# US 2014/0193086 A1 · Image processing methods and apparatuses > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2014/0193086 A1 - Document ID: US20140193086A1 - Status: Published application - Kind: A1 - Application number: 14/146,263 - Filed: 2014-01-02 - Published: 2014-07-10 - Earliest claimed: 2013-01-10 - Assignees: Nuctech Company Limited; Tsinghua University - Applicants: Nuctech Company Limited; Tsinghua University - Inventors: Li Zhang; Zhiqiang Chen; Jia Hao - Web page: https://patentgrep.com/patent/US20140193086A1 ## CPC - G06T 5/50: using two or more images, e.g. averaging or subtraction - G06T 2211/408: Dual energy - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract Provided are image processing method and apparatus. The method includes selecting one of a plurality of images matched in structure as reference image, and select another of the images as subject image; for a subject pixel in the subject image, determining a pixel corresponding to the subject pixel in the reference image; calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel; establishing weight coefficients based on the similarity values, and weighted averaging the subject pixel in the subject image to obtain a processed pixel value. With the above solutions, it is possible to use structure information of a higher-quality image in processing another image, and thus improve quality of the other image. ## Claims ### Claim 1 An image processing method comprising: selecting one of a plurality of images matched in structure as reference image, and selecting another of the images as subject image; for a subject pixel in the subject image, determining a pixel corresponding to the subject pixel in the reference image; calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel; establishing weight coefficients based on the similarity values, and weighted averaging the subject pixel in the subject image to obtain a processed pixel value. ### Claim 2 (depends on 1) The image processing method of claim 1, wherein said calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: calculating, as similarity value, a Gaussian weighted Euclidean distance between each of image blocks where the at least part of pixels in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 3 (depends on 1) The image processing method of claim 1, wherein said calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: calculating, as similarity value, a structure similarity index between each of image blocks where the at least part of pixels in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 4 (depends on 2) The image processing method of claim 2, wherein if a difference between an average of pixels in an image block where certain pixel in the reference image is located and an average of pixels in the image bock where the subject pixel is located is greater than a predetermined threshold, no similarity value is calculated for the two pixels. ### Claim 5 (depends on 1) The image processing method of claim 1, wherein a specific region containing the subject pixel is selected from the subject image, and said calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: calculating, as similarity value, a Gaussian weighted Euclidean distance between each of image blocks where pixels in a corresponding specific region in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 6 (depends on 1) The image processing method of claim 1, wherein a specific region containing the subject pixel is selected from the subject image, and said calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: calculating, as similarity value, a structure similarity index between each of image blocks where pixels in a corresponding specific region in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 7 (depends on 5) The image processing method of claim 5, wherein if a difference between an average of pixels in an image block where certain pixel in the corresponding specific region in the reference image is located and an average of pixels in the image bock where the subject pixel is located is greater than a predetermined threshold, no similarity value is calculated for the two pixels. ### Claim 8 (depends on 1) The image processing method of claim 1, wherein said determining a pixel corresponding to the subject pixel in the reference image comprises: determining the corresponding pixel in the reference image based on location information of the subject pixel in the subject image. ### Claim 9 (depends on 1) The image processing method of claim 1, wherein the reference image has higher image quality than the subject image, and the image quality comprises one of signal-to-noise ratio, sharpness, contrast, resolution, and definition. ### Claim 10 (depends on 1) The image processing method of claim 1, wherein the reference image comprises a high-energy or low-energy attenuation coefficient image obtained by a dual-energy CT imaging system, and the subject image comprises an atomic number image or an electron density image. ### Claim 11 (depends on 10) The image processing method of claim 10, wherein the high-energy or low-energy attenuation coefficient image is used to correct metal artifact in the electron density image or the atomic number image. ### Claim 12 (depends on 1) The image processing method of claim 1, wherein the reference image comprises a full-sample image, and the subject image comprises a sparse-sample image. ### Claim 13 (depends on 12) The image processing method of claim 12, wherein the full-sample image is used to correct aliasing artifact in the sparse-sample image. ### Claim 14 (depends on 1) The image processing method of claim 1, wherein a pixel having a higher similarity value corresponds to larger weight, and a pixel having a lower similarity value corresponds to smaller weight. ### Claim 15 (depends on 1) The image processing method of claim 1, wherein the reference image comprises CT image, and the subject image comprises an image obtained from PET or SPECT modality. ### Claim 16 An image processing apparatus comprising: means for selecting one of a plurality of images matched in structure as reference image, and selecting another of the images as subject image; means for determining, for a subject pixel in the subject image, a pixel corresponding to the subject pixel in the reference image; means for calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel; means for establishing weight coefficients based on the similarity values, and weighted averaging the subject pixel in the subject image to obtain a processed pixel value. ### Claim 17 (depends on 16) The image processing apparatus of claim 16, wherein said means for calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: means for calculating, as similarity value, a Gaussian weighted Euclidean distance between each of image blocks where the at least part of pixels in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 18 (depends on 16) The image processing apparatus of claim 16, wherein said means for calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: means for calculating, as similarity value, a structure similarity index between each of image blocks where the at least part of pixels in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 19 (depends on 16) The image processing apparatus of claim 16, wherein a specific region containing the subject pixel is selected from the subject image, and said means for calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: means for calculating, as similarity value, a Gaussian weighted Euclidean distance between each of image blocks where pixels in a corresponding specific region in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 20 (depends on 16) The image processing apparatus of claim 16, wherein a specific region containing the subject pixel is selected from the subject image, and said means for calculating similarity values of at least part of pixels in the reference image with respect to the pixel corresponding to the subject pixel comprises: means for calculating, as similarity value, a structure similarity index between each of image blocks where pixels in a corresponding specific region in the reference image are respectively located and an image block corresponding to an image block where the subject pixel is located. ### Claim 21 (depends on 16) The image processing apparatus of claim 16, wherein said means for determining a pixel corresponding to the subject pixel in the reference image comprises: means for determining the corresponding pixel in the reference image based on location information of the subject pixel in the subject image.