基于局部二元模式Gabor小波变换的人脸识别

摘要

This paper proposed a local binary pattern Gabor transform which combines LBP and Gabor transform,encodes the local variety of the multitude of the Gabor features by LBP operator, using histogram to analyze the property of the local variety after LBP encoding,so that preventing information losses as a result of down-sampling and reducing dimensionality. This method combines the advantages of both LBP and Gabor transform,which not only improves the discrimination of the histogram but also significantly reduces dimensionality, Experiments show that the proposed method is more precise than traditional ones.%将局部二元模式(LBP)与Gabor小波变换相结合,用LBP算子对Gabor特征幅值的邻域变化进行编码,并用直方图分析LBP编码后的局部变化属性,这样避免了因下采样造成的信息损失,同时又通过直方图统计达到了降维的目的.实验结果表明,该方法结合了LBP与Gabor变换的优点,增强了直方图的鉴别能力,又能有效降低维数,取得了较理想的识别效果.