基于视频流的林火烟雾运动目标检测方法的研究

运动目标检测是实现视频图像分类与识别的前提.烟雾是森林火灾发生早期的显著特点和视觉现象,通过对林火烟雾图像的特征分析,研究了几种常用的运动目标检测方法,即帧间差分法、背景估计法等,分析了其实现过程,对比了它们的优缺点,并寻求最佳的视频林火烟雾运动目标检测方法.实验结果表明:改进的背景估计法结合色彩判断准则的方法不仅具有更好的烟雾捕捉能力,而且抗干扰能力强,将大大减轻后续图像识别的压力.

Based on Forest Fire Smoke Moving Object Detection Video Stream

Moving object detection is the precondition for video image classification and recognition.Smoke is a distinctive feature of the early forest fire,forest fire smoke through the characteristics of the image analysis,several common moving target detection methods,analyzes the implementation process,comparing their advantages and disadvantages,and to seek the best forest fire smoke video moving target detection methods.The experiments also showed that the improved method with color background estimation criterion method not only has a better ability to capture smoke and anti-jamming capability,which greatly reduces the subsequent image recognition pressure.