Entropy maximization based auto-exposure

1 . An image capture apparatus comprising:

an image sensor;

a motion sensor; and

an auto-exposure unit configured to:

obtain, from the image sensor, an input image captured in accordance with an exposure interval;

obtain, from the motion sensor, motion data indicating motion of the image capture apparatus temporally corresponding to the exposure interval; and

obtain a target exposure interval and a target gain based on the input image, the exposure interval, and the motion data, wherein the image capture apparatus is configured to control the image sensor to obtain a subsequent input image captured in accordance with the target exposure interval and the target gain;

wherein the image capture apparatus is configured to output or store data representing the subsequent input image.

2 . The image capture apparatus of claim 1 , wherein, to obtain the target exposure interval, the auto-exposure unit is configured to:

obtain image information-amount data for the input image based on the motion data;

obtain derivative information-amount data based on the image information-amount data and a candidate exposure adjustment; and

obtain, as the target exposure interval, an information-amount maximizing exposure interval based on the image information-amount data and the derivative information-amount data.

3 . The image capture apparatus of claim 2 , wherein:

the input image includes pixels; and

to obtain the target exposure interval, the auto-exposure unit is configured to:

obtain pixel information-amount data for the pixels; and

obtain a sum of the pixel information-amount data as the image information-amount data.

4 . The image capture apparatus of claim 3 , wherein, to obtain the pixel information-amount data, the auto-exposure unit is configured to:

for each pixel from the pixels:

determine a pixel information-amount value for the pixel based on a signal-to-noise ratio for the pixel, pixel saturation, and motion blur.

5 . The image capture apparatus of claim 4 , wherein, to determine the respective pixel information-amount value, the auto-exposure unit is configured to:

in response to a determination that an input value for the pixel is a maximum pixel value, use a defined value as the pixel information-amount value for the pixel; and

in response to a determination that the input value for the pixel is less than the maximum pixel value:

obtain a signal level information-amount for the pixel as the base two logarithm of a signal-to-noise ratio for the pixel, wherein the signal-to-noise ratio is a ratio of the input value to a noise value for the pixel, and wherein the noise value for the pixel is modeled as a square root of a sum of a first defined value and a product of a second defined value and the input value of the pixel; and

use a motion blur adjusted value for the pixel as the pixel information-amount value for the pixel.

6 . The image capture apparatus of claim 5 , wherein the defined value is zero.

7 . The image capture apparatus of claim 6 , wherein, to obtain the motion blur adjusted value for the pixel, the auto-exposure unit is configured to:

determine an amount of motion blur based on the motion data, the exposure interval, and a size of the pixel; and

use, as the motion blur adjusted value, a result of dividing the signal level information-amount by a sum of the size of the pixel and the amount of motion blur.

8 . The image capture apparatus of claim 3 , wherein, to obtain the pixel information-amount data, the auto-exposure unit is configured to:

for each pixel from the pixels:

determine a pixel information-amount value for the pixel based on a histogram of the input image.

9 . The image capture apparatus of claim 8 , wherein, to obtain the pixel information-amount data, the auto-exposure unit is configured to:

obtain a saliency map indicating a relative importance of respective pixels within the input image; and

for each pixel from the pixels:

determine a saliency adjusted information-amount by multiplying a saliency value from the saliency map for the pixel by the pixel information-amount value; and

use the saliency adjusted information-amount as the pixel information-amount value.

10 . The image capture apparatus of claim 9 , wherein, to obtain the pixel information-amount data, the auto-exposure unit is configured to:

in response to a determination that a pixel is a saturated pixel along a stitch line for image merging, weight the pixel to minimize saturated pixels along the stitch line.

11 . A method comprising:

obtaining, by an image capture apparatus, a target exposure interval and a target gain for capturing an image by:

obtaining, from an image sensor of the image capture apparatus, a previously captured image captured in accordance with a previously identified exposure interval;

obtaining, from a motion sensor of the image capture apparatus, motion data indicating motion of the image capture apparatus temporally corresponding with the previously identified exposure interval; and

obtaining the target exposure interval and the target gain based on the previously captured image, the previously identified exposure interval, and the motion data;

controlling the image sensor to obtain a subsequent input image captured in accordance with the target exposure interval and the target gain; and

outputting or storing data representing the subsequent input image.

12 . The method of claim 11 , wherein, obtaining the target exposure interval includes:

obtaining image information-amount data for the previously captured image based on the motion data;

obtaining derivative information-amount data based on the image information-amount data and a candidate exposure adjustment; and

obtaining, as the target exposure interval, an information-amount maximizing exposure interval based on the image information-amount data and the derivative information-amount data.

13 . The method of claim 12 , wherein:

the previously captured image includes pixels; and

obtaining the target exposure interval includes:

obtaining pixel information-amount data for the pixels; and

obtaining a sum of the pixel information-amount data as the image information-amount data.

14 . The method of claim 13 , wherein obtaining the pixel information-amount data includes:

for each pixel from the pixels:

determining a pixel information-amount value for the pixel based on a signal-to-noise ratio for the pixel, pixel saturation, and motion blur.

15 . The method of claim 14 , wherein determining the pixel information-amount value includes:

in response to a determination that an input value for the pixel is a maximum pixel value, using a defined value as the pixel information-amount value for the pixel; and

in response to a determination that the input value for the pixel is less than the maximum pixel value:

obtaining a signal level information-amount for the pixel as the base two logarithm of a signal-to-noise ratio for the pixel, wherein the signal-to-noise ratio is a ratio of the input value to a noise value for the pixel, and wherein the noise value for the pixel is modeled as a square root of a sum of a first defined value and a product of a second defined value and the input value of the pixel; and

using a motion blur adjusted value for the pixel as the pixel information-amount value for the pixel.

16 . The method of claim 15 , wherein obtaining the motion blur adjusted value for the pixel includes:

determining an amount of motion blur based on the motion data, the previously identified exposure interval, and a size of the pixel; and

using, as the motion blur adjusted value, a result of dividing the signal level information-amount by a sum of the size of the pixel and the amount of motion blur.

17 . The method of claim 13 , wherein obtaining the pixel information-amount data includes: for each pixel from the pixels:

determining a pixel information-amount value for the pixel based on a histogram of the previously captured image.

18 . The method of claim 17 , wherein obtaining the pixel information-amount data includes:

obtaining a saliency map indicating a relative importance of respective pixels within the previously captured image; and

for each pixel from the pixels:

determining a saliency adjusted information-amount by multiplying a saliency value from the saliency map for the pixel by the pixel information-amount value; and

using the saliency adjusted information-amount as the pixel information-amount value.

19 . The method of claim 18 , wherein obtaining the pixel information-amount data includes:

in response to a determination that a pixel is a saturated pixel along a stitch line for image merging, weighting the pixel to minimize saturated pixels along the stitch line.

20 . An image capture apparatus comprising:

an image sensor;

a motion sensor; and

an auto-exposure unit configured to:

obtain, from the image sensor, an input image captured in accordance with an exposure interval;

obtain, from the motion sensor, motion data indicating motion of the image capture apparatus temporally corresponding to the exposure interval;

obtain a saliency map indicating a relative importance of respective pixels within the input image; and

obtain a target exposure interval and a target gain based on the input image, the exposure interval, the saliency map, and the motion data, wherein the image capture apparatus is configured to control the image sensor to obtain a subsequent input image captured in accordance with the target exposure interval and the target gain;

wherein the image capture apparatus is configured to output or store data representing the subsequent input image.