Clark School Home UMD
Return to Patents

Human Visual Model for Data Hiding

Abstract 

A method and apparatus of hiding identification data in visual media. When image or video data is received, frequency masking is performed to divide the image or video data into blocks of smooth regions and blocks of non-smooth regions and to obtain preliminary just-noticeable-difference. Edge detection is performed to divide the non-smooth region of the image or video data into texture blocks and edge blocks. Then blocks of regions that are substantially proximate to blocks of smooth regions of the image or video data are determined. The image or video data is then adjusted by applying different strength of watermark in association with the type of each block.

Inventor(s)

Hong Heather Yu

Date Issued 

08/26/2003

Patent No. 

6,611,608