Quaternion-based deformation of cloth point clouds and its verification

Reconstructing detailed, high-resolution cloth point clouds is a significant challenge in computer graphics and computer vision, often hindered by the noise and gaps present in single-frame captures. Furthermore, modeling the complex dynamics of non-rigid cloth deformation remains difficult. As an initial step toward a novel method for deforming and integrating cloth point clouds across multiple frames, this paper presents a deformation approach based on dual quaternions (DQ). Our methodology utilizes Dual Quaternion Blending (DQB), a technique that models non-rigid deformations using quaternions. This approach is explored for its potential to smoothly represent complex dynamics, overcoming the unnatural distortions often produced by conventional methods like Linear Blend Skinning (LBS). To validate this approach, we conducted experiments synthesizing multiple distinct transformations. Specifically, a transformation involving a 150-degree rotation around the x-axis and another involving a 30-degree rotation were blended based on spatial coordinates. The results demonstrate that DQB successfully generates a smooth and continuous deformation, effectively connecting the multiple transformations without the artifacts associated with LBS. This experiment confirms that dual quaternion synthesis is a viable and effective method for handling complex, sequential deformations. This foundational study establishes the viability of our approach, paving the way for future work in generating complete, high-resolution cloth point clouds by integrating deformation and point cloud data across consecutive frames.