人脸图像虚假皮沟特征剔除算法的仿真分析  

Face the False Skin Channel Characteristics for the Algorithm Eliminate to Simulation Analysis

作　　者：李菊霞[1] 李建平[2] 

机构地区：[1]山西农业大学信息科学与工程学院,山西太谷030801 [2]山西农业大学工学院,山西太谷030801

出　　处：《计算机仿真》2013年第2期418-421,共4页Computer Simulation

基　　金：山西农业大学创新基金(2009009)

摘　　要：由于人脸皮肤沟壑细微褶皱和断纹分布复杂,光照不均匀、纹路较浅和皮肤的多颗粒特征,会使得皮肤的老化沟壑出现很多虚假特征,对正常纹理特征造成干扰,造成人脸细微的皮沟特征发生形变。传统的高斯滤波和Mean Shift提取算法在虚假特征的干扰下,对细微皮沟特征的老化表达出现偏差,造成发生形变的皮沟特征分割结果不准。为优化分割,提出一种人脸虚假皮沟特征剔除分割新算法。首先采用wold人脸纹理模型,利用真实皮沟纹理确定性随机场和虚假皮沟特征不确定性随机场的谱属性不同的特征,将它们分离开。然后利用Harmonic滤波器,允许随机成分通过,滤掉虚假皮沟的不确定性信息。仿真结果证明,对不同粒度的皮肤图像,提出的方法能够较好过滤虚假信息,完成高精度人脸皮沟特征分割。The face skin fine fold and fault distribution pattern complex, uneven illumination, grain is shallow and skin more grain characteristics will make the skin aging ravine appear a lot of false features of the normal texture characteristics cause interference. The traditional gaussian filter and pyramid feature extraction algorithm in false characteristics of interference, the subtle skin channel characteristics of the aging expression appears deviation, cause skin furrow segmentation results no problem. Therefore, this paper puts forward a false face skin channel characteristics to eliminate split the new algorithm. First the wold face texture model, using the real leather channel uncertainty texture with airport and false leather channel uncertainty characteristics with the airport spectral properties of different features, will be separated them. And by using the Harmonic filter, allowing the random component through, filter out false skin channel uncertainty information. The computer simulation proved to different size skin image, the pro- posed method has better filtering false information, to complete the high accuracy men have groove feature segmentation.

关 键 词：皮沟分割 虚假皮沟 剔除算法 

分 类 号：TP127[自动化与计算机技术—控制理论与控制工程]