Yusuke Tokunaga, Toshihide Hakukawa, and Takahiro Inoue

Abstract:
We propose an algorithm and design of an intelligent digital integrated circuit for recognition of circular patterns in a binary image based on template matching using a modified matching degree. The proposed system consists of a preprocessor #1 (noise reduction, coarsening, and edge detection), a preprocessor #2 (noise filtering and location parameter detection), and a circular pattern recognition block. The proposed system is implementable onto field programmable gate arrays (FPGAs) and forming part of the vision system for a watermelon harvesting robot. Functional verification, logic synthesis, and implementation are detailed for the FPGA circular pattern recognition block.

Keywords: noise reduction, coarsening, edge detection, noise filtering, pattern recognition, FPGA