FPGA based connected component labeling

Abstract

Connected component labeling is very useful for separating object and background, and counting objects. The sequential processing architecture proposed by Von Neumann has limits in real-time processing when large data is treated. In this study, a connected component labeling system using parallel hardware architecture is implemented. This system is able to calculate over 200 frames per second (fps) and is labeled a maximum of 255 components. This is a stand-alone system that can receive input image data from a camera and display the resulting image through a monitor.