【摘要】 提出一种基于三元空间融合的模糊人脸图像特征相似度识别方法。通过建立模糊人脸图像特征相似度投影矩阵,将模糊人脸图像映射到同一个特征表示空间,得到维数一致的特征向量;针对10种LIH特征相似度进行分析,确定模糊人脸图像特征相似度信息能量百分比;使用三元空间融合技术中编码的方式描述模糊人脸图像中的特征描述子,基于三元空间融合提取局部特征描述子;通过设计对称表示相似性度量方法,取得最完整的特征相似度识别结果。实验结果表明,该方法可以实现模糊人脸图像特征相似度精准识别,其特征相似度识别分辨率明显高于传统方法。更多还原

【Abstract】 In this paper, a fuzzy face image feature similarity recognition method based on ternary space fusion is proposed. By establishing the projection matrix of fuzzy face image feature similarity, the fuzzy face image is mapped to the same feature representation space to obtain the feature vector with the same dimension; the 10 LIH feature similarities are analyzed to determine the percentage of information energy of the fuzzy face image feature similarity; the feature descriptors in the fuzzy face image are described by using the coding method in the ternary space fusion technology, and the local feature descriptors are extracted based on ternary space fusion; the most complete feature similarity recognition results are obtained by designing a symmetric representation similarity metric. Experiment results show that this method can realize accurate recognition of fuzzy face image feature similarity, and its feature similarity recognition resolution is significantly higher than that of traditional methods.更多还原