Smart Vision & Robotic Sensing

Professor, Robotics Laboratory
Smart Innovation Program, Graduate School of Advanced Science and Engineering
Hiroshima University
>> Research Contents
In order to establish high-speed robot senses that are much faster than human senses, we are conducting research and development of information systems and devices that can achieve real-time image processing at 1000 frames/s or greater. As well as integrated algorithms to accelerate sensor information processing, we are also studying new sensing methodologies based on vibration and flow dynamics; they are too fast for humans to sense.

Real-Time Feature-Point-Tracking Algorithm for High-Speed Vision

In this study, real-time feature point tracking at 1000 fps was performed by implementing a feature point tracking algorithm on a high-speed vision platform, which is improved for hardware integration and high-speed processing at real time. To solve the problems of high-speed hardware processing with less memory consumption and low computational load, we improved the feature point tracking algorithm by considering (1) maximum value search in divided block regions, and (2) correspondence in limited block regions by assuming high frame rates. The processing flow of the improved algorithm in the below figure.

The high-speed vision platform on which the improved algorithm is hardware-implemented can be used to track feature points of 1024×1024 pixel images at 1000 fps. By considering fast-moving objects in the real world, we verified the performance of our developed real-time feature point tracking system.

rotating star-shaped objec MPEG movie(1.0M)
rotating star-shaped object
waving checkered cloth MPEG movie(2.1M)
waving checkered cloth
moving printed characters MPEG movie(1.6M)
moving printed characters