论文部分内容阅读
烟雾探测是预见火灾隐患的重要手段,具有快速响应和非接触等优点,但现有方法在减少计算量和减少误报上尚需改进。提出了一种基于运动块追踪的视频烟雾探测方法。采用混合高斯模型对背景进行建模,提取可疑运动区域并划分运动块,根据运动块灰度变化及与周围8邻域运动块的灰度梯度,筛选需要全程跟踪的目标运动块。在随后追踪过程中,分析各个运动块的运动方向、灰度衰减变化及半透明性特征,综合各特征值对烟雾进行判断。在测试烟雾视频中,60帧内便可以探测到烟雾块,而在干扰视频中只出现了1次误报。试验结果表明该方法减少了计算量,提高了烟雾检测的准确性,降低了误报率。
Smoke detection is an important means of predicting fire hazards, and has the advantages of fast response and non-contact. However, the existing methods need to be improved in terms of reducing the calculation and reducing false positives. A video smoke detection method based on motion block tracking is proposed. The mixed Gaussian model was used to model the background, the suspicious motion regions were extracted and the motion blocks were divided. According to the gray level of the motion blocks and the gray gradient of the neighboring motion blocks with the surrounding 8 neighborhoods, the target motion blocks that needed tracking were selected. In the subsequent tracking process, the movement direction of each moving block, the change of gray attenuation and the translucent characteristics were analyzed, and the smoke was judged according to each feature value. In the test smoke video, smoke blocks can be detected within 60 frames and only 1 false alarm appears in the interference video. The experimental results show that this method reduces the computational load, improves the accuracy of smoke detection and reduces the false alarm rate.