Image processing noise reduction
Noise reduction in images is provided by performing a noise reduction step on blocks of pixels within a video-processing pipeline. The noise reduction step consists of applying a discrete cosine transform (DCT) to the block of pixels, quantizing the resulting DCT coefficients, and performing an inverse of the DCT to the quantized coefficients. The output of that noise reduction step is a block of image pixels similar to the input pixels, but with significantly less image noise. Because the noise reduction step can be performed quickly on small blocks of pixels, the noise reduction can be performed in real-time in a video processing pipeline.
1 . A method for removing noise from a real-time stream of digital image data, the method comprising:
receiving data from at least one image sensor of a plurality of image sensors;
converting the data into a YCbCr color space;
applying a noise reduction pipeline to remove noise from the data, wherein the pipeline includes:
downsampling at least a portion of a Chroma component of the data in order to remove a first portion of the noise,
frequency domain processing the data to remove a second portion of the noise, wherein the processing includes implementing a noise reduction step that consists of:
performing discrete cosine transform (DCT) on the data to obtain DCT values,
quantizing the DCT values, and
conducting an inverse DCT to yield noise-reduced image data; and
providing the data for displaying, broadcasting, or storing as a digital image.
2 . The method of claim 1 , further comprising exposing at least a portion of said data to a low pass filter.
3 . The method of claim 2 , wherein the low-pass filter is inserted into the noise reduction pipeline at any point to filter image data represented in the frequency domain, the low-pass filter configured to remove a third portion of the noise.
4 . The method of claim 3 , wherein the low-pass filter is selected from a plurality of low-pass filters and the selected low-pass filter includes filtering parameters configured to filter noise that is specific to a particular digital image acquisition device and environmental conditions at the time of image acquisition.
5 . The method of claim 1 , wherein a luminance component is also downsampled.
6 . The method of claim 1 , further comprising processing said noise-reduced image data through a video processing pipeline.
7 . The method of claim 6 , wherein the video processing pipeline is a high dynamic range pipeline.
8 . The method of claim 6 , wherein the video processing pipeline comprises streaming pixel values from each of the plurality of image sensors in a frame independent manner through a kernel operation that identifies saturated pixel values and a merge module to merge the pixel values to produce a high-dynamic range (HDR) image.
9 . The method of claim 8 , wherein each sensor of the plurality of image sensors includes a Bayer filter.
10 . The method of claim 8 , wherein the plurality of image sensors are each positioned with respect to at least one beamsplitter and a lens of a video camera such that incoming light is split onto the plurality of image sensors so that each image sensor senses an image that is identical but for light level.
11 . The method of claim 1 , further comprising exposing said noise-reduced image data to a low-pass filter.