In this paper we discuss the use of elastic body splines to improve the performance of the fluid deformation algorithm for non-linear registration. It is found that the fluid deformation algorithm gives a good registration at the expense of spatial blurring. By applying elastic body splines to landmarks created automatically within the fluid algorithm, an improved performance is achieved.
The generation of virtual environments has recently attracted much attention. In this paper we address our recent work in this area. Unlike most of methods of 3D scene construction, instead of pursuing the precise 3D structure, we consider the 3D pseudo-structure of 3D scenes to construct the virtual environments from multi-images with forward motion. For each frame, visual 3D geometry of the scene's background is defined as a simple polygonal model, by roughly specifying the vanishing point. Modeled 3D scene is decomposed into sub-images, which are corresponding to the polygons in the modeled 3D scene. Sub-images whose polygons have the same normal are registered using the Levenberg-Marquardt iterative algorithm to estimate motion parameters. These registered sub-images are correspondingly pasted together to form a pseudo-3D space. By controlling the position and direction, the virtual camera can walk through in this virtual space to create novel 2D views and to acquire an immersive impression. Even if the virtual camera goes deeper into this virtual environment, it also can obtain a novel view while maintaining high resolution.
This paper presents a new formalism for the concept of fractal iteration applied to the dynamics of petri nets by translating the evolution of it into graphics output. The idea is derived from the comcept of fractal iteration principles and the approach use a petri nets as an abstract powerful modeling tool for the generation of self-similar fractal images via its duality, deadlock, inhibitor arcs, firing sequence and marking reachability. Generating fractal images via the dynamics of a petri nets allows an easy and direct proof for the similarity and correspondence between the dynamics of complex quadratic fractals via a recursive procedure and the state of a petri nets via a reachability problem. The reachability problem will be solved in terms of dynamics of the fractal in order to generate images via two proposed methods. Validation of our approach is given in terms of experimental results. An investigation of the relationships between the generated images and their corresponding petri nets is also discussed.
This paper addresses the problem of articulated figure positioning using dynamic and kinematic constraints, concentrating on balancing when there is a significant force interaction with the environment. The dynamic and kinematic constraints are satisfied simultaneously allowing an end-effector to move to a desired position and apply a certain amount of force and torque within the physical limits of the figure. The balance of the figure is controlled using a new technique based on the precise manipulation of joint torques, which reduces the torque of the system considering all the joints in proportion to their influence on the balance. The proposed algorithm can be easily embedded in inverse kinematics based systems since it has the same order of time complexity and a similar structure. The algorithm can also be used to correct a kinematic based posture to make it dynamically sound and robust. If there are multiple supports, the supporting ratio can be shifted smoothly and continuously between supporting sites even if a considerable force interaction exists with the environment.
In this report, one of watermarking methods by using spread spectrum and wavelet is proposed. The original image is spread its frequency domain at first. The energy of high friqency domain is calculated by decomposing the spread image with wavelet. Watermark is embedded into the subband region of multi-resolution analysis with using the energy of multi-resolution representation. The watermark is spread and become low power and wide band signal. In result, invisible watermark can be made. Our proposal scheme provides a good tool as one of copyright protection techniques for multimedia network.
This paper presents a lock-control system using face identification. There are three main processes in the face identification : 1) extraction of a face from an input image, 2) normalization of the size and inclination of the face image, 3) matching of its facial features (eye, nose and mouth). Thresholds of matching are determined by statistical estimations. As results of experiments, the correct acceptance rate was 92.2% and the correct rejection rate was 99.6%. Additionally, sectionally adaptive correlation method for higher robustness against fluctuation of lighting condition is proposed.
An extraction method of personal features based on on-line handwritten characters including pen-pressure and pen-inclination information is proposed. In the proposed method, feature parameters extracted from handwriting information are classified into several categories and each category is weighted suitably according to the uniqueness of personal features. The proposed feature extraction method can be applied to text-independent writer verification. In this paper, the reliability of the proposed method with some simulation results using Japanese handwriting data is also discussed.
Recently, network services such as banking, electronic commerce, database access services, information services over the Internet and telephone networks have become popular, and user verification technology that is essential to those services has become important. Voice verification technology is one type of user verification technology, and demand for this should be anticipated in an environment where services such as those over a telephone network can use only voice as the means of communication. This paper describes practical speaker recognition technology for constructing voice verification systems. Moreover, this paper introduces new software developer kits for speaker recognition, reports on speaker recognition experiments using telephone speech that we recently conducted, and shows that the recognition performance is seriously affected by text and handset conditions that are the same/different for training and testing.
The method of the extraction and the evaluation of the individual with the pinna is important to raise the identification accuracy in individual identification which uses the pinna. Moreover, the extraction of the pinna area is important to make the identification method easily. In this thesis, the individual with the pinna is investigated, and the result of examining the evaluation method and the extraction method of the pinna area is described.
Rotation Invariant Phase Only Correlation (RIPOC) is one of the applications of Phase Only Correlation (POC). RIPOC can correlate two images even though there is rotation between two images. It is also able to measure the rotation angle between two images. Like POC, RIPOC processes images macroscopically so it can correlate the images of which characteristic points are difficult to be extracted. We present the experimental results of the ability to detect the rotation angle by RIPOC.