煤矿监控中火灾图像分割方法研究

Research on Fires Image Segmentation Technology in Coal Mine Surveillance

在分析现有火灾图像分割方法特点的基础上,针对最大类间方差法(Otsu法)对双峰陡峭且双峰相距过远的图像不能精确提取目标的问题,提出一种利用直方图和遗传算法的最大类间方差法的改进方法。研究结果表明,该方法提取目标更精确,且对噪声的抑制能力更强。

Based on analysing the characteristics of fire image segmentation,aimed at the problem that Otsu method can not accurately extract the target from the steep peaks image and the image that the peaks are too far away, the paper presents an improved Otsu method by means of histogram and Genetic Algorithm. The results indicate that the method extracts more precise targets and it has more capable of noise suppression.