In recent years, new active range sensors have been developed for 3D data acquirement, such as time-of-flight cameras. These sensors enable acquiring of range data at video rate and are suited for dynamic environment. Unfortunately, the resolution of the range data is quite limited and the captured data are typically contaminated by noise. In this paper, we propose a novel method for depth video enhancement. Using high resolution color video as guidance reference, we iteratively refine the input depth map based on a newly presented linear filter model, in terms of both its spatial resolution and depth precision. The linear filter has a good edge-preserving property and a runtime independent of filter size, which fulfills both accuracy and speed requirements. For temporally consistent estimate on depth video, we extend the method into temporally neighboring frames. Simple optical flow and patch-based similarity measure are used to obtain accurate depth in an efficient manner. Experimental results show that the proposed method greatly improves the quality and boosts the resolution of range data while achieving high computational efficiency. We also show that the temporally consistent constraint addresses a flickering problem and improves the accuracy of depth video.
In the ECE regulation, any surface of the interior part of automobiles must have sufficient roundness where a sphere of diameter 165mm, which is equivalent to the average head size of infants, may collide. A system for automatically detecting sphere contacting shape on the automobile part is developed. In the current practice, some Japanese automobile manufactures detect the shape by using the virtual milling method. In this method, a milling simulation with a ball end cutter of diameter 165mm is executed on the part to detect the sphere contacting shape. The method generally needs a lot of computation time for the detection. In this paper, the authors propose an improved virtual milling method for fast detecting the sphere contacting shape. Our algorithm initially generates points sufficiently covering the visible surface of the model. For each point, the contact condition of the cutter is evaluated by using the inverted offset surface of the part. An inverted offsetting method accelerated with the depth buffer mechanism of GPU is introduced. An experimental system is implemented and some computational experiments are performed. Our system can detect the sphere contacting shape on the part surface with approximately 1 million polygons in a few seconds.
Using paper models is very attractive for tangible user interaction in an augmented reality (AR) environment. In this paper, we propose an approach to tangible interaction using paper models, which is available at very low cost without hardwire connections, and accessible without restriction on location. For tangible interaction, we facilitate product- and ring-types of paper models which are all fabricated by paper crafting. The approach has been tested and applied in virtual design evaluation of digital handheld products. We found that the approach is accurate enough to be applied to the design evaluation process and tangible enough to provide a pseudo feeling of manipulating virtual products with human hands.
A novel algorithm is proposed for registration of two partially overlapping range images that are arbitrarily oriented. The algorithm is inspired by the Shape Index (SI) description of 3D geometry and the Scale Invariant Feature Transform (SIFT) algorithm for detecting and matching features between intensity images. For a range image pair to be registered, we first construct an artificial intensity image pair based on SI. Then, the SIFT algorithm is used to extract and match features between the artificial images, and accordingly the 3D feature matches between the range images can be established with ease. A RANSAC procedure, combined with a hysteresis thresholding scheme, is carefully designed to filter out the false 3D feature matches to ensure the robustness. Finally, the transformation is computed according to all correct 3D feature pairs to register the two range images. Experiments and comparisons are included to verify the superior robustness and effectiveness of the proposed approach.
The objective of this paper is to propose a method to evaluate an aesthetically designed product from the viewpoint of grasping. Ease of grasping, which is considered to affect the ergonomic aspect of aesthetic design, is estimated by using the moment of force in the hand when grasping the aesthetically designed product. Two grasping position are used to evaluate the ease of grasp. These grasping positions are achieved when the hand exerts a minimum moment of force. A design support system was also developed based on the design method from the previous work and current proposed evaluation method. The system is capable of providing an aesthetic design solution and to subsequently perform the ease of grasping evaluation. The effectiveness of the proposed method was confirmed in ergonomics experiment using a PET bottle shape.
Engineering objects often exhibit symmetries on their surfaces for functionality or aesthetics. Detecting such symmetries from their scanned meshes enables advanced applications, such as feature-based CAD model reconstruction or mesh size reduction for efficient CAE. In this paper, we propose a new method for detecting a common class of symmetries in engineering, such as translation, rotation and reflection, from scanned meshes of a variety of objects ranging from mechanical parts to aesthetic products. Major advantages of our proposed method over conventional ones are considerable robustness for scanner noise and exhaustive detection of multiple symmetries from a mesh. We demonstrate its effectiveness from various experiments on scanned meshes acquired by laser or CT scanners. We also show that surface distortions or deformations on the object surfaces can be evaluated using symmetries quantitatively.