Patent №
US 6,714,693
Granted
2004-03-30
Filed 2000
Owner
CANON KABUSHIKI KAISHA
Lab
—
AI components
1
vision
Assignment
Recorded
Dataset
AIPD
2023_r1 edition
Application
09575243
The present invention uses an image processing apparatus for inputting low-resolution video information, expanding one pixel to N×M pixels, and converting a low-resolution image into a high-resolution image, which comprises a first window formation unit for referring to peripheral pixels including a noticeable pixel, plural conversion units different from each other in conversion method for converting one pixel to N×M pixels, and a selection unit for selecting a proper conversion method out of the plural conversion units in accordance with a distribution state of pixels in a first window and in which the conversion units calculates conversion values for N×M pixels of the noticeable pixel in accordance with an evaluation result of the type of a selected conversion unit in adjacent pixels including the noticeable pixel in an evaluation unit. Thereby, when converting low-resolution information into high-resolution information, it is possible to realize a preferable conversion free from jaggies without causing halation due to interpolation and generate high-resolution information in which continuity is not broken at a boundary point for each conversion in conversions provided with nonlinearity.
AI classification
Ownership
CANON KABUSHIKI KAISHA
assignment · 112210817
Assignors
MIYAKE, NOBUTAKA
On an employer assignment, the assignors are typically the inventors.