| CPC G06T 5/73 (2024.01) [G06T 7/90 (2017.01); G06T 2207/10024 (2013.01); G06T 2207/10068 (2013.01)] | 12 Claims |

|
1. A method for smoke reduction in images comprising:
accessing an RGB image of an object obscured by smoke, the RGB image including a plurality of pixels;
determining a dark channel matrix of the RGB image, where the dark channel matrix includes, for each pixel of the plurality of pixels, a minimum color component intensity for a respective pixel area centered at the respective pixel;
estimating an atmospheric light matrix for the RGB image based on the dark channel matrix, wherein the atmospheric light matrix includes an atmospheric light component value for each of the plurality of pixels;
determining a transmission map based on the atmospheric light matrix and the dark channel matrix;
de-hazing the RGB image based on the transmission map to reduce the smoke in the RGB image, wherein the de-hazing the RGB image includes:
converting the RGB image to a YUV image;
performing a de-hazing operation on the YUV image to provide a Y′UV image, wherein the performing the de-hazing operation on the YUV image includes, for each pixel x of the plurality of pixels:
determining Y′ as
![]() wherein:
T(x) is a transmission component for the pixel x, and
A(x) is the atmospheric light component value for the pixel x; and
converting the Y′UV image to the de-hazed RGB image; and
displaying the de-hazed RGB image on a display device.
|