改进Vibe前景提取的YOLOv3烟雾目标检测方法

Smoke object detection method based on YOLOv3 and improved Vibe foreground extraction

在使用计算机视觉方法对烟雾目标进行检测时需要考虑环境因素的影响,直接进行检测易出现漏检、误检问题。因此提出一种基于改进Vibe前景提取的YOLOv3烟雾目标检测方法。首先使用CLAHE增强图像对比度,通过改进的Vibe前景提取方法检测并提取疑似的烟雾目标区域,最后整合样本并使用目标检测算法YOLOv3对烟雾目标进行识别与定位。实验结果表明所提出的方法对复杂环境的烟雾目标检测具有更高的性能,更适用于烟雾目标检测任务。

When using computer vision methods to detect dynamic smoke object,the influence of environmental factors needs to be considered,and direct detection is prone to problems of missed detection and false detection.Therefore,a YOLOv3 smoke object detection method based on improved Vibe foreground extraction is proposed.Firstly,CLAHE is used to enhance the contrast of the image,and the suspected smoke object area is detected and extracted by the improved Vibe foreground extraction method.Finally,all the sample data are integrated and the object detection algorithm YOLOv3 is used to identify and locate the smoke object.The experiment results show that the proposed method has higher performance for smoke object detection in complex environments,and is more suitable for dynamic smoke object detection tasks.