Real-time classification of objects by color using field programmable gate array

Abstract

This paper presents a system for real-time classification by color of objects carried out on a video sequence at 20 Frames Per Second (FPS). The Spherical Coordinate Transform (SCT) is used for the color analysis. The histogram for a partition of a bidimensional color space is computed on a concurrent digital circuit, allowing the classification of five different colors. The results have shown that green, yellow, and orange colors can be reliably classified. However, red and magenta colors are confused each other. The efficiency of the classifica-tion is 95.2%. The prototype was implemented using an Altera Cyclone® II 2C70 FPGA device on a DE2-70 educational board from TERASIC and Quartus II for the design.