Systems for multi-exposure imaging
Systems and methods for multi-exposure imaging are provided. Multiple images may be captured having different exposure times. The multiple images may be processed to produce an output image. Multi-exposure images may be summed prior to linearization. Pixel values in multi-exposure images may be summed to nonlinear pixel values. The nonlinear pixel values may be linearized using one or more knee points. Multi-exposure imaging may be used to motion-intensive application such as automotive applications.
1. A method for multi-exposure imaging, comprising:
with an image sensor, capturing a first image having a first exposure time;
with the image sensor, capturing a second image having a second exposure time that is less than the first exposure time;
with an image processor, processing the first and second images to produce a processed image, wherein processing the first and second images comprises adding pixel signals P 1 from the first image to pixel signals P 2 from the second image to produce summed pixel signals P lo ;
with the image processor, linearizing the summed pixel signals P lo to produce pixel signals P in the processed image, wherein linearizing the summed pixel signals P lo comprises scaling the summed pixel signals P lo that are above a knee point, wherein scaling the summed pixel signals P lo that are above a knee point comprises scaling the summed pixel signals P lo that are above a knee point by a factor, wherein the factor comprises an exposure ratio g plus one and wherein the exposure ratio g comprises the first exposure time divided by the second exposure time.
2. The method defined in claim 1 , wherein the image sensor has an associated scale, and wherein the knee point comprises the associated scale multiplied by (1+1/g).
3. A method for multi-exposure imaging, comprising:
with an image sensor, capturing a first image having a first exposure time;
with the image sensor, capturing a second image having a second exposure time that is less than the first exposure time; and
with an image processor, processing the first and second images to produce a processed image, wherein processing the first and second images comprises adding pixel signals P 1 from the first image to pixel signals P 2 from the second image to produce summed pixel signals P lo , wherein, for summed pixel signals P lo below a knee point Kp, pixel signals P in the processed image are P lo , wherein, for summed pixel signals above the knee point Kp, pixel signals P in the processed image comprise Kp+(P 1 +P 2 −Kp)*(g+1), wherein g is an exposure ratio comprising the first exposure time divided by the second exposure time.
4. A method for multi-exposure imaging, comprising:
with an image sensor, capturing a plurality of images having different exposure times;
with an image processor, processing the plurality of images to produce a processed image, wherein processing the plurality of images to produce a processed image comprises adding pixel signals from the plurality of images to produce summed pixel signals P lo , wherein processing the plurality of images to produce the processed image further comprises linearizing the summed pixel signals P lo to produce linearized pixel signals P, wherein linearizing the summed pixel signals P lo comprises linearizing the summed pixel signals P lo using a knee point, wherein the plurality of images comprises at least first, second, and third images, wherein g 1 is an exposure ratio between the first and second images, wherein g 2 is an exposure ratio between the second and third images, wherein the knee point comprises a scale of the image sensor multiplied by (1+1/g 1 +1/g 2 ), and wherein linearizing the summed pixel signals P lo further comprises linearizing the summed pixel signals P lo using an additional knee point.
5. A method for multi-exposure imaging, comprising:
with an image sensor, capturing a plurality of images having different exposure times;
with an image processor, processing the plurality of images to produce a processed image, wherein processing the plurality of images to produce a processed image comprises adding pixel signals from the plurality of images to produce summed pixel signals P lo , wherein processing the plurality of images to produce the processed image further comprises linearizing the summed pixel signals P lo to produce linearized pixel signals P, wherein linearizing the summed pixel signals P lo comprises linearizing the summed pixel signals P lo using a knee point, wherein the plurality of images comprises at least first, second, and third images, wherein g 1 is an exposure ratio between the first and second images, wherein g 2 is an exposure ratio between the second and third images, wherein the knee point comprises a scale of the image sensor multiplied by (1+1/g 1+1 /g 2 ), and wherein linearizing the summed pixel signals P lo further comprises comparing the summed pixel signals P lo with an additional knee point, wherein the additional knee point comprises the scale of the image sensor multiplied by (2+1/g 2 ).
6. An imaging device, comprising:
an image sensor that captures a plurality of images having different exposure times; and
an image processor that produces a summed image having summed pixel signals P lo by adding the plurality of images, wherein the image processor linearizes the summed pixel signals P lo , wherein the plurality of images comprise first and second images, wherein the summed pixel signals P lo have a knee point Kp given
Kp
=
fullscale
·
(
1
+
1
g
)
,
by wherein g is an exposure ratio between the first and second images, and wherein fullscale is a scale of the imaging device.
7. An imaging device, comprising:
an image sensor that captures a plurality of images having different exposure times; and
an image processor that produces a summed image having summed pixel signals P lo by adding the plurality of images, wherein the image processor linearizes the summed pixel signals P lo , wherein the plurality of images comprise first, second, and third images, wherein the pixel signals P lo have a knee point Kp given by
Kp
1
=
fullscale
·
(
1
+
1
g
1
+
1
g
1
·
g
2
)
,
wherein g 1 is an exposure ratio between the first and second images, wherein g 2 is an exposure ratio between the second and third images, and wherein fullscale is a scale of the imaging device.
8. The imaging device defined in claim 7 , wherein the imaging device comprises an automobile imaging device and wherein the pixel signals P lo have an additional knee point K2p given by
Kp
2
=
fullscale
·
(
2
+
1
g
2
)
.