a floating point numerical format, said circuitry for detecting and flagging readings which are abnormally low or abnormally high for an integration range selected and used to acquire said reading of a light induced charge for said pixel, such that a least significant bit of a mantissa value is not utilized for said converted value, but is used to add information about a selected pixel value without significantly changing an actual value of said pixel.
17. An imaging device as in claim 16 wherein said circuitry is configured to normalize said first value corresponding to said charge level.
18. An imaging device as in claim 16 further comprising a set of selectable integration periods for said pixels and a second value corresponding to a selected integration period for a given pixel, said circuitry is configured for conversion of said first value and said second value directly to a floating point numerical format.
19. An imaging device as in claim 18 wherein said circuitry is configured to normalize said first value corresponding to said charge level and said second value combined.
20. An imaging device as in claim 16 further comprising circuitry to normalize said representation of a selected integration period along with said corresponding reading of an integrated charge level.
21. An imaging device as in claim 16 wherein a first selected integration period for any given pixel is independent of a second selected integration period for a second pixel.
22. An imaging device as in claim 16 wherein one of said set of available integration periods is selected and recorded in memory for substantially every active pixel; and at the end of an integration period, a reading of an integrated charge level on each pixel is digitized and output in a form that may be mapped directly into a mantissa of a floating point representation along with a corresponding indication of said integration period used to read said pixel is output in a form which may be mapped into an exponent of a floating point representation of a pixel value.
23. An imaging device as in claim 22 wherein a ratio in sensitivity for a pixel using a first available integration period relative to a second sensitivity using another of said available integration periods is approximately an integral power of 2.
24. An imaging device including:
a two dimensional array of pixels configured to independently select one of a plurality of available integration periods;
wherein each pixel of said two dimensional array of pixels is sampled to select one of said plurality of available integration periods;
wherein a first sample of said each pixel sampling is for a longest of said plurality of available integration periods; and
an integral circuit which provide image output from the array of pixels in a normalized floating point format;
said circuit for detecting and flagging readings which are abnormally low or abnormally high for an integration range selected and used to acquire said reading of a light induced charge for said pixel, such that a least significant bit of a mantissa value is not utilized for a converted value of said pixel, but is used to add information about a selected pixel value without significantly changing an actual value of said pixel.
25. A device as in claim 24 wherein a pixel data conforms to a proposed 16 bit extension of the IEEE 754 floating point format.
26. A device as in claim 24 wherein in addition to the said floating point normalization, a pixel data is scaled to provide output levels with a maximum pixel reading of approximately one.
27. A device as in claim 24 wherein electronic circuits to normalize the floating point value are included on the same integrated circuit substrate as the pixel array.
28. A device as in claim 24 wherein a pixel output value varies approximately linearly with light level.
29. A device as in claim 24 wherein pixel data provided to the circuit which normalizes an output data is already in a non-normalized floating point form.
30. A device as in claim 24 wherein pixel values have a range which exceeds 64000 to one for non-zero light level readings.
31. A device as in claim 24 wherein pixel values include an encoding to indicate that a pixel value is lower than expected for the integration range selected to acquire pixel reading.
32. A device as in claim 24 wherein pixel values include an encoding to indicate that the pixel value is higher than expected for the integration range selected to acquire the pixel reading.
33. A device as in claim 24 wherein pixel output is provided in logarithmic form by circuits included on the same substrate as the imaging array
34. A device as in claim 24 wherein a histogram of image data is generated in conjunction with the floating point processing circuits.
35. A device as in claim 34 wherein the data path to transmit control signals to circuits to generate the histogram and the data path to transmit histogram data are at least partially separated from the path over which pixel data is communicated.