Yasunori Nagasaka

Fast 3D position measurement with two unsynchronized cameras

Stereo vision systems have been proposed and used in various fields. They require cameras to synchronize with each other for tracking objects. In this paper, we present a vision system, which uses two normal, unsynchronized cameras for calculating the 3D position of objects. The unsynchronization in shutter timing of cameras is used to process more images per second than a synchronized camera system. 3D positions of an object are measured as a crossing point of lines in 3D through the detected position on the latest frame and estimated 3D position calculated by fast frames. This makes it possible for our system to consist of two unsynchronized cameras and allows for a PC to calculate the 3D position of the moving object at 60 fps. This can be useful for controlling a robot as a fast visual feedback system.

Discrimination among individuals using orthogonal transformed face images

We investigated the discrimination among individuals by comparing the individual face image which applied an orthogonal transform with the reference images. Image recognition is affected by the conditions of the image, like the distance from the camera to a face or the brightness of the illumination. Therefore a normalizing method which adjusts to the changes of the distance and brightness is needed for correct recognition. Adjustment of the distance was carried out by enlarging/reducing the image based on the positions of the eyes, nose and mouth. The orthogonal transform is suitable for extracting the characteristic parts of an image. Moreover, it is not affected by the changes of the brightness of the illumination. We applied five kinds of transforms (discrete Fourier, Walsh-Hadamard, Haar, discrete cosine and wavelet) to the human face image. For each transform, two types of pattern matching methods were examined. One used the whole transformed image with no change, and the other used the weighted images. Weighting is performed to extract only a small part which includes individual characteristics from the image. We compared the discrimination rates for the five transforms. As a result, it has been shown that the weightings was effective to improve the discrimination rate. For all five transforms, the discrimination rate was improved by 5%-40% using weighting. In addition only small parts of the transformed image were useful for discrimination. When the discrete Fourier transform was used, the highest discrimination rate of 96.8% was obtained.

Fast Image Processing and Flexible Path Generation System for RoboCup Small Size League

Two key features of successful multi-robot systems in RoboCup are robustness of a vision system and optimization of feedback control in order to reach the goal point and generate the action under the team strategy. This paper proposes two new methods. One is a fast image processing method, which is coped with the spatial variance of color parameters in the field, to extract the positions of robots and ball in 1/30 sec. The separation problem in the interlaced format image is solved. Another one is a path generation method in which the robot approaches the goal by changing its direction convergently. With these two algorithms, the real time processing system is realized by generating a stable path under a low quality input image.

A method of pseudo stereo vision from images of cameras shutter timing adjusted

Multiple cameras have been used to get a view of a large area. In some cases, the cameras are placed so that their views are overlapped to get a more complete view. 3D information of the overlapping areas that are covered with two or three cameras can be obtained by stereo vision methods. By shifting the shutter timings of cameras and using our pseudo stereo vision method, we can output 3D information faster than 30 fps. In this paper, we propose a pseudo stereo vision method using three cameras with different shutter timings. Using three cameras, two types of shutter timings are discussed. In three different shutter timings, 90 points of 3D position for a sec are obtained because the proposed method can output 3D positions at every shutter timing of three cameras. In two different shutter timings, it is possible to calculate the 3D position at 60 fps with better accuracy.

Watching elderly and disabled person's physical condition by remotely controlled monorail robot

We are developing a nursing system using robots and cameras. The cameras are mounted on a remote controlled monorail robot which moves inside a room and watches the elderly. It is necessary to pay attention to the elderly at home or nursing homes all time. This requires staffs to pay attention to them at every time. The purpose of our system is to help those staffs. This study intends to improve such situation. A host computer controls a monorail robot to go in front of the elderly using the images taken by cameras on the ceiling. A CCD camera is mounted on the monorail robot to take pictures of their facial expression or movements. The robot sends the images to a host computer that checks them whether something unusual happens or not. We propose a simple calibration method for positioning the monorail robots to track the moves of the elderly for keeping their faces at center of camera view. We built a small experiment system, and evaluated our camera calibration method and image processing algorithm.

A Proposal of Bridging Activities between RoboCupJunior and Senior Leagues

We propose new bridging activities between RoboCupJunior (RCJ) and RoboCup senior leagues. Many of RCJ graduates cannot find a suitable way to progress to the senior leagues. Therefore, they often retire from the RoboCup at this stage. A major problem is the technical gap between RCJ and senior leagues. However, if some top-level RCJ teams are provided with sufficient technical advice, they can join senior leagues in a short time. Our proposal aims to achieve this. Since authers have managed RCJ Soccer Secondary class (RCJSS) and RoboCup Soccer Small-Size League (SSL) in RoboCup national competition and know them well, the bridging activities for them are proposed. The first-level activities include the following. (1) RCJSS members visit an SSL competition, watch the game, and understand its rules, and vice versa. (2) We have realized an inter-league game using only the current field and robots in RoboCup National Open 2007. (3) A greater level of interaction is required. When the RCJSS teams require technical assistance or advice, if they are friendly with the members of any SSL team, they can freely seek assistance. The second-level is to form a bridge league between RCJSS and SSL. (1) Rules are introduced from SSL since this league aims to be a stepping-stone to full SSL. The field size is reduced to half the current size of an SSL field. The number of robots is reduced from five to three in order to reduce the parts cost. (2) We prepare development kits for transition. RCJSS teams can develop their robot systems rapidly by utilizing these kits for transition.

A Pseudo Stereo Vision Method using Asynchronous Multiple Cameras

Systems of multiple baseline stereo vision (Okutomi & Kanade, 1993 and Kanade et al., 1997) have been proposed and used in various fields. They require cameras to synchronize with one another to track objects accurately to measure depth. However, inexpensive cameras such as Web-cams do not have synchronous systems. A system of tracking human motion (Mori et al., 2001) and a method of recovering depth dynamically (Zhou & Tao, 2003) from unsynchronized video streams have been reported as approaches to measuring depth using asynchronous cameras. In the former, the system can obtain the 2D position of a contact point between a human and the floor, and the cycle of visual feedback is 5 fps on average. In the latter, the method creates a non-existing image, which is used for stereo triangulation. The non-existing image is created from the estimated time delay between unsynchronized cameras and optical flow fields computed in each view. This method can output a depth map at the moment of frame t-1 (one frame before the current one), not the current frame. We propose a method of pseudo-stereo vision using asynchronous multiple cameras. Timing the shutters of cameras asynchronously has an advantage in that it can output more 3D positions than a synchronous camera system. The 3D position of an object is measured as a crossing point of lines in 3D space through the observation position on the last frame and the estimated 3D position using the previous two frames. This makes it possible for the vision system to consist of asynchronous multiple cameras. Since the 3D position is calculated at the shutter timing of each camera, the 3D position can be obtained from the number of cameras x 30 points. This chapter is organized as follows. Section 2 describes a method of measuring the 3D position using asynchronous multiple cameras. Section 3 reports experimental result on the recovery of motion when an object is moved in virtual 3D space, and discusses the effectiveness of the proposed method. Section 4 describes some experimental setups and reports experimental results using real images of an object moving at high speed. Section 5 discusses the processing time and enhancement of n cameras. Finally, Section 6 summarizes the method of pseudo-stereo vision.

A Design Method of Various Filters Utilizing an Adaptive Filter with Restricted Algebraic Moment of Weight

In this paper we propose a design method of the various filters utilizing an adaptive filter that has the restricted algebraic moment of the weight. When an amplitude sampling filter such as the surface acoustic wave filter or a digital filter are realized, the dispersion of the weight of these filters becomes a problem. This dispersion is given by the algebraic moment of degree N of the weight in this paper. The adaptive system consists of the ideal filter and an adaptive filter that estimates the impulse response of the ideal filter. The adaptive algorithm is a gradient algorithm by which the algebraic moment of the weight was restricted. We prove the existence of the unique convergence value of the weight and formulate the convergence condition of the adaptive algorithm. As a result, the design of the filter that has the restricted algebraic moment of the weight becomes easy, and the application of this design technology extends widely. Finally, the convergence performances of this adaptive filter are verified by the computer simulation.

Owaribito - A Team Description

In this paper, we describe image processing method for high-speed robot ID recognition with black and white markers, path generation algorithm which controls left and right wheel velocity independently, strategy and communication software with error correcting code (ECC).