# US 2020/0005440 A1 · Systems and methods for reducing low-frequency non-uniformity in images

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0005440 A1
- Document ID: US20200005440A1
- Status: Published application
- Kind: A1
- Application number: 16/568,139
- Filed: 2019-09-11
- Published: 2020-01-02
- Earliest claimed: 2017-03-15
- Assignees: FLIR Systems, Inc.; Noiseless Imaging Oy LTD
- Applicants: FLIR Systems, Inc.; Noiseless Imaging Oy Ltd.
- Inventors: Enrique Sanchez-Monge; Alessandro Foi
- Web page: https://patentgrep.com/patent/US20200005440A1

## CPC

- G06T 5/94: based on local image properties, e.g. for local contrast enhancement
- G06T 3/40: Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T 5/10: using non-spatial domain filtering
- G06T 5/73: Deblurring; Sharpening
- G06T 2207/10048: Infrared image
- H04N 25/61: the noise originating only from the lens unit, e.g. flare, shading, vignetting or "cos4"
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

Various techniques are disclosed for separating and removing low-frequency shadow or shading (also referred to herein as “non-uniformity”) from images that have been corrupted by the non-uniformity. A non-uniformity estimate that approximates the non-uniformity effect on the corrupted image may be generated by iteratively adding new blotches of non-uniformity data represented by two-dimensional (2D) functions, such as 2D Gaussian functions, to the non-uniformity estimate and applying filters to smoothen the 2D functions. In each iteration of the non-uniformity estimate generation process, a new non-uniformity update candidate that minimizes a cost function is identified. The corrupted image is processed based on the non-uniformity estimate to generate a corrected image.

## Claims

### Claim 1

A method of correcting non-uniformity in the form of low frequency shading corrupting an image, the method comprising:
generating a non-uniformity estimate for the image by iteratively:
estimating a new non-uniformity update that minimizes a cost function based on a total variation among pixel values of the image;
updating the non-uniformity estimate by adding the new non-uniformity update to the non-uniformity estimate; and
applying the non-uniformity estimate to the image to obtain a corrected image with the low frequency shading suppressed or removed.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
resizing the image from an original size to a smaller size before generating the non-uniformity estimate to obtain a resized version of the image, wherein the generating the non-uniformity estimate is performed with respect to the resized version of the image; and
resizing the generated non-uniformity estimate to the original size before applying the non-uniformity estimate to the image.

### Claim 3 (depends on 2)

The method of claim 2, further comprising:
rescaling pixel values of the resized version of the image from an original scale to a smaller scale before generating the non-uniformity estimate to obtain a resized version of the image; and
resealing values of the non-uniformity estimate from the smaller scale to the original scale before applying the non-uniformity estimate to the image.

### Claim 4 (depends on 1)

The method of claim 1, further comprising applying, to the image, a transformation that takes into account an additive nature and a multiplicative nature of the low frequency shading to generate a transformed image, wherein the generating and the applying of the non-uniformity estimate are performed with respect to the transformed image.

### Claim 5 (depends on 4)

The method of claim 4, wherein the transformation comprises a parameter that balances an additive nature of the low frequency shading and a multiplicative nature of the low frequency shading.

### Claim 6 (depends on 4)

The method of claim 4, further comprising, after applying the non-uniformity estimate to the transformed image, performing an inverse of the transformation to the corrected image.

### Claim 7 (depends on 1)

The method of claim 1, wherein the cost function is further based on an energy level of a gradient of the non-uniformity estimate.

### Claim 8 (depends on 7)

The method of claim 7, wherein the cost function further comprises a weight associated with the total variation among pixel values of the processed image and a weight associated with the energy level of the gradient of the new non-uniformity update.

### Claim 9 (depends on 1)

The method of claim 1, wherein estimating the new non-uniformity update comprises:
determining a non-uniformity update candidate for the processed image; and
updating the non-uniformity estimate with the determined non-uniformity update only when the identified non-uniformity update reduces a cost calculated for the non-uniformity estimate based on the cost function; and
wherein determining the non-uniformity update candidate comprises selecting a global filter, wherein updating the non-uniformity estimate comprises applying the global filter to the non-uniformity estimate; and wherein applying the non-uniformity estimate to the image comprises subtracting the non-uniformity estimate from the image.

### Claim 10 (depends on 1)

The method of claim 1, wherein estimating the new non-uniformity update comprises:
determining a non-uniformity update candidate for the processed image; and
updating the non-uniformity estimate with the determined non-uniformity update only when the identified non-uniformity update reduces a cost calculated for the non-uniformity estimate based on the cost function; and
wherein determining the non-uniformity update candidate comprises selecting a two-dimensional 2D Gaussian function having a location and a peak amplitude, and wherein updating the non-uniformity estimate comprises adding the 2D Gaussian function to the non-uniformity estimate.

### Claim 11 (depends on 10)

The method of claim 10, wherein determining the non-uniformity update candidate comprises selecting a spatial location within the processed image as the location for the 2D Gaussian function.

### Claim 12 (depends on 10)

The method of claim 10, wherein determining the non-uniformity update candidate comprises selecting a spatial location in or around a corner of the processed image as the location for the 2D Gaussian function.

### Claim 13 (depends on 10)

The method of claim 10, wherein determining the non-uniformity update candidate comprises selecting a location and a peak amplitude for the 2D Gaussian function that minimize the cost function.

### Claim 14

A system for correcting non-uniformity in the form of low-frequency shading corrupting an image, the system comprising:
an image interface configured to receive the image; and
a processor communicatively coupled with the image interface and configured to:
generate a non-uniformity estimate for the image by iteratively:
estimate a new non-uniformity update that minimizes a cost function based on a total variation among pixel values of the image;
update the non-uniformity estimate by adding the new non-uniformity update to the non-uniformity estimate; and
apply the non-uniformity estimate to the image to obtain a corrected image with the low frequency shading suppressed or removed.

### Claim 15 (depends on 14)

The system of claim 14, wherein the processor is further configured to apply, to the image, a transformation that takes into account an additive nature and a multiplicative nature of the low frequency shading to generate a transformed image, wherein the generating and the applying of the non-uniformity estimate are performed with respect to the transformed image.

### Claim 16 (depends on 14)

The system of claim 14, wherein the cost function is further based on an energy level of a gradient of the non-uniformity estimate.

### Claim 17 (depends on 16)

The system of claim 16, wherein the cost function further comprises a weight associated with the total variation among pixel values of the processed image and a weight associated with the energy level of the gradient of the new non-uniformity update.

### Claim 18 (depends on 14)

The system of claim 14, wherein the processor is configured to estimate the new non-uniformity update at least by:
determining a non-uniformity update candidate for the processed image; and
updating the non-uniformity estimate with the determined non-uniformity update only when the identified non-uniformity update reduces a cost calculated for the non-uniformity estimate based on the cost function.

### Claim 19 (depends on 14)

The system of claim 14, wherein the processor is further configured to:
perform, to images received from the image interface, a series of image processes in a main signal path, wherein the generating of the non-uniformity estimate for the image is performed in a background processing path outside of the main signal path; and
apply the non-uniformity estimate to the images received from the image interface as part of the series of image processes in the main signal path.

### Claim 20 (depends on 14)

The system of claim 14, further comprising an infrared imaging sensor configured to capture an infrared image to provide to the image interface, wherein the image comprises the captured infrared image.
