基于动态特征融合的监控视频图像小目标检测研究

Research on small object detection of surveillance video images based on dynamic feature fusion

在实际监控场景中,监控视频图像往往会受到复杂的背景干扰和光照变化的影响,使得小目标的检测变得非常困难且准确性较差。因此,文章提出基于动态特征融合的监控视频图像小目标检测方法。首先对监控视频图像进行预处理,主要是对不同类型的噪声进行噪声滤除操作,以提高图像质量并突出小目标;其次融合数学形态学处理技术,提取预处理后图像的小目标相关特征;最后采用动态特征融合方式进行小目标检测,将多个特征融合在一起,以更全面地描述图像中的小目标。通过这种动态特征融合方法完成了监控视频图像小目标检测方法设计。实验结果表明:新方法在检测不同类型的小目标中具有明显优势,能够实现对小目标的准确监测,说明其在监控视频图像小目标检测中具有应用价值。

In actual monitoring scenes,monitoring video images are often affected by complex background interference and light changes,which makes the detection of small targets very difficult and accurate.Therefore,small objcct detection methods for surveillance video images based on dynamic feature fusion are proposed.Firstly,the monitoring video images are preprocessed,mainly for noise filtering to improve the image quality and highlight the small targets;secondly,the mathematical morphological processing techniques to cxtract the small target related features of the prcproccssed image;finally,the dynamic feature fusion mcthod dctcct the multiple features together to describe the small targets in the image more comprehensively.The small object detection method of surveillance video image is designed through this dynamic feature fusion method.The experimental results show that the new mcthod has obvious advantages in dectecting different types of small targcts,and can realize the accurate monitoring of small targcts,indicating that it has application value in the detection of small objects in surveillance video images.