基于分块的伪Zernike矩归一化的人脸识别方法

以改进的伪Zemike矩相关知识为背景，提出了一种新颖的人脸识别方法。该方法通过将人脸分块，降低了光照条件、人脸表情等外在因素对人脸识别的影响。首先将人脸图像分块并重组矩阵的行和列，然后计算重组后人脸图像的伪Zemike矩，并对其进行归一化处理，最后采用最小邻近分类器进行判别。实验结果表明，该方法对于人脸光照、姿态和表情变化均具有良好的鲁棒性。在前人的研究基础上，进一步探讨了伪Zemike矩在人脸识别方面的应用。

The Face Recognition Sheme Based on Image Segmentation and Normalization of Pseudo-Zernike Moment

A novel face recognition based on normalization of pseudo-Zernike moment is proposed in this paper. Partitioning the facial image into a few blocks reduces the influence of some factors such as lighting condition and facial expression on face recognition. Firstly, the original sample images are divided into smaller modular images, and matrix is reconstructed. Secondly, the pseudo-Zernike moments of the reconstructed image are computed. Then, the result is normalized. At last, the nearest neighborhood algorithm is used to construct classes . Experimental results demonstrate the presented approach has better robust to variations of illumination and facial expression. On the basis of previous study, a further discussion is made on the application of pseudo-Zernike moments in the aspect of face recognition.