Parallel Extraction Architecture for Numerous Particles


We propose new processor architecture for real-time particle measurement. This architecture is able to extract particle information at the microsecond level, even when an image has several thousands of particles. The particle information includes various features such as the size, centroid, orientation, rough shape information, and so on. This kind of processing enables to construct a system which could realize high performance on fluid analysis as well as product inspection. The covering fields are medical science, chemistry and industry application.

Fig.1 Parallel Extraction

The proposed architecture simultaneously extracts the moments of multiple particles in parallel by utilizing the massively parallel architecture. The basic concept of this parallel extraction is shown above. This enables a significant reduction in the amount of calculation required. In addition, asynchronous operation between processing elements allows fast processing.

Fig.2 Example of Parallel Extraction


  1. Yoshihiro Watanabe, Takashi Komuro, Shingo Kagami and Masatoshi Ishikawa: Parallel Extraction Architecture for Information of Numerous Particles in Real-time Image Measurement, Journal of Robotics and Mechatronics, Vol. 17, No. 4, pp. 420-427, 2005. [PDF]
  2. Yoshihiro Watanabe, Takashi Komuro, Shingo Kagami, Masatoshi Ishikawa: Parallel Extraction Architecture for Image Moments of Numerous Objects, IEEE 7th International Workshop on Computer Architecture for Machine Perception (CAMP'05) (Palermo, 2005.7.4) / Proceedings pp. 105-110. [PDF]