System, method, and apparatus for determining a high dynamic range image

1. A system, comprising:

an image sensor configured to capture image data using a plurality of selectable exposure times; and

a processing apparatus configured to:

receive a first image from the image sensor, wherein the first image is captured with a first exposure time;

receive a second image from the image sensor, wherein the second image is captured with a second exposure time that is less than the first exposure time;

determine a high dynamic range image based on the first image and the second image, wherein an image portion of the high dynamic range image is based on a corresponding image portion of the second image when a pixel of a corresponding image portion of the first image is saturated, wherein to determine the high dynamic range image includes:

responsive to a value of a pixel of an image component of the first image being in a range near a saturation level, determine a blending ratio based on a difference between the saturation level and the value of the pixel; and

determine a corresponding image component of the high dynamic range image based on a weighted sum, using the blending ratio as a weight, of the image component of the first image and a corresponding image component of the second image; and

store, display, or transmit an output image that is based on the high dynamic range image.

2. The system of claim 1 , wherein the processing apparatus is further configured to determine the high dynamic range image by performing operations to:

determine initial blending ratios for respective image portions of the high dynamic range image to obtain an initial blending ratio map;

apply a low-pass spatial filter to the initial blending ratio map to obtain a blending ratio map; and

combine the first image and the second image using the blending ratio map to obtain the high dynamic range image.

3. The system of claim 1 , wherein the processing apparatus is further configured to:

determine a noise map for the high dynamic range image based on noise level estimates for pixels of the first image, noise level estimates for pixels of the second image, and a blending ratio map that specifies how image components of the first image and the second image are combined to form the high dynamic range image; and

apply temporal noise reduction processing to the high dynamic range image based on the noise map.

4. The system of claim 1 , wherein the processing apparatus comprises an image signal processor that is configured to perform image processing operations on the high dynamic range image.

5. The system of claim 1 , wherein:

the first image is in a raw format; and

the second image is in a raw format.

6. The system of claim 1 , wherein the processing apparatus is further configured to:

apply motion compensation to align corresponding pixels of the first image and the second image when determining the high dynamic range image.

7. The system of claim 6 , wherein the processing apparatus is configured to apply motion compensation by performing operations to:

select the first image or the second image as a target image;

apply a local motion compensation transformation to a reference image to obtain a first candidate image;

apply a global motion compensation transformation to the reference image to obtain a second candidate image;

obtain a first quality metric based on the first candidate image and the target image;

obtain a second quality metric based on the second candidate image and the target image; and

based on the first quality metric and the second quality metric, select a motion compensation transformation from among the local motion compensation transformation and the global motion compensation transformation.

8. A method, comprising:

receiving a first image from an image sensor, wherein the first image is captured with a first exposure time;

receiving a second image from the image sensor, wherein the second image is captured with a second exposure time that is less than the first exposure time;

determining a high dynamic range image based on the first image and the second image,

wherein a portion of the high dynamic range image is based on a corresponding image portion of the second image when a pixel of a corresponding image portion of the first image is saturated, and

wherein determining the high dynamic range image includes using a blending ratio map to specify how the image portion of the second image is combined with the corresponding image portion of the first image to determine the image portion of the high dynamic range image; and

storing, displaying, or transmitting an output image that is based on the high dynamic range image.

9. The method of claim 8 , wherein:

the first image is in a raw format; and

the second image is in a raw format.

10. The method of claim 8 , wherein determining the high dynamic range image includes:

determining initial blending ratios for respective image portions of the high dynamic range image to obtain an initial blending ratio map;

applying a low-pass spatial filter to the initial blending ratio map to obtain a blending ratio map; and

combining the first image and the second image using the blending ratio map to obtain the high dynamic range image.

11. The method of claim 10 , wherein determining the initial blending ratio includes:

responsive to a value of a pixel of an image component of the first image being in a range near a saturation level, determining a blending ratio based on a difference between the saturation level and the value of the pixel; and

determining a corresponding image component of the high dynamic range image based on a weighted sum, using the blending ratio as a weight, of the image component of the first image and a corresponding image component of the second image.

12. The method of claim 10 , wherein determining the initial blending ratio includes:

identifying a maximum pixel value for pixels in an image portion of the first image;

if the maximum pixel value is saturated, determining the initial blending ratio to be one, wherein an image portion of the second image is used to determine the corresponding image portion of the high dynamic range image;

if the maximum pixel value is not saturated and the maximum pixel value is outside of a range near a saturation level, determining the initial blending ratio to be zero, wherein an image portion of the first image is used to determine the corresponding image portion of the high dynamic range image; and

if the maximum pixel value is not saturated and the maximum pixel value is in the range near the saturation level, determining the initial blending ratio based on a difference between the saturation level and the maximum pixel value, wherein the image portion of the first image and the corresponding image portion of the second image are combined using the initial blending ratio to determine the corresponding image portion of the high dynamic range image.

13. The method of claim 8 , further comprising:

applying motion compensation to align corresponding pixels of the first image and the second image when determining the high dynamic range image.

14. The method of claim 13 , wherein the applying motion compensation includes:

selecting the first image or the second image as a target image;

applying a local motion compensation transformation to a reference image to obtain a first candidate image;

applying a global motion compensation transformation to the reference image to obtain a second candidate image;

obtaining a first quality metric based on the first candidate image and the target image;

obtaining a second quality metric based on the second candidate image and the target image; and

based on the first quality metric and the second quality metric, selecting a motion compensation transformation from among the local motion compensation transformation and the global motion compensation transformation.

15. The method of claim 8 , further comprising:

determining a noise map for the high dynamic range image based on noise level estimates for pixels of the first image, noise level estimates for pixels of the second image, and a blending ratio map that specifies how image components of the first image and the second image are combined to form the high dynamic range image; and

applying temporal noise reduction processing to the high dynamic range image based on the noise map.

16. An image processing apparatus, comprising:

an image sensor configured to capture image data using a plurality of selectable exposure times;

a high dynamic range module configured to:

receive a first image from the image sensor, wherein the first image is captured with a first exposure time; and

receive a second image from the image sensor, wherein the second image is captured with a second exposure time that is less than the first exposure time;

determine a high dynamic range image based on the first image and the second image, wherein an image portion of the high dynamic range image is based on a corresponding image portion of the second image when a pixel of a corresponding image portion of the first image is saturated; and

determine a noise map for the high dynamic range image, wherein the noise map is determined based on a particular combination of the first image and the second image and estimates of a noise level for pixels of the first image and pixels of the second image.

17. The apparatus of claim 16 , further comprising:

a three-dimensional noise reduction module configured to combine the high dynamic range image with a recirculated image to obtain a noise reduced image, wherein the recirculated image is based on one or more previous images of a sequence of images from the image sensor.

18. The apparatus of claim 17 , wherein the three-dimensional noise reduction module includes a temporal noise reduction module configured to recursively combine the high dynamic range image with the recirculated image using a set of mixing weights for respective image portions of the recirculated image, to obtain a temporal noise reduced image.

19. The apparatus of claim 17 , further comprising:

a motion compensation module configured to apply transformations to align pixels in the recirculated image with corresponding pixels in a current image.

20. The apparatus of claim 19 , wherein the motion compensation module is further configured to apply transformations by performing operations to:

select the first image or the second image as a target image;

apply a local motion compensation transformation to a reference image to obtain a first candidate image;

apply a global motion compensation transformation to the reference image to obtain a second candidate image;

obtain a first quality metric based on the first candidate image and the target image;

obtain a second quality metric based on the second candidate image and the target image; and

based on the first quality metric and the second quality metric, select a motion compensation transformation from among the local motion compensation transformation and the global motion compensation transformation.