摘要
为解决传统背景减法在动态背景下受噪声干扰和运动目标检测准确性不高的问题,提出一种基于改进背景减法的视频图像运动目标检测方法。在背景建模阶段,为易于计算和提高检测精度,采用基于GMM的图像块均值方法重构背景模型;在目标检测阶段,采用数学形态学和小波半软阈值函数相结合的方法对检测到的运动目标进行去噪处理;在背景更新阶段,采用自适应背景更新方法进行背景更新。实验结果表明,所提方法提高了运动目标检测的准确性,验证了其有效性。
To solve the problems of noise interference and inaccuracy of moving object detection in dynamic background in traditional background subtraction,a method of moving object detection in video image based on improved background subtraction was proposed.In the background modeling stage,GMM-based image block mean method was used to reconstruct the background model to facilitate calculation and improve detection accuracy.In the phase of target detection,the method of combining mathematical morphology with wavelet semi-soft threshold function was used to denoise the detected moving target.In the background updating stage,the adaptive background updating method was used to update the background.Experimental results show that the improved background subtraction method improves the accuracy of moving object detection,thus demonstrating the effectiveness of the improved algorithm.
作者
左军辉
贾振红
杨杰
Nikola KASABOV
ZUO Jun-hui;JIA Zhen-hong;YANG Jie;Nikola KASABOV(College of Information Science and Engineering,Xinjiang University,Urumqi 830046,China;Institute of Image Processing and Pattern Recognition,Shanghai Jiao Tong University,Shanghai 200240,China;Knowledge Engineering and Discovery Research Institute,Auckland University of Technology,Auckland 1020,New Zealand)
出处
《计算机工程与设计》
北大核心
2020年第5期1367-1372,共6页
Computer Engineering and Design
基金
国家自然科学基金项目(U1803261)。
关键词
背景减法
背景模型
数学形态学
自适应背景更新
运动目标检测
background subtraction
background model
mathematical morphology
adaptive background updating
moving object detection