Image compounding based on independent noise constraint
AssigneeSiemens Medical Solutions USA, Inc.
InventorsYunqiang Chen, Hongcheng Wang, Tong Fang, Jason Tyan
Life of this patent
Estimating the term…
Loading patent…
AssigneeSiemens Medical Solutions USA, Inc.
InventorsYunqiang Chen, Hongcheng Wang, Tong Fang, Jason Tyan
Estimating the term…
A method and system for improving image quality by compounding a plurality of images to mitigate the effects of image noise. The method utilizes the independency between noise components for multiple image compounding. An effective measurement is designed to regularize the independency between noise in a traditional generative model based filtering framework, thereby enabling a more robust algorithmic solution to inaccurate signal/noise modeling. The method generally comprises selecting a plurality of images, calculating the residual error on each image; calculating the noise likelihood of each image, calculating the signal likelihood of the image, performing an independence analysis to regularize an independence constraint between the residual errors of the images, and summing the signal likelihood, noise likelihood and pairwise independency to approximate the joint independency between the residual errors.
A method for improving image quality by compounding a plurality of images, comprising the steps of:
A memory medium containing program instructions stored in the memory medium which, when executed by a processor, enable a computer to improve image quality by compounding a plurality of images, by:
A system for improving image quality by compounding a plurality of images, comprising:
Dependencies are inferred from the claim text.