Summary
International Symposium on Nonlinear Theory and Its Applications
2016
Session Number:C1L-F
Session:
Number:C1L-F-1
A Hardware Cellular-Automaton Architecture for Spatial Pattern Generation Towards Motion-Vector Estimation of Textureless Objects
Aoi Tanibata, Miho Ushida, Alexandre Schmid, Masayuki Ikebe, Tetsuya Asai, Masato Motomura,
pp.-
Publication Date:2016/11/27
Online ISSN:2188-5079
Summary:
We have already proposed a hardware-oriented cellular-automaton algorithm that generates spatial patterns on textureless objects and backgrounds, aiming at motion-vector estimation of textureless moving objects (IJUC 2016, NOLTA 2015). In this report, we propose a fundamental architecture for the algorithm. The system consists of 1-D shift-register arrays (FIFOs) and arithmetic operators for diffusion, a lookup table acting as a nonlinear reaction function, and several state controllers (counters, multiplexers, etc.), and they act as 1-D reaction-diffusion streaming processor (RDSP). By arranging 1-D RDSP on a 2-D mesh in a time-division manner, 2-D image processing for texture generation will be performed, which results in reduced complexity of the circuit system as a whole. We will demonstrate the operation of the proposed hardware system on our FPGA board.