项目名称: 手机人脸识别与验证关键技术研究
项目编号: No.61262040
项目类型: 地区科学基金项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 陈熙
作者单位: 昆明理工大学
项目金额: 43万元
中文摘要: 基于手机的人脸识别/认证技术在手机权限保护、移动远程人脸识别/认证和移动电子商务等领域发挥着越来越重要的作用。但在现实环境下,手机的计算能力、内存大小和人脸表现形式的复杂性限制了其实际应用。因此,本项目结合手机人脸识别/认证的实际需求,围绕复杂环境下低复杂度手机人脸检测和特征提取两个科学问题,开展手机人脸识别/认证关键技术研究。主要创新研究包括:(1)真实光照环境下手机人脸检测与特征提取算法研究;(2)低分辨率条件下的手机人脸特征描述研究;(3)基于分数低阶统计理论的带遮挡手机人脸特征提取算法研究;(4)手机视频人脸建模研究;(5)稀疏样本情况下手机人脸特征提取研究;(6)可撤销人脸模板研究;
中文关键词: 基于Hammerstein多项式展开的直接鉴别局部保持投影;局部分形自相关;局部凸凹模式;非高斯二维Gabor滤波器;矩阵随机低秩近似
英文摘要: Cell phone based face recognition / authentication technologies in the mobile phone rights protection, the mobile remote recognition / authentication, and mobile e-commerce plays an increasingly important role. But in the real environment, the cell phone's computing power, memory size and the complexity of the face limit its practical application. Therefore, this project combined with the actual needs of the cell phone face recognition / certification, centered on two scientific issues about low complexity degree face detection and feature extraction in the complex environment, studied the cell phone face recognition / authentication key technologies. The main innovations include: (1) Cell phone face detection and feature extraction algorithm in the real illumination environment;(2) Face characterization of cell phones in the low-resolution conditions;(3) Facial feature extraction algorithm for cell phone face with block based on the fractional low-order statistic theory ;(4) Research on face modeling in cell phone video; (5)Research on face modeling with sparse samples; (6)Research on cancellable facial template generation;
英文关键词: Direct discriminant locality preserving projection;local fractal auto-correlation;local convex-and-concave pattern;non-Gaussian two-dimensional Gabor filters;matrix random low-rank approximate