Defogging images and video

Patent metadata

Field	Value
Publication number	US-9965835-B2
Application number	US-201414555803-A
Country	US
Kind code	B2
Filing date	Nov 28, 2014
Priority date	Nov 28, 2014
Publication date	May 8, 2018
Grant date	May 8, 2018

Abstract

A method and system are disclosed. The method may include determining a histogram of intensity values for pixels in image sensor data in which the histogram is bimodal. The method may include determining a breakpoint between the two modes. The histogram may include a first distribution of intensity values below the breakpoint and a second distribution of intensity values above the breakpoint. The method may include generating output intensity values. Generating output intensity values may include compressing the first distribution of intensity values of the pixels with intensity values below the breakpoint, stretching the second distribution of intensity values of the pixels with intensity values above the breakpoint, and generating an output image based on the output intensity values.

First claim

What is claimed is:

1. A method comprising:
determining a histogram of input intensity values for pixels in image sensor data, wherein the image sensor data includes a video stream having a plurality of images;
determining, by a processor, whether the histogram is bimodal, wherein when the histogram is determined to be bimodal with two modes, the method further comprising:
determining a breakpoint between the two modes such that the breakpoint is a low-pass filtering of a plurality of breakpoints corresponding to the plurality of images,
wherein the histogram includes a first distribution of the input intensity values below the breakpoint and a second distribution of the input intensity values above the breakpoint;

generating output intensity values, corresponding to and based on the input intensity values, such that:
a third distribution of the output intensity values, corresponding to the input intensity values below the breakpoint, is compressed relative to the first distribution, and
a fourth distribution of the output intensity values, corresponding to the input intensity values above the breakpoint, is stretched relative to the second distribution; and

generating an output image based on the output intensity values.

2. The method of claim 1 , wherein the plurality of images includes a current image in the video stream and a previous image in the video stream, the method further comprising:
determining a previous histogram of intensity values for pixels in the previous image, wherein the previous histogram is bimodal; and
determining the breakpoint for the current image based on a previous breakpoint between the two modes of the previous histogram.

3. The method of claim 2 ,
wherein the video stream includes at least three images, and
wherein the breakpoint for the current image is determined based on a plurality of previous breakpoints between modes in the corresponding plurality of previous images.

4. The method of claim 3 , further comprising:
generating an output video stream based on the output intensity values;
compressing the output video stream to generate a compressed video stream; and
transmitting the compressed video stream to a monitoring station.

5. The method of claim 1 , wherein determining that the histogram is bimodal includes determining that the histogram includes two local maxima.

6. The method of claim 1 , wherein determining that the histogram is bimodal includes:
determining that the pixel intensities are clustered around two separated values.

7. The method of claim 1 , further comprising generating output intensity values such that the output intensity values, corresponding to the input intensity values above the breakpoint, include an intensity value below the breakpoint.

8. The method of claim 7 , wherein a user-defined parameter determines an extent to which the fourth distribution of intensity values includes output intensity values below the breakpoint.

9. The method of claim 1 ,
wherein the fourth distribution of the output intensity values, corresponding to the input intensity values above the breakpoint, is stretched, relative to the second distribution, based on the compression of the third distribution relative to the first distribution.

10. The method of claim 9 , wherein the breakpoint between the two modes is an input breakpoint, the method further including:
generating the output image such that a histogram of output intensity values for pixels of the output image is bimodal with an output breakpoint between two modes, and
wherein the output image breakpoint is at a lower intensity value than the input breakpoint.

11. A device comprising:
a camera to capture an image and generate image sensor data; and
a processor to
determine a histogram of input intensity values for pixels in the image sensor data, wherein the image sensor data includes a video stream having a plurality of images;
determine whether the histogram is bimodal such that the breakpoint is a low-pass filtering of a plurality of breakpoints corresponding to the plurality of images, wherein the processor determines that the histogram is bimodal with two modes;
determine a breakpoint between the two modes, wherein the histogram includes a first distribution of the input intensity values below the breakpoint and a second distribution of the input intensity values above the breakpoint;
generate output intensity values, corresponding to and based on the input intensity values, such that:
a third distribution of the output intensity values, corresponding to the input intensity values below the breakpoint, is compressed relative to the first distribution, and
a fourth distribution of the output intensity values, corresponding to the input intensity values above the breakpoint, is stretched relative to the second distribution; and

generate an output image based on the output intensity values.

12. The device of claim 11 , wherein the plurality of images includes a current image in the video stream and a previous image, and wherein the processor is further configured to:
determine a previous histogram of intensity values for pixels in the previous image, wherein the previous histogram is bimodal;
determine a current histogram of intensity values for pixels in the current image, wherein the current histogram is bimodal; and
determine the breakpoint based on a previous breakpoint between the two modes of the previous histogram and a current breakpoint between the two modes of the current histogram.

13. The device of claim 12 ,
wherein the video stream includes at least three images, and
wherein, when determining the breakpoint, the processor is further configured to determine the breakpoint based on at least three breakpoints between modes in the corresponding at least three images.

14. The device of claim 13 , wherein the processor is further configured to:
generate an output video stream based on the output intensity values; and
compress the output video stream to generate a compressed video stream, wherein the device further includes a transmitter to send the compressed video stream across a network to a monitoring station.

15. The device of claim 11 , wherein when the processor determines that the histogram is bimodal, the processor is further configured to determine that the histogram includes two local maxima.

16. The device of claim 11 , wherein when the processor determines that the histogram is bimodal, the processor is further configured to:
determine that pixel intensities are clustered around two separated values.

17. The device of claim 11 , wherein the processor is configured to generate output intensity values such that the output intensity values, corresponding to the input intensity values above the breakpoint, include an intensity value below the breakpoint.

18. The device of claim 17 , wherein a user-defined parameter determines an extent to which the processor is configured to generate output intensity values such that the fourth distribution output intensity values below the breakpoint.

Assignees

• Axis Ab

Inventors

• Svensson Fredrik
• Kannermark Per
• Nilsson Andreas

Classifications

• G06T5/40

  using histogram techniques · CPC title

• G06T5/009 Primary

  Physics · mapped topic

• G06T5/92 Primary

  based on global image properties · CPC title