Hiroki Imamura

Monocular 3D palm posture estimation based on feature-points robust against finger motion

The usability of wearable augmented reality (AR) systems would improve by letting users arbitrarily display virtual information on their palm and simultaneously manipulate it as tablet computers or smartphones. To realize such interaction between users and virtual information, we aim to robustly estimate 3-D palm posture against finger motion. This is based on the assumption that finger motion is separately estimated from palm posture and applied to manipulation of displayed information. In addition, the capability of electric sources, sensors, and processors are very limited in wearable computers. For this reason, by using a monocular camera and estimating palm posture from only a few image feature-points, we achieve an efficient estimation that satisfies real-time constraint on wearable computers. The accuracy and the robustness of our method are demonstrated by qualitative and quantitative comparison with a widely-used cardboard maker. Additionally, we confirmed that our method is run on a mobile computer at the average of 12.44 msec per frame.

Estimation for a hand position and orientation robust to motion of fingers using a web camera

We present a method to estimate the position and the orientation of a hand from a captured image for applications which need to manipulate instinctively and freely using a hand. We use a hand as a tool instead of an input device such as a mouse or a marker, and we use only a web camera instead of multiple-camera or an infrared camera, in order to improve mobility and usability. We focused on four valley-points between neighboring fingers as invariant feature points. By detecting four valley-points, the method makes possible the robust estimation for the position and the orientation of a hand regardless of the bending and the closing of fingers. In the method, we are able to move fingers freely when all valley-points are viewable. In an experiment, we qualitatively evaluated the robustness and usefulness of the proposed method by drawing 3DCG objects.

Development of A Finger Mounted Type Haptic Device Using A Plane Approximated to Tangent Plane

In recent years, several researches of haptic devices have been conducted. By using conventional haptic devices, users can perceive touching an object, such as CG (computer graphics) by a force feedback. Since conventional haptic devices provide a force feedback from a single point on an object surface where users touched, users touch an object by point contact. However, conventional haptic devices cannot provide users with a sense such as humans touching an object with a finger pad because a finger pad does not touch an object by point contact but surface contact in reality. In this paper, we propose a novel haptic device. By using this haptic device, users can perceive the sense of the slope on a CG object surface when they put fingers on it without tracing. Moreover, users can perceive the sense of grabbing a CG object with finger pads. To perceive the sense of the slope on a CG object and to grab it, we mount the plane interface of the haptic device on each two finger pads. Each plane interface provides a finger pad with the sense of the slope approximated to a tangent plane of area where they are touching. In the evaluation experiments, the subjects in this experiment evaluated this haptic device. From the results, the subjects could perceive the sense of the slope on a CG object surface. In addition, they could perceive the sense of grabbing a CG object.

Improvement of a finger-mounted haptic device using surface contact

Recently, several researches of haptic device have been conducted. Haptic devices provide users with perception of touching an object, such as Computer Graphics (CG) by force feedback. Since they provide force feedback from a single point on an object surface where users touched, users touch an object by point contact. However, they cannot provide sense such as humans touching an object with a finger pad because humans do not touch an object by point contact but surface contact. We focused on this characteristic and developed a surface contact haptic device. To touch a CG object in surface contact, we use a plane interface. It provides force feedback by being approximated to tangent plane on a CG object surface where users touched and the sense of grabbing a CG object with finger pads.

A 3-Dimensional Object Recognition Method Using Relationship of Distances and Angles in Feature Points

In recent years, human support robots have been receiving attention. Especially, object recognition task is important in case that people request the robots to transport and rearrange an object. We consider that there are four necessary properties to recognize in domestic environment as follows. (1) Robustness against occlusion. (2) Fast recognition. (3) Pose estimation with high accuracy. (4) Coping with erroneous correspondences. As conventional object recognition methods using 3-dimensional information, there are model-based recognition methods such as the SHOT and the Spin Image. The SHOT and the Spin Image do not satisfy all four properties for the robots. Therefore, to satisfy the four properties of recognition, we propose a 3-dimensional object recognition method by using relationship of distances and angles in feature points. As per our approach, the proposed method achieves to solve problems of conventional methods by using not only the feature points but also relationship between feature points. To achieve this purpose, firstly, the proposed method uses a curvature as a feature in a local region. Secondly, the proposed method uses points having high curvature as feature points. Finally, the proposed method generates a list by listing relationship of distances and angles between feature points and matches lists.

DEVELOPMENT OF AN INTUITIVE APPLICATION TO SIMULATE SEAMLESSLY THE INTERIOR AND EXTERIOR IN A BUILDING BY USING AR

Recently, some simulation systems of the building using Augmented Rarity (AR) have been proposed. In these systems, these are separated by interior simulation systems and exterior simulation systems, and two systems are independent respectively. Therefore, in performing the synthetic simulation of the buildings, the unification of two systems is required. In this paper, to satisfy a demand, we propose a system which seamlessly transfer between interior and exterior by expanding and shrinking AR object of the building. Thereby, we attempt to construct the intuitive and seamless synthetic system of the interior and exterior in a building. By the proposed system, we expect that we can intuitively and seamless simulate the synthetic simulation system of the building comparing with conventional simulation systems.

Recognition for objects by relationship between attributes

The object recognition method based on attributes has been studied. The conventional method recognizes objects by the presence or absence of attributes. However, the conventional method has two problems. Firstly, the conventional method is not able to recognize a target object of which a part of attribute is occluded. Secondly, the conventional method misrecognizes a target object which has irrelevant attributes. Therefore, to solve these two problems, we propose the object recognition by relationship between attributes. In this paper, we focus on the face as recognition object. The proposed method uses relationship as constraints for object recognition using attributes. The proposed method applies two major type constraints. First constraint is a local constraint which is applied to a part of attributes. To achieve robust face recognition against occlusion scenes, the proposed method uses the local constraint. And then, Second constraint is a global constraint which is applied to all attributes. To achieve robust face recognition against irrelevant attributes, the proposed method uses the global constraint. In this paper, to evaluate effectiveness of the proposed method, we compare the proposed method with the conventional method. We experimented in normal face, occlusion and irrelevant attributes. We used 43 images of a face which are changed in scale and rotation. Experimental results showed that the recognition ratio of the proposed method is equal to or more than the conventional method in normal face, occlusion and irrelevant attributes.

A glove-typed input device using hetero-core fiber sensors for 3D-CG modeling

In this paper, we describe development of a glove-typed input device using hetero-core fiber optic sensors. A hetero-core fiber sensor is suited to the glove-typed input device because its advantage of optical fiber such as flexibility, light weight, and thin size structure. In addition, the hetero-core fiber has been found to have highly sensitive and reproducible perfomance with bending detection. The developed glove-typed input device was successfully operated for its application to the 3D-CG modeling.