基于HMRF先验模型的HBE卫星遥感图像超分辨率重建
HBE Satellite Image Super-resolution Reconstruction with HRMF Prior Model
-
摘要: 提出了一种在Bayes概率统计框架下的混合Bayes超分辨率重建算法,该算法采用Huber马尔可夫随机场(Huber Markov random field,HMRF)模型对理想图像进行先验建模,可以较好地突出重建图像的不连续边缘特征信息。实验结果表明,该算法克服了极大后验概率估计(maximum a posteriori,MAP)算法中的若干缺陷,取得了良好的重建结果,图像边缘特征清晰,纹理信息突出。Abstract: A hybrid Bayes estimation(HBE) algorithm for satellite superresolution reconstruction is proposed.Under the Bayesian framework,the optimal value of the ideal high resolution satellite image and the point spread function(PSF) are estimated,and the Huber Markov random field(HMRF) probability density distribution model for the ideal high resolution image by maximum likelihood(ML) estimation are set up.The ill-posed reconstruction problem into well-posed one through adding HMRF prior constraint model is turned into the Bayesian estimation.Experimental results prove that after HBE reconstruction the spatial resolution of remote sensing images gets efficient improvement.
下载:
