摘要
采用传统混合高斯模型(GMM)进行监控视频内运动目标检测时,检测结果受光照、阴影影响较大,且运算速度慢,针对这一缺陷,提出了一种改进的GMM和阴影抑制相结合的运动目标检测算法。该算法通过对GMM添加平衡系数、合并冗余高斯分布,完成自适应改进,利用2级学习率αh和αl的设置提升背景建模精确率度,经霍特林变换完成阴影抑制,得到轮廓清晰的运动目标。仿真实验结果显示:改进的算法克服了传统GMM在动态背景下检测精度低的缺陷,提高了算法的实时性和准确性,解决了光照变化、环境噪声等因素引起的误检、漏检问题。
Aiming at the defects of traditional Gaussian mixture model in monitoring video moving target detection,the detection result is affected by illumination and shadow,and the operation speed is slow.An improved moving target de-tection algorithm is proposed by combining GMM and shadow suppression.The adaptive improvement is completed by adding the balance coefficient to the GMM and combining the redundant Gaussian distribution.The accuracy of the background modeling is improved by the setting of the two-level learning rates αh and αi,and the shadow suppression is achieved by the Hotlin transform to obtain the clear moving target.The simulation results show that the algorithm over-comes the defect of low accuracy of traditional GMM in dynamic background,improves the real-time performance and accuracy of the algorithm,and solves the problem of false detection and missed detection caused by illumination changes and environmental noise.
作者
范超男
李士心
张海
郭荣
FAN Chao-nan;LI Shi-xin;ZHANG Hai;GUO Rong(School of Electronic Engineering,Tianjin University of Technology and Education,Tianjin 300222,China)
出处
《天津职业技术师范大学学报》
2019年第4期25-31,共7页
Journal of Tianjin University of Technology and Education
基金
天津市高等学校科技发展基金资助项目（20140818）
天津市自然科学基金资助项目（18JCYBJC16400）
