METHOD AND APPARATUS FOR DE-NOISING AN IMAGE USING VIDEO EPITOME
A method and an apparatus for de-noising an image, and in particular, de-noising an image using video epitome based on a source video image. An embodiment of the present principles provides a method of processing an image in a video, comprising: decoding an encoded version of the image to generate a decoded version of the image; and generating a de-noised version of the image using the decoded version of the image and a video image epitome which is a texture epitome associated with the image, wherein the video image epitome was extracted from a source version of the image, wherein the generating comprises: de-noising a current patch using corresponding patches located in the video image epitome that correspond to at least one of a plurality of nearest neighbor patches.
1 . A method of processing an image in a video, comprising:
decoding an encoded version of the image to generate a decoded version of the image; and
generating a de-noised version of the image using the decoded version of the image and a video image epitome which is a texture epitome associated with the image, wherein the video image epitome was extracted from a source version of the image,
wherein the generating comprises:
de-noising a current patch using corresponding patches located in the video image epitome that correspond to at least one of a plurality of nearest neighbor patches.
2 . The method of claim 1 , wherein the de-noising comprises performing a Non Local Means method of de-noising using the video image epitome.
3 . The method of claim 2 , wherein the de-noising comprises setting a filtering parameter by estimating a noise level as the mean squared error between the image epitome patches and the corresponding noisy patches, wherein the filtering parameter is set as a product of the estimated noise level and a pre-defined user parameter.
4 . The method of claim 1 , wherein the de-noising comprises using a method that includes a hard thresholding step and a Wiener filtering step.
5 . The method of claim 4 , wherein the hard thresholding step comprises adaptively choosing the threshold by: performing a 3D transform on a group of noisy patches and their corresponding image epitome patches; determining a thresholding rule between the transformed patches; substituting the current patch in a patch of the group of noisy patches; applying the thresholding rule to the group of noisy patches including the current patch, and performing an inverse transform to generate a first de-noised version of the current patch.
6 . The method of claim 5 , wherein the first de-noised version of the current patch is used as oracle for the Wiener filtering step.
7 . The method of claim 1 , wherein the video image epitome and the encoded version of the image are accessed via a bitstream received over a communications channel, and wherein the video image epitome is encoded, and the bitstream includes a flag indicating that the video image epitome is included with the encoded version of the image.
8 . An apparatus for processing an image of a video, comprising:
a communications interface configured to access an encoded version of the image and generating a decoded version of the image, and a video image epitome which is a texture epitome associated with the image, wherein the video image epitome was extracted from a source version of the image;
a processor, coupled to the communications interface, and configured to generate an output for display including a de-noised version of the decoded image using the decoded version of the video and the video image epitome, and
wherein the image epitome and the processor is configured to generate a de-noised version of the decoded image by:
de-noising a current patch using the corresponding patches located in the video image epitome that correspond to at least one of a plurality of nearest neighbor patches.
9 . The apparatus of claim 8 , wherein the processor is configured to perform a Non Local Means method of de-noising using the video image epitome.
10 . The apparatus of claim 8 , wherein the processor is configured to set a filtering parameter by estimating a noise level as the mean squared error between the image epitome patches and the corresponding noisy patches, wherein the filtering parameter is set as a product of the estimated noise level and a pre-defined user parameter.
11 . The apparatus of claim 10 , wherein the processor is configured to de-noise by using a method that includes a hard thresholding step and a Wiener filtering step.
12 . The apparatus of claim 11 , wherein the processor is configured to perform the hard thresholding step by adaptively choosing the threshold by: performing a 3D transform on a group of noisy patches and their corresponding image epitome patches; determining a thresholding rule between the transformed patches; substituting the current patch in a patch of the group of noisy patches; applying the thresholding rule to the group of noisy patches including the current patch, and performing an inverse transform to generate a first de-noised version of the current patch.
13 . The apparatus of claim 12 , wherein the first de-noised version of the current patch is used as oracle for the Wiener filtering step.