Method and device for generating a high-dynamic range (HDR) image
In one implementation, a method includes obtaining an image. The method includes generating a plurality of denoised images by denoising the image using a respective plurality of denoisers, wherein the plurality of denoisers includes a first denoiser associated with a first level of exposure and a second denoiser associated with a second level of exposure. The method includes generating a combined image by combining the plurality of denoised images.
1 . A method comprising:
obtaining an image;
generating a plurality of denoised images by denoising the image using a respective plurality of denoisers, wherein the plurality of denoisers includes a first denoiser that applies a first denoising algorithm associated with a first level of exposure and a first level of noise, a second denoiser that applies a second denoising algorithm associated with a second level of exposure, and a third denoiser that applies a third denoising algorithm associated with the first level of exposure and a second level of noise; and
generating a combined image by combining the plurality of denoised images.
2 . The method of claim 1 , wherein the image is a frame of video.
3 . The method of claim 1 , wherein the image includes motion blur.
4 . The method of claim 1 , wherein the first denoiser includes a first neural network trained on training data including first images having the first level of exposure and noisy versions of the first images and the second denoiser includes a second neural network trained on training data including second images having the second level of exposure and noisy versions of the second images.
5 . The method of claim 1 , wherein the first denoiser includes a first neural network trained on training data including first images having the first level of exposure and noisy versions of the first images having the first level of noise and the third denoiser includes a second neural network trained on training data including third images having the first level of exposure and noisy versions of the third images having the second level of noise.
6 . The method of claim 1 , wherein generating the combined image includes performing exposure compensation on the plurality of denoised images.
7 . The method of claim 1 , wherein generating the combined image includes generating a weighted blend of the plurality of denoised images.
8 . The method of claim 1 , wherein generating the combined image includes combining the plurality of denoised images and one or more supplemental images.
9 . The method of claim 1 , wherein the combined image is a high-dynamic range (HDR) image.
10 . An electronic device comprising:
a non-transitory memory; and
one or more processors configured to:
obtain an image;
generate a plurality of denoised images by denoising the image using a respective plurality of denoisers, wherein the plurality of denoisers includes a first denoiser that applies a first denoising algorithm associated with a first level of exposure and a first level of noise, a second denoiser that applies a second denoising algorithm associated with a second level of exposure, and a third denoiser that applies a third denoising algorithm associated with the first level of exposure and a second level of noise; and
generate a combined image by combining the plurality of denoised images.
11 . The electronic device of claim 10 , wherein the image is a frame of video.
12 . The electronic device of claim 10 , wherein the image includes motion blur.
13 . The electronic device of claim 10 , wherein the first denoiser includes a first neural network trained on training data including first images having the first level of exposure and noisy versions of the first images and the second denoiser includes a second neural network trained on training data including second images having the second level of exposure and noisy versions of the second images.
14 . The electronic device of claim 10 , wherein the first denoiser includes a first neural network trained on training data including first images having the first level of exposure and noisy versions of the first images having the first level of noise and the third denoiser includes a second neural network trained on training data including third images having the first level of exposure and noisy versions of the third images having the second level of noise.
15 . The electronic device of claim 10 , wherein the one or more processors are configured to generate the combined image by performing exposure compensation on the plurality of denoised images.
16 . The electronic device of claim 10 , wherein the one or more processors are configured to generate the combined image by generating a weighted blend of the plurality of denoised images.
17 . The electronic device of claim 10 , wherein the one or more processors are configured to generate the combined image by combining the plurality of denoised images and one or more supplemental images.
18 . A non-transitory computer-readable medium having instructions encoded thereon that, when executed by one or more processors of an electronic device, causes the electronic device to perform a method comprising:
obtaining an image;
generating a plurality of denoised images by denoising the image using a respective plurality of denoisers, wherein the plurality of denoisers includes a first denoiser that applies a first denoising algorithm associated with a first level of exposure and a first level of noise, a second denoiser that applies a second denoising algorithm associated with a second level of exposure, and a third denoiser that applies a third denoising algorithm associated with the first level of exposure and a second level of noise; and
generating a combined image by combining the plurality of denoised images.