摘要
提出一种用于彩色序列图像复原的模型更新算法,计算退化图像序列各帧的图像质量,统计序列图像质量的方差,以方差差异作为判断准则,选择适当的模型进行复原。该算法扩展了基于支持向量机的彩色图像复原算法。仿真实验中,测试图像采用视频监控和智能交通领域常见的运动模糊进行退化处理。实验结果表明,该算法能有效标记出图像序列中质量发生显著变化的关键帧,复原效率得以提高,同时复原也更有针对性。
A model updating algorithm is proposed for support vector machine（SVM） based color image sequence restoration. The image quality is calculated for each frame of a degraded image sequence. The quality variances of the image sequence are counted. The variance difference is applied as the judgment criterion of the model selection in the restoration. The SVM based color image restoration algorithm is extended and can be applied to restore sequential images. In the simulation experiments, the testing images are degraded by motion blur, which is a common degradation type in video surveillance and intelligent transportation related fields. Experimental results show that the proposed algorithm can effectively mark out the key frames of an image sequence which are degraded most significantly. The restoration efficiency is improved and the restoration algorithm becomes more pertinent.
出处
《上海电机学院学报》
2009年第4期310-313,共4页
Journal of Shanghai Dianji University
基金
国家自然科学基金资助项目(60675023)
关键词
图像复原
序列图像
支持向量机
image restoration
image sequence
support vector machine（SVM）
