Affiliations: Department of Computer Science and Information
Engineering, National Chung Cheng University, Chiayi, Taiwan 62107, R.O.C.
E-mail: mnwu@cs.ccu.edu.tw | Department of Computer Science and Information
Management, Providence University, Taichung, Taiwan 433, R.O.C.
E-mail: mhlin3@pu.edu.tw | Department of Information Engineering and Computer
Science, Feng Chia University, Taichung 407, Taiwan, R.O.C.
E-mail: ccc@cs.ccu.edu.tw
Abstract: Content-based image copy detection can be used as either an
alternative approach or a complementary approach to watermarking. To detect
copied versions of digital images, which are generated by various image
processes, several schemes have been proposed. However, none of them can
successfully detect tampering when those images are rotated 90 or 270 degrees
from the original image. To solve this problem, we propose a new copy detection
scheme based on Y component of image and block strategy. A test image is just
converted to a YUV color model where only Y component is used to generate the
image signatures. Then, the transformed image is divided into 8×8 blocks.
Finally, the relationship between each block and its neighboring eight blocks
are recorded by comparing these block's mean values. This image signature
generated by our proposed scheme contains the comparative relationship between
each block and its neighboring blocks. Our proposed scheme can detect
modifications those are generated by 90 or 270 rotation of original one, which
can not be detected by Kim's scheme. The optimal block numbers used for
image partition have been explored and the efficiency of the proposed scheme is
extensively tested with 20,000 test images.
Keywords: Content-based image copy detection, copyright protection, image signature, mean value
