| 摘要 |
PURPOSE: An image watermarking method using an HVS(Human Visual System) is provided to insert a watermark into a motion picture according to a global masking method for combining plural individual masking methods, thereby satisfying invisibility, robustness, and security. CONSTITUTION: A frame of a test motion picture is inputted(S600). A process of performing a frequency masking is performed as follows. The frame is divided into blocks(S610). A DCT conversion is performed(S620), and a masking value in a frequency area is obtained(S630). By carrying out an IDCT, a frequency masking value is obtained in a spatial area(S640). A process of performing a spatial masking is carried out as follows. A contrast of an image is increased(S611), and an outline of the frame is extracted(S621), to obtain a spatial masking value(S631). A process of performing a motion masking is conducted as follows. A luminosity difference between a current frame pixel and a previous frame pixel is obtained(S612), and an outline of the frame is extracted(S622), then a motion masking value is obtained(S632). Each masking result is combined(S650), and a watermark value is weighted(S660), then an original frame is added(S670). A PSNR value of a watermark-inserted frame is compared with a target PSNR(S680), to properly control a control constant(S690). |
