复杂背景下火灾疑似区域快速提取算法

Rapid extracting algorithm of suspected fire area under complex background

针对火灾场景图像容易受到噪声干扰、光照变化等干扰因素的影响,难以实时有效地提取出火灾运动区域的问题,设计了一种适合于全天候高效工作的火灾火焰运动区域的快速提取算法,该算法根据像素点灰度信息分布和序列帧相关性,从时间域对背景模型和阈值进行自适应更新,并利用投影原理的二维统计原理,在基于运动目标区域空间相关性的基础上,实现了对火焰疑似区域的快速提取。实验结果表明,在1 920×1 080分辨率下,该算法总共消耗时间为0.232 ms。该算法解决了外界光线对目标区域提取的影响,同时火灾疑似区域检测的时间消耗以及算法复杂度也比区域聚类算法大大降低,较大地提高了算法的执行效率。

Aiming at solving the problem of extracting real-time fire movement area caused by the sensitive of fire scene image in variable noises interference and illumination changes, a kind of rapid extraction algorithm suitable for all-weather efficient fire flame movement area is designed in this paper. According to the distribution of pixel gray information and the correlation of sequence frame, the algorithm, using the 2d-statistical principle of projection principle, adaptive updat-ing from the time domain for background model and the threshold value, can realize the extraction of the flame suspected area rapidly based on the movement of the target area on spatial correlation. The experimental results show that the pro-posed algorithm consumption time is 0.232 ms in total at a 1920×1080 resolution. The algorithm solves outside light influ-ence in the extraction in the target area. Compared with the area clustering algorithm, the algorithm proposed in this paper, reduces the detecting time consuming of suspected fire area and algorithm calculation complexity greatly.