Image Contour Extraction using Bacterial Quorum Sensing

In this paper we present an image contour extraction strategy based on bacterial Quorum Sensing (QS). By imitating bacterial behavior, we get the bacteria or agents to navigate the image and accumulate in the areas of attraction, which correspond to special features of color. When the bacterial population begins to cluster in one or several areas of the image, the population virulence is activated, which accelerates the process of convergence. Population boundaries identify the contour sought in the images. As a result, the method achieves accurate estimation of the area related to the flame in real-time. We present experiments where our strategy accurately identifies the flame inside the furnace, which increases the performance of the oxygen estimation algorithm in which it is used.