In recent years, streaming content services through OTT (Over The Top) platforms have become a popular form of media consumption and continue to grow. However, the widespread use of these services has led to various copyright protection issues, with illegal piracy becoming a serious problem. To protect content from illegal reproduction, DRM and watermarking technologies are used; however, illegal distributors evade these protections by altering the content’s angle or size before uploading it. In this study, feature extraction algorithms such as HCD, SIFT, ORB, and AKAZE were applied to evaluate watermark restoration performance on geometrically transformed, rotated images. The watermark was inserted by selecting the frequency components with the lowest energy after applying DWT to the original image, and the restoration performance was compared at different rotation angles (0, 10, 30, 60, 90 degrees). Performance evaluation metrics, including SSIM and PSNR, were used to assess the structural similarity and image quality loss, while NC and BER values were used to measure restoration accuracy.
The experimental results showed that the AKAZE algorithm achieved the highest watermark insertion and restoration accuracy. This study presents a method for addressing content rotation transformations and demonstrates practical applicability in real-time streaming services and content protection technologies.