A simple hierarchical clustering method for improving flame pixel classification Kleber J. F. Souza, Silvio J. F. Guimar˜ aes, Zenilton Patroc´ ınio Jr. Computer Science Department PUC Minas Belo Horizonte, MG, Brazil kleberjfsouza@gmail.com, {sjamil,zeniltong}@pucminas.br Arnaldo de A. Ara´ ujo Computer Science Department UFMG Belo Horizonte, MG, Brazil arnaldo@dcc.ufmg.br Jean Cousty Lab. d’Informatique Gaspard-Monge Universit´ e Paris-Est Paris j.cousty@esiee.fr Abstract—In this paper, we propose a new approach for color image simplification in order to improve flame pixel classification. The fire detection performance depends critically on the performance of the flame pixel classifier. Color image simplification is the process of simplifying an image in order to decrease the number of colors while preserving, as much as possible, shapes. In this work, a hierarchical clustering method in a given color space is used to map the original colors into a smaller set of representative ones, allowing the use of a simple heuristic rule for classifying the clusters related to candidate flame colors. Using reverse mapping, we identify possible flame colors in the image. Main contributions of our work are the application of a simple hierarchical clustering method to color simplification, that decreases the number of possible flame colors, and a filtering methodology to reduce the influence of outliers. Several color spaces and distance measures were used to evaluate the proposed method. Experimental results demonstrate that color simplification is essential to successfully employ heuristic classification of flame colors. Keywords-Flame pixel classification; Hierarchical clustering; Minimum Spanning Tree; I. I NTRODUCTION Fire detection based on image analysis has become one of the most promising techniques and it also presents a low cost. It addresses fire detection problem without using expensive infrared and ultraviolet sensors which were mainly adopted until mid-1990s [1]. These sensors present several disadvantages: high acquisition and maintenance costs, huge number of false positives, great noise sensibility, impractical use for large open spaces and lack of additional data, such as fire growth rate [1]. Visual fire detection is based on analysis of several aspects, such as color, movement, and geometry. But the most important one is color analysis, which tries to locate pixels associated with flame. Even the number of false positive is high, the presence of object colors similar to flame may produce a high false alarm rate (Fig. 1). In general computer vision-based fire detection systems employ three major stages [2], [3], [4], [5]. The first stage is the flame pixel classification; the second stage is the moving object segmentation, and the last part is the analysis of candidate regions. This analysis is usually based on two figures of merit: the shape of the region and its temporal changes. The fire detection performance depends critically on the performance of the flame pixel classifier which generates seed areas on which the rest of the system operates. The flame pixel classifier is thus required to have a very high detection rate and preferably a low false alarm rate. (a) Sun (b) Fire (c) Fire Figure 1. some images with objects colors similar to flame In this paper, we propose a new approach for color image simplification in order to improve flame pixel classification. Color image simplification is the process of simplifying an image in order to decrease the number of colors while preserving, as much as possible, shapes and object contours. In this work, a hierarchical clustering method in a given color space is used to map the original colors into a smaller set of representative ones, allowing the use of a simple heuristic rule for classifying the clusters related to candidate flame colors. Using reverse mapping, we identify possible flame colors in the image. The main contributions of our work are the application of a simple hierarchical clustering method to color simplification, that decreases the number of 2011 23rd IEEE International Conference on Tools with Artificial Intelligence 1082-3409/11 $26.00 © 2011 IEEE DOI 10.1109/ICTAI.2011.25 110