Title

基于自适应高斯混合模型的图像稳定方法

Translated Titles

Image Stabilization Based on Adaptive Gaussian Mixture Model

Authors

张敏(Min Zhang);赵猛(Meng Zhao);贾云得(Yun-De Jia);王俊(Jun Wang)

Key Words

图像稳定 ; 运动估计 ; 运动补偿 ; 高斯混合模型 ; image stabilization ; motion estimation ; motion compensation ; Gaussian mixture model

PublicationName

北京理工大學學報

Volume or Term/Year and Month of Publication

24卷10期(2004 / 10 / 01)

Page #

897 - 900

Content Language

簡體中文

Chinese Abstract

提出了一种基于自适应高斯混合模型(Gaussian mixture model)的运动滤波器的图像稳定方法。首先采用多分辨力金字塔技术,通过特征跟踪进行运动估计;然后分析前n帧图像的统计特征,求取GMM参数;最后利用GMM运动滤波器对运动参数进行滤波,输出稳定的图像序列。实验结果表明,该方法对于图像的不规则运动的参数补偿是有效的,并且可以实现实时处理。

English Abstract

A novel approach is presented to stabilize image sequences. A Gaussian mixture model for motion compensation is proposed in order to smooth down the unwanted vibrations of image sequences. Digital image stabilization system integrates three modules: pyramid-based motion estimation, adaptive Gaussian mixture model and motion compensation. Experimental results show that, using the method discussed for motion parameter compensation, the system can work in real time.

Topic Category 基礎與應用科學 > 基礎與應用科學綜合
工程學 > 工程學綜合