REMOVAL OF NOISE IN IMAGE AREA Russian patent published in 2017 - IPC G06T5/00 G06K9/40 

FIELD: physics.

SUBSTANCE: data processing element of the reconstructed volumetric image from the imaging scanner contains: an algorithm memory including one or more iterative noise elimination algorithms only in the image area based on minimizing the Huber penalty function; a processor eliminating noise in the reconstructed volumetric image data exclusively in the image area based on one of the iterative minimization algorithms of the Huber penalty function that includes a first parameter that is a free variable and in the first expression determines the noise elimination level and a second parameter that is a free variable and in the second expression determines the homogeneity of noise elimination. The algorithm includes a first expression that is based on minimizing the Huber penalty function and which includes the product of the first parameter and the weighted sum of neighboring voxels with respect to the voxel adjusted with respect to the size that determines the level of noise elimination.

EFFECT: to ensure the elimination of noise in the image data.

14 cl, 5 dwg