基于Markov随机场的自适应正则化三维显微图像复原

提出了基于马尔可夫随机场模型的正则化因子自适应调整三维显微图像复原算法,并用模拟序列样本和真实生物样本进行了实验.为了保持复原图像的边缘等细节信息,以Markov随机场模型作为图像的先验概率模型,对代价函数添加边缘约束惩罚项.其中,正则化因子在迭代过程中自适应地进行更新.实验结果表明此算法在对原始图像进行估计的同时,能够有效地保留图像的边缘等细节信息.而EM算法虽然能够有效地去除层间干扰,却丢失了大量的细节信息.

Restoration of 3-Dimensional Microscopical Images Based on Markov Random Field with Adaptive Regularization Parameter

An algorithm based on Markov random field model for 3-Dimensional microscopical image restoration with adaptive choice of regularization parameter is proposed.And its performance is illustrated with synthetical and real biological data.To preserve the discontinuities during the restoration process,an additional edge penalized term with Markov prior probability is taken into account in the cost function.The regularized parameter is automatically updated based on the current iterative results during the iteration process.Experiments show that the proposed algorithm has the ability of preserving discontinuities while estimating the original images from degraded ones.In contrast,although the EM algorithm has the ability to effectively remove interference between layers,it also lost lots of detail information of the original images.