Abstract

The title of my doctor thesis is ``Multiple-Viewpoint Range-Finding System for Accurate Three-Dimensional Model-Based Movie Acquisition". Meeting the components of the "real-time", "high-speed", "high-accuracy", "3-D model completeness" is the key for many advanced application such as virtual reality. For the realization, we work on three topics: "[1]Multiple-viewpoint system", "[2]Hardware signal processing", and "[3]Calibration"

[1] Multiple-viewpoint system

Rangefinders, each of which consists of camera and projector for the sheet light projection based on light-section method, are placed in multiple viewpoints to configure the multiple-viewpoint system. We employ the time-division light-projection technique to prevent light interference from multiple range finders. And we use the smart image sensor with VGA(640x480) pixel resolution for high-speed high-accuracy 3-D imaging. The present system achieves 2.5mm of maximum error and 0.71 mm of average error by measuring a plane placed at 1200 mm. The measured capture speed is 6.1 and 10.6 models per second with 2 and 3 rangefinders, respectively. The maximum speeds are 32.5 and 21.7 models per second. The result achieves high-speed and high-accuracy 3-D model-based movie acquisition compared with the other systems.

(Original 3-D model-based movie) (Acquired real-time 3-D model-based movie)
Please click the photo if you want to watch the movie.

Related publications

Y.Yachide, M. Ikeda, and K. Asada, ``Time-Division-Based Multiple-Viewpoint 3-D Measurement System for Real-Time, High-Speed, and High-Accuracy Model Movie Acquisition,''
Journal of Image Information and Television Engineers, Vol. 62, No.3, pp. 392 -- 397, Mar. 2008(in Japanese).

Y. Yachide, Y. Oike, M. Ikeda, and K. Asada, ``Real-Time 3-D Measurement System Based on Light-Section Method Using Smart Image Sensor,''
in Proc. of IEEE International Conference on Image Processing(ICIP), pp. 1008 -- 1011, Sep. 2005.

And more.

[2]Hardware signal processing

We present the FPGA-based 3-D measurement system aiming for the parallel control, high-speed 3-D position computation, and display of the multiple rangefinders. The huge amounts of data processing are required for the 3-D model-based movie acquisition using multiple image sensors with high-pixel resolution. And it is necessary to control a sensor and to transfer data in multiple-viewpoint system. Therefore, it is not appropriate to process such huge data in multiple-viewpoint system with a standard PC. We develop the standalone 3-D measurement system with FPGA-Based 3-D computational engine for parallel sensor controlling and high-speed 3-D computation.

Related publications

Y. Yachide, M. Ikeda, and K. Asada, ``High-Speed 3-D Measurement System Using Smart Image Sensor and FPGA Based 3-D Engine,''
in Proc. of IEEE International Conference on Electronics, Circuits and Systems (ICECS), pp. 764 -- 767, Dec. 2006.

Y. Yachide, M. Ikeda, and K. Asada, ``FPGA-Based 3-D Engine for High-speed 3-D Measurement Based on Light-Section Method,''
Proc. of IEEE International Conference on Field-Programmable Technology (ICFPT), pp. 293 - 296, Dec., 2007.

And more.

(Acquired 3-D model-based movie)
Please click the photo if you want to watch the movie.

[3]Calibration

We propose the corresponding-points free calibration method using spheres for multiple-viewpoint system aiming large area calibration with low cost.

Related publications

Y. Yachide, M. Ikeda, and K. Asada,
``Triangulation-based calibration method based on light-section method using spheres,''
Proc. of International Conference on Sensing Technology (ICST), pp. 399-403, Nov., 2007.

And more.