부산대학교 도서관

High dynamic range images are shared widely online, however they might be used improperly. Some HDR image-based watermarking (HDR-IW) techniques were proposed to secure these images. They did, however, inherit the same issue found in the traditional IW approaches for standard dynamic range images, in which only certain trade-offs among robustness, imperceptibility, and capacity are achieved. To con-currently improve robustness, imperceptibility, and payload, a novel saliency (eye-catching region) detection-based trade-off independent HDR-IW is proposed in this study. The host HDR image is first converted into an SDR image and the salient objects are detected to create a saliency map, which is then used to separate the host image's foreground and background. With the use of the same mask, the watermark (binary image) is then split into the foreground and background regions. Next, both regions are encrypted using multiple homomorphic algorithms. Subsequently, discrete wavelet and lifting transforms followed by the fast Fourier transforms are applied to the host image to generate the real parts of the subbands. Finally, the first 4 bits of each coefficient in the LL and HH subbands of the host image are utilized to embed watermark bits. Experiment results suggest that the proposed approach performs better at resolving competing criteria than cutting-edge techniques.