摘要
为了提高图像编码预测器的预测性能,提出了一种低复杂度,高效的自适应预测方法。采用LMS(Least MeanSquare)自适应滤波技术进行预测,并对预测值进行减邻域均值的改进,有效克服了图像的非零均值和非平稳性特征,满足LMS算法的要求,使预测性能得以提高。通过对不同图像的仿真结果表明,该方法的预测差值图像的熵比GAP算法和MED算法的差值图像的熵要小0.1 bit/piexl左右,均方误差(MSE)也要小于后两者的均方误差。
A low complexity and efficient adaptive prediction method for image predictive coding was presented. The predictor employed LMS（Least Mean Square） adaptive filtering technology and improved predicted values by using local mean estimation substructure method,which overcame the non-zero and non-stationary of the image＇s characteristics,met the requirements of LMS algorithm well and improved the capability of predicting. The simulation results of different images show that the entropy of the prediction error image using the presented method is less than the results of GAP algorithm and MED algorithm about 0.1 bil/pixel,and the Mean Square Error（MSE） of the prediction error image is also smaller than the latter two algorithms.
出处
《电子设计工程》
2011年第4期109-112,共4页
Electronic Design Engineering
基金
国家重点实验室基金资助项目(9140C5305020706)
关键词
图像压缩
二维LMS算法
预测编码
自适应预测
image compression
2-D LMS algorithm
prediction coding
adaptive prediction
