结合多方向特征与最优匹配的幻觉脸技术

Face hallucination based on multi-orientation features and optimal matching

研究了基于多分辨率塔式结构的幻觉脸算法。在Baker方法水平和垂直一、二阶灰度特征的基础上,增加了两个斜方向的一、二阶导数灰度特征,弥补了Baker方法建立的金字塔提取的高频信息不够丰富的缺点,提取了更全面的图像特征信息,使匹配更为准确。提出了一种自适应最优匹配方法,使匹配复原结果获得了更全面和准确的高频信息,消除了噪声干扰。对IMDB人脸库进行了实验对比,结果显示,新方法得到了33.5889dB的平均峰值信噪比,高于Baker方法和传统的插值算法,显示出了更好的视觉效果。

A novel face hallucination algorithm based on multi-resolution pyramid structure was proposed.The algorithm added the first and second obliquity derivative gray features,which were combined with the first and second order gray features in horizontal and vertical direction of Baker method.The new method can extract more image feature information and make the matching process more accurate.In addition,a new adaptive optimal matching algorithm was used to matching process.The recovered results can obtain more accurate and comprehensive high frequency information.The experimental results show that the average PSNR of the proposed algorithm can reach 33.588 9 dB in IMDB face database,which is higher than that of other methods and the recovery of face image has a better visual effect.