Real-Time Segmentation of Color Images based on the K-means Clustering on FPGA

Abstract

In this paper, we describe a segmentation method of color images based on the k-means clustering. With a k-means clustering algorithm, we can reduce the number of colors in a given image to K while maintaining the quality of the image. Based on these K colors, we can segment color images by recognizing contiguous pixels of the same color as a region. However, the k-means clustering is a very time consuming task, particularly for large size images and large number of clusters. Therefore, in order to use a k-means clustering algorithm for image segmentation, we need to recognize the regions in parallel with the k-means clustering algorithm. In our implementation, the regions can be recognized in parallel with each iteration of the k-means clustering algorithm.