Hardware-driven adaptive k-means clustering for real-time video imaging

Abstract

A new adaptive k-means clustering algorithm for real-time video imaging is presented. In the proposed solution, a weighted contribution of both pixel intensity and distance between the pixels is used for cluster identification. The weight adaptation of each parameter reduces the computation complexity and makes it possible to implement the algorithm in hardware. The algorithm is designed for real-time video imaging in a VLSI implementation. It was implemented with 15 kgates and maximum clock rate of 80 MHz. Simulation results prove that a QCIF image could be handled in 15 f/s.