SYSTEM FOR RECOVERY OF DEGRADED IMAGES

Patent №

US 7,729,010

Granted

2010-06-01

Filed 2006

Owner

UNIVERSITY OF ROCHESTER

Lab

AI components

2

vision · kr

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

11501207

A system for recovering degraded images captured through atmospheric turbulence, or other atmospheric inhomogeneities, such as snow, rain, smoke, fog, or underwater fluctuations, is provided having an imager for capturing through such turbulence both a degraded image of a scene having at least one object, and an image of a point source associated with the object. The imager converts the degraded image into first image data signals representing the degraded image, and converts the image of the point source into second image data signals representing a point spread function. A computer of the system receives the first and second image data signals and produces third image data signals representing a recovered image of the object of the degraded image in accordance with the first and second image data signals. In another embodiment, the imager captures a degraded image through atmospheric turbulence of a scene having a known reference object and an unknown object, and converts the degraded image into first image data signals. After receiving the first image data signals, the computer identifies in the first image data signals such image data signals representing the reference object. The computer produces image data signals representing a recovered image of the degraded image in accordance with the first image data signals, the image data signals representing the reference in the first image data signals, and image data signals representing an undegraded image of the reference. The computer may output the image data signals representing a recovered image to an output device to display or print the recovered image.

VisionKnowledge representationG06T 5/73G06T 5/10G06T 5/50G06T 2207/20056

AI classification

Vision1.00
Knowledge representation0.99
Planning0.24
Evolutionary computation0.02
Natural language0.00
Machine learning0.00
AI hardware0.00
Speech0.00

Ownership

UNIVERSITY OF ROCHESTER

assignment · 242200522

Assignors

GEORGE, NICHOLAS, DR.

On an employer assignment, the assignors are typically the inventors.

From the same owner

© 2026 NYSGPT2525 LLC