Abstract:
Traditional Gaussian Mixture Model(GMM) is very sensitive to light mutations and is slow for convergence speed. This paper presents a detection method for moving object based on improved GMM. The method can eliminate the effect of illumination by mismatching pixel. Background image is exacted by improved GMM. Binary difference image is got by background image difference, then moving object is got from difference image. Experimental results show that the detection can adapt illumination changes well and improve the accuracy and robustness of moving object diction.
Key words:
Gaussian Mixture Model(GMM),
background image difference,
background updating,
illumination change
摘要： 传统高斯混合模型(GMM)对于光照突变十分敏感，且收敛速度较慢。为此，提出一种基于改进GMM的运动目标检测方法。利用不匹配像素消除光照影响，使用改进的GMM提取背景图像。通过差分当前帧与背景图像获得二值差分图像，从该差分图像中获取运动目标。实验结果表明，该方法能适应光照变化，提高检测的准确性和鲁棒性。
关键词:
高斯混合模型,
背景图像差分,
背景更新,
光照变化
CLC Number:
SU Bing, LI Gang, WANG Hong-Yuan. Detection Method for Moving Object Based on Improved Gaussian Mixture Model[J]. Computer Engineering, 2012, 38(2): 210-212.
苏兵, 李刚, 王洪元. 基于改进高斯混合模型的运动目标检测方法[J]. 计算机工程, 2012, 38(2): 210-212.