# US 9,883,079 B2 · Method for noise-robust color changes in digital images > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 9,883,079 B2 - Document ID: US9883079B2 - Status: Granted - Kind: B2 - Application number: 15/214,238 - Filed: 2016-07-19 - Published: 2016-11-10 - Granted: 2018-01-30 - Earliest claimed: 2007-06-26 - Assignees: Google Inc. - Applicants: Google Inc. - Inventors: Nils Kokemohr - Web page: https://patentgrep.com/patent/US9883079B2 ## CPC - H04N 1/58: Edge or detail enhancement; Noise or error suppression, e.g. colour misregistration correction (H04N1/62 takes precedence) - G06T 5/20: using local operators - G06T 5/70: Denoising; Smoothing - G06T 11/10: Texturing; Colouring; Generation of textures or colours (retouching, inpainting or scratch removal G06T5/77) - H04N 1/4092: {Edge or detail enhancement} - H04N 1/62: Retouching, i.e. modification of isolated colours only or in isolated picture areas only - G06T 2207/10024: Color image - G06T 2207/20004: Adaptive image processing - G06T 2207/20024: Filtering details - G06T 2207/20192: Edge enhancement; Edge preservation - G06T 2207/20221: Image fusion; Image merging - H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A method for applying filters to digital images with minimal amplification of image noise, comprising filtering the digital image with an EPDR edge-preserving detail-reducing filter, determining a matrix from the filtered image as a result of one or more structure adaptive functions, and modifying the digital image using the filter, adjusted by the matrix values, to produce an enhanced digital image. The order of processing may be inverted, by first determining the matrix and then filtering the matrix with the edge-preserving detail-reducing filter. ## Claims ### Claim 1 A computer-implemented method comprising: filtering a digital image with an edge-preserving detail-reducing (EPDR) filter, wherein the digital image comprises a plurality of pixels; generating a first image by filtering the EPDR-filtered digital image with an image-processing filter that includes a structure-adaptive filter, wherein the structure-adaptive filter provides an intensity of a filter effect on a plurality of pixels of the EPDR-filtered digital image based on a respective geometric distance of each respective pixel of the plurality of pixels of the EPDR-filtered digital image to a reference location in the EPDR-filtered digital image, wherein the reference location is one of a reference pixel and a reference region; generating a second image by filtering the EPDR-filtered digital image with a highpass filter; and producing a third image with a suppressed noise level by adding generated the first image and generated the second image. ### Claim 2 (depends on 1) The method of claim 1, wherein the EPDR filter is selected from the group consisting of an arctan blurring filter, a wavelet-based detail-reduction filter, a bilateral filter, a Beltrami flow filter, an adaptive Gaussian blurring filter, and an edge-preserving smoothing filter using geodesic distances. ### Claim 3 (depends on 1) The method of claim 1, wherein first pixels of the EPDR-filtered digital image having a particular color and located closer to the reference location are modified to a greater degree by the structure-adaptive filter than second pixels having the particular color and that are located farther from the reference location. ### Claim 4 (depends on 1) The method of claim 1, wherein the reference location includes the reference region, wherein the reference region has user-defined boundaries. ### Claim 5 (depends on 1) The method of claim 1, wherein the reference location is the reference pixel, and wherein the structure-adaptive filter further provides the filter effect on each respective pixel of the plurality of pixels based on a color similarity between a color of the respective pixel in the EPDR-filtered digital image and a reference color of the reference pixel. ### Claim 6 (depends on 1) The method of claim 1, wherein the reference location is the reference pixel, and wherein the structure-adaptive filter scales a color change at the plurality of pixels based on the respective geometric distances of the plurality of pixels to the reference pixel. ### Claim 7 (depends on 1) The method of claim 1, wherein the structure-adaptive filter provides a plurality of different filter effects at a plurality of corresponding different locations in the EPDR-filtered digital image, wherein the structure-adaptive filter provides a respective intensity of the different filter effects on the respective corresponding different locations based on respective geometric distances of the respective corresponding different locations to the reference location in the EPDR-filtered digital image. ### Claim 8 (depends on 5) The method of claim 5, wherein the filter effect based on the color similarity is further based on a maximum color change selected by a user. ### Claim 9 (depends on 1) The method of claim 1, wherein the reference location in the EPDR-filtered digital image is determined based on user input from a user. ### Claim 10 A system comprising: one or more memory devices; one or more processors; and logic encoded in the one or more memory devices for execution by the one or more processors and when executed operable to perform operations comprising: filtering a digital image with an edge-preserving detail-reducing (EPDR) filter, wherein the digital image comprises a plurality of pixels; generating a first image by filtering the EPDR-filtered digital image with an image-processing filter that includes a structure-adaptive filter, wherein the structure-adaptive filter is configured to provide an intensity of a filter effect on a plurality of pixels of the EPDR-filtered digital image based on a respective color similarity between a color of each respective pixel of the plurality of pixels of the EPDR-filtered digital image and a reference color of a reference pixel in the EPDR-filtered digital image; generating a second image by filtering the EPDR-filtered digital image with a highpass filter; and producing a third image with a suppressed noise level by adding generated the first image and generated the second image. ### Claim 11 (depends on 10) The system of claim 10, wherein the structure-adaptive filter provides the intensity of the filter effect on each respective pixel of the plurality of pixels based additionally on a respective geometric distance of each respective pixel to the reference pixel in the EPDR-filtered digital image. ### Claim 12 (depends on 10) The system of claim 10, wherein the reference pixel in the EPDR-filtered digital image is determined based on user input from a user. ### Claim 13 (depends on 11) The system of claim 11, wherein first pixels of the EPDR-filtered digital image having a particular color and located closer to the reference pixel are modified to a greater degree by the structure-adaptive filter than second pixels having the particular color and that are located farther from the reference pixel. ### Claim 14 (depends on 10) The system of claim 10, wherein the structure-adaptive filter scales a color change at the plurality of pixels based on the respective geometric distances of the plurality of pixels to the reference pixel. ### Claim 15 (depends on 10) The system of claim 10, wherein the color similarity includes similarity of at least one of: brightness, luminosity, saturation, hue, and color temperature. ### Claim 16 (depends on 10) The system of claim 10, wherein the EPDR filter is selected from the group consisting of an arctan blurring filter, a wavelet-based detail-reduction filter, a bilateral filter, a Beltrami flow filter, an adaptive Gaussian blurring filter, and an edge-preserving smoothing filter using geodesic distances. ### Claim 17 A non-transitory computer readable medium having stored thereon program instructions that, when executed by a processor, cause the processor to perform operations including: filtering a digital image with an edge-preserving detail-reducing (EPDR) filter, wherein the digital image comprises a plurality of pixels; generating a first image by filtering the EPDR-filtered digital image with an image-processing filter that includes a structure-adaptive filter, wherein the structure-adaptive filter provides an intensity of a filter effect on a plurality of pixels of the EPDR-filtered digital image based on a respective geometric distance of the each respective pixel of the plurality of pixels of the EPDR-filtered digital image to a reference pixel in the EPDR-filtered digital image; generating a second image by filtering the EPDR-filtered digital image with a highpass filter; and producing a third image with a suppressed noise level by adding generated the first image and generated the second image. ### Claim 18 (depends on 17) The computer readable medium of claim 17, wherein the reference pixel is determined based on user input from a user. ### Claim 19 (depends on 17) The computer readable medium of claim 17, wherein the EPDR filter is selected from the group consisting of an arctan blurring filter, a wavelet-based detail-reduction filter, a bilateral filter, a Beltrami flow filter, an adaptive Gaussian blurring filter, and an edge-preserving smoothing filter using geodesic distances. ### Claim 20 (depends on 17) The computer readable medium of claim 17, wherein the structure-adaptive filter further provides the filter effect on each respective pixel of the plurality of pixels based on a color similarity between a color of the respective pixel and a reference color of the reference pixel.