基于吉伯斯随机场和最大后验概率的医学序列图像伪光流估计

Pseudo-optical flow estimation for medical image sequences based on maximum a posteriori probability and Gibbs random field

目的提出一种医学序列图像消除帧间相关的新算法，以利于图像信号的压缩、存储和传输。方法把医学序列图 像假想为运动序列，然后根据最大后验规则，利用迭代条件模型来求解这个伪运动序列的伪光流场，达到消除帧间相 关的目的。结果本算法能够有效地消除医学序列图像的帧间相关，它比仅利用帧内相关的算法更为有效。结论医学 序列图像存在大量帧间冗余信息，消除这种冗余信息将显著地提高医学序列图像的压缩比。

Objective: To propose a new scheme for removing inter-frame redundancies of medical image sequences to facilitate the compression, memory and transmission of the images. Methods The medical image sequences were assumed to be in motion, and the pseudo-optical flow field was estimated by using iterative condition mode based on maximum a posteriori probability. Results By using the algorism proposed in this paper, the inter-frame redundancies were significantly reduced. This algorism was more effective in medical image sequence compression than those involving merely inter-frame correlations. Conclusion There are large amount of inter-frame correlations in medical image sequences, which can be reduced to improve the compression ratio.