Palette-based image protection via shape self-similar subdivision
Yu-Kuen Ho, Mei-Yi Wu, Jia-Houg Lee · 2005
Watermarking techniques have been widely discussed and developed for image protection in recent years. However, there are still very few studies to focus on watermarking palette images. This study proposes a new watermarking scheme for protecting palette images especially cartoon images using eflcient shape segmentation in the sparial domain. The lengths and orientations of two principal axes of segmented image objects were used for image normalization and shape segmentation. Watermarks then are embedded into object images by replacing the palette indices of pixels according to the gradient computation on each pixel. Experimental results reveal that the proposed method effectiveb resists geometrical attach such as rotation, scaling and handation and also against general imagefiltering attacks without using the original image.