Roughness-adaptive three-dimensional visuohaptic watermarking

Kwangtaek Kim, Purdue University

Abstract

As contents for visuohaptic rendering of virtual objects continue to be developed, the need will soon arise for the protection of visuohaptic media from illegal reproduction. Until very recently, digital watermarking has been concerned with only visual and audio contents. Several studies on visuohaptic watermarking demonstrated that the haptic sensory channel is sometimes more sensitive to watermarks embedded in a triangular mesh data structure than the visual channel. This attests to the need to extend current state-of-the-art visual 3D watermarking techniques to visuohaptic algorithms where the sensorimotor capabilities of the haptic channel are taken into account. This dissertation presents a series of studies towards the development of roughness-adaptive visuohaptic 3D watermarking techniques. In the first study, a roughness-adaptive visual watermarking scheme is presented that adaptively maximizes watermark strength based on local surface roughness. It takes advantage of the fact that a rougher surface patch can mask a stronger watermark than a smoother surface patch. A visual perception study was conducted to quantify watermark perceptibility as a function of surface roughness. It was found that the maximum watermark strength for invisibility is linearly related to measured surface roughness: the rougher the host surface, the stronger the watermark that can be successfully hidden. The stronger watermarks in turn contribute to improved robustness of watermarks. In the second study, the roughness-adaptive watermarking approach was extended to include the haptic sensory modality. A perception study was conducted to estimate watermark detection thresholds under vision alone, touch alone and visuohaptic conditions. The results showed that watermark detection threshold was determined by the visual channel (i.e., vision is more sensitive) at rougher surface patches, and by the haptic channel (i.e., touch is more sensitive) at smoother surface patches. The third study sought to equalize the extent of shading during the visual and haptic rendering of 3D objects, in order to achieve a more general approach to visuohaptic watermarking. A new approach based on spherigon smoothing is adapted to up-sample 3D triangular meshes to achieve the same smoothing for both visual and haptic rendering. A human perception study showed that watermark strength can be increased by 5 to 26 times depending on local surface roughness while preserving imperceptibility. The increase in watermark strength led to 309% increase in robustness as compared with the second study. Finally, we examined the potential of inserting watermarks by altering the gross shape of 3D objects as opposed to surface details. It was found that vision is dominant in 3D gross shape discrimination in that the visuohaptic thresholds were similar to the visual thresholds. These studies exemplify the use of human perception data in specifying watermark strengths. The findings form the foundation for future development of 3D visuohaptic watermarking schemes that can have exciting applications in cultural heritage preservation, entertainment and commerce.

Degree

Ph.D.

Advisors

Tan, Purdue University.

Subject Area

Electrical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS