基于动画头像的特征提取和识别研究

本文从动画头像入手,将头像进行灰度变换,然后进行了三种图像压缩方法,从而进行特征提取,应用2DPCA,2DLDA,PCA.最后对图像进行识别操作,采用SVM和最近邻法操作,从结果中可以看出2DLDA,2DPCA的正确率很高,2DLDA算法的计算速度更快。

Feature extraction and recognition based on game head portrait

This paper starts from the game head portrait, do gray-scale transformation, and then use the three kinds of image compression method, which is used for feature extraction, the main application is 2DPCA, 2DLDA, PCA. and then do recognition operation on the image, the main use of SVM and nearest neighbor method, it can be seen from the results that the correct rate of 2DPCA and 2DLDA is very high, the 2DLDA algorithm has faster calculation speed.