摘要： 研究先更新再预测的经典自适应提升小波算法，提出一种基于自回归(AR)模型的二维自适应提升小波变换算法。根据图像局部特性选择自适应更新算子，利用更新后的系数位置关系给出基于AR模型的预测算子，使预测误差功率最小。实验结果表明，与使用最小均方误差标准的自适应预测算法相比，该算法能够降低高频系数能量，且峰值信噪比也有所提高。

关键词: 自适应小波变换, 提升方案, 图像压缩, 自回归模型, 最小预测误差

Abstract: This paper proposes a new algorithm for 2D adaptive lifting wavelet transform, which suites for the task of image compression applications. It is based on an update lifting operator and a prediction lifting operator according with p-order AR model of an image. It can get the coefficients of the predict filter to minimize the power of predictor error. Experimental results show that the proposed algorithm is competitive for the image compression, in terms of the decrease of the entropy of the detail coefficients and the increase of the PSNR.

Key words: adaptive wavelet transform, lifting scheme, image compression, AR model, minimum prediction error

中图分类号: 

• TN911.73