摘要
传统的运动目标检测算法具有实时性不高和鲁棒性差等缺点,未解决上述问题,本文提出了一种Subsense检测算法,该算法基于Color-LBSP算子原理,可以根据检测目标自适应调整算法。为了对非参数模型中的像素特征进行描述,本文对局部二值相似模型描述符(LBSP)进行了改进,并利用改进后的LBSP协同颜色强度(RGB)进行像素特征表述。背景建模的搭建使用了获取的像素级模型样本,使用保守的更新策略和区域扩张规则保证样本的动态化更新。通过实验结果可以证明,在不影响分割性能的情况下,本文所提出的基于Subsense的运动目标检测算法能够根据局部环境变化自动化调整模型参数,对复杂的运动目标有良好的识别特性。
In this paper,in order to overcome the shortcomings of traditional moving object detection algorithms,such as poor robustness and low real-time performance,an adaptive adaptive target detection algorithm based on color-lbsp operator is proposed,namely the sense detection algorithm.The improved local binary similarity model descriptor(lbsp)and color intensity(RGB)are used to describe the pixel features in the nonparametric model.According to the obtained pixel level model samples,the background modeling is carried out,update samples using conservative update policies and area expansion rules.Finally,the experiments show that the algorithm based on subsense can recognize complex moving objects and adjust the model parameters locally without affecting the segmentation performance.
作者
徐晓明
王雪梅
Xu Xiaoming;Wang Xuemei(Institute of physical and chemical engineering of nuclear industry,Tianjin,300180)
出处
《电子测试》
2021年第19期68-70,共3页
Electronic Test
