光学学报, 2018, 38 (2): 0210004, 网络出版: 2018-08-30
基于快速仿射模板匹配和改进仿射迭代最近点算法的人脸稳健精确仿射配准 下载: 794次
Robust and Precise Affine Registration for Faces Based on Fast Affine Template Matching and Modified Affine Iterative Closet Point Algorithm
图像处理 图像配准 人脸仿射配准 仿射变换 粗搜索 精确配准 image processing image registration face affine registration affine transformation rough search fine registration
摘要
提出了一种仅基于单个模板的人脸仿射配准方法。首先,为了克服人脸仿射变换而产生的局部形变,引入颜色特征来平衡模板人脸和目标人脸之间的颜色相似性和形状错配率,进而提出了一种基于颜色特征的人脸粗搜索算法。接着,采用人脸粗搜索算法得到的仿射变换作为初始约束,建立上步仿射约束下的人脸形状精确配准算法。在算法的每一步迭代中,利用前一步迭代得到的仿射变换,建立最近点的对应关系,并利用前步仿射约束下的目标函数求解新的仿射变换。本文算法成功解决了旋转、缩放和噪声干扰情况下人脸形状难以配准的问题。与传统人脸仿射配准算法相比,本文算法有效提高了人脸仿射配准的稳健性和精确性。
Abstract
We propose a method for face affine registration based on a single template. Firstly, in order to overcome the local deformation caused by face affine transformation, the color feature is introduced to balance the color similarity and shape mismatch rate between the template face image and the target face image, and then the face rough search algorithm is proposed based on the color feature. Secondly, the affine transformation obtained by face rough search algorithm is used as an initial constraint to establish the face shape fine registration algorithm based on the previous affine constraints. In each iteration step of the algorithm, the affine transformation obtained in the previous iteration is used to establish the correspondence of the nearest point, and the new affine transformation is solved by use of the objective function based on the previous affine constraint. The proposed algorithm successfully solves the problem that a face shape is difficult to register under the conditions of presence of rotation, scale transformation, and noise interference. Compared with the traditional face affine registration algorithm, the proposed algorithm effectively improves the robustness and accuracy of face affine registration.
吴礼洋, 熊磊, 杜少毅, 毕笃彦, 方挺. 基于快速仿射模板匹配和改进仿射迭代最近点算法的人脸稳健精确仿射配准[J]. 光学学报, 2018, 38(2): 0210004. Liyang Wu, Lei Xiong, Shaoyi Du, Duyan Bi, Ting Fang. Robust and Precise Affine Registration for Faces Based on Fast Affine Template Matching and Modified Affine Iterative Closet Point Algorithm[J]. Acta Optica Sinica, 2018, 38(2): 0210004.