DETERMINING IF AN IMAGE IS BLURRED

Patent №

US 7,929,797

Granted

2011-04-19

Filed 2007

Owner

SEIKO EPSON CORPORATION

Lab

AI components

1

vision

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

11809609

A technology for making determinations as to whether the image of an image data file is blurred, with low processing load, is provided. First, photographic image data PID containing pixel color information, as well as thumbnail image data for the photographic image data, are prepared. A specific pixel is then selected from among the pixels of the thumbnail image data. Specifically, on the basis of data of a pixel under examination and data of a pixel neighboring the pixel under examination, it is determined whether the pixel under examination should be designated as the specific pixel. Then, for a specific region Ape which is a region constituting part of the photographic image data PID and which corresponds to the specific pixel of the thumbnail image data, analysis of the photographic image PID is carried out, and a localized blur level that represents indistinctness of the image in the specific region Ape is calculated. Subsequently, in the event that the localized blur level of the specific region Ape contained in the photographic image data PID meets a prescribed condition, it is determined that the image of the photographic image data PID is blurred.

VisionG06K 15/02G06K 15/1822G06K 15/186G06T 7/13G06T 7/168H04N 1/00175H04N 1/00188H04N 1/00132

AI classification

Vision1.00
Knowledge representation0.03
Evolutionary computation0.02
Planning0.00
Natural language0.00
AI hardware0.00
Machine learning0.00
Speech0.00

Ownership

SEIKO EPSON CORPORATION

assignment · 194350059

Assignors

MATSUZAKA, KENJI

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

From the same owner

© 2026 NYSGPT2525 LLC