System and methods for low-light image enhancement utilizing denoising preprocessing with wavelet decomposition
A system and method are disclosed for low-light image enhancement using denoising preprocessing with wavelet decomposition AI-based techniques to enhance image quality of low-light images. Subsampled images are created from a raw input image. A wavelet decomposition process is performed on each subimage to create multiple frequency domain subimages. Each frequency domain subimage is input into a corresponding neural network. The output of each corresponding network is input to an inverse wavelet module. The output of the inverse wavelet module is a denoised image that is input to an image signal processing pipeline, where additional processing may be performed on the denoised image.
1 . A system for image enhancement, comprising:
a computing device comprising at least a memory and a processor;
a denoising preprocessing module comprising a plurality of programming instructions that, when operating on the processor, cause the computing device to:
generate a plurality of subimages from an input image;
apply a frequency domain transformation to each subimage to produce frequency domain subimages;
process each of the frequency domain subimages using corresponding machine learning models and then using an inverse frequency domain transformation on the outputs of the machine learning models; and
provide a combined output to an image signal processing pipeline.
2 . The system of claim 1 , wherein the input image is in raw image format.
3 . The system of claim 1 , wherein the frequency domain transformation utilizes wavelet decomposition.
4 . The system of claim 1 , wherein the denoising preprocessing module generates four frequency domain subimages.
5 . The system of claim 1 , wherein the frequency domain transformation includes a downsampling operation.
6 . The system of claim 5 , wherein the denoising preprocessing module processes image data in multiple dimensions.
7 . The system of claim 1 , further comprising a machine learning module implementing the machine learning models, wherein at least one model includes a non-linear activation function.
8 . The system of claim 7 , wherein the machine learning module implements multiple types of neural network layers or blocks.
9 . A method for image enhancement, comprising steps of:
generating a plurality of subimages from an input image;
applying a frequency domain transformation to each subimage to produce frequency domain subimages;
processing each of the frequency domain subimages using corresponding machine learning models;
combining outputs from the machine learning models using an inverse frequency domain transformation; and
providing the combined output to an image signal processing pipeline.
10 . The method of claim 9 , wherein the input image is in a raw image format.
11 . The method of claim 9 , wherein the frequency domain transformation utilizes wavelet decomposition.
12 . The method of claim 9 , further comprising generating at least four frequency domain subimages.
13 . The method of claim 9 , wherein the frequency domain transformation includes a downsampling operation.
14 . The method of claim 13 , further comprising processing image data in multiple dimensions.
15 . A non-transitory computer program product for an electronic computation device comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the electronic computation device to:
generate a plurality of subimages from an input image;
apply a frequency domain transformation to each subimage to produce frequency domain subimages;
process each of the frequency domain subimages using corresponding machine learning models;
combine outputs from the machine learning models using an inverse frequency domain transformation; and
provide the combined output to an image signal processing pipeline.
16 . The non-transitory computer program product of claim 15 , wherein the computer readable storage medium further comprises program instructions that, when executed by the processor, cause the electronic computation device to perform a downsampling operation as part of the frequency domain transformation.
17 . The non-transitory computer program product of claim 15 , wherein the computer readable storage medium further comprises program instructions that, when executed by the processor, cause the electronic computation device to cause the electronic computation device to utilize wavelet decomposition as the frequency domain transformation.
18 . The non-transitory computer program product of claim 15 , wherein the computer readable storage medium further comprises program instructions that, when executed by the processor, cause the electronic computation device to process the input image in a raw image format.
19 . The non-transitory computer program product of claim 15 , wherein the computer readable storage medium further comprises program instructions that, when executed by the processor, cause the electronic computation device to implement at least one machine learning model with a non-linear activation function.
20 . The non-transitory computer program product of claim 19 , wherein the computer readable storage medium further comprises program instructions that, when executed by the processor, cause the electronic computation device to implement multiple types of neural network layers or blocks.