基于小波域HMT模型融合的图像分辨率增强

图像的分辨率可以通过预测细节子带中的小波系数来得到提高.采用Gaussian混合模型的小波域隐马尔可夫树可以精确地描述真实图像的统计特性,非常适合于预测图像的细节子带.但是这类方法的缺点是被预测的小波系数是随机生成的,每次算法得到的结果均不相同,只能从中选择一个作为最终结果.本文提出了一种算法,将随机结果根据局部方差融合规则融合到一起,从而产生一幅更适合人类视觉系统的图像.实验结果证明了本文算法的有效性,其主观和客观评价指标均优于传统算法.

Image Resolution Enhancement Based on Wavelet Domain HMT and Fusion

Image resolution enhancement algorithms based on the estimation of detail wavelet coefficients at high resolution scales have been proposed recently. These algorithms assume that the low resolution image is the approximation sub-band of a higher resolution image and interpolates images by predicting coefficients at finer scales. Hidden Markov Tree (HMT) in the wavelet domain using Gaussian mixture models are capable of accurately modeling the statistical behavior of real world images by exploiting relationships between coefficients in different scales and have shown to produce promising results. However, one drawback of these methods is that, the coefficients to be estimated are generated randomly, so the results are different every time, only one of them is chosen finally. In this paper, we propose an algorithm which fuses these random results together by means of the fusion rules based on area-based standard deviation. This makes the enhanced image more suitable for the human vision and reduces the disorder degree of the image. Experiments demonstrate the effectiveness of the proposed method and show the superiority to previous methods in objective and subjective qualities.