基于PCA空间的自适应相似近邻保持投影算法
非相度保持投影算法是一种有效的特征提取算法,该算法无须参数设置且识别性能稳定,但算法的运算量大,并且存在着小样本问题,因此提出了一种基于PCA空间的自适应相似近邻保持投影算法。该算法为了减少权值的计算量,提出直接利用PCA处理过的样本进行近邻计算,以归一化的样本平方欧氏距离来构造相似权值;而为了解决小样本问题提出了最大化差异形式的目标函数,从而有效地解决了小样本问题,同时也提高了算法的识别精度。最后人脸库上的实验结果表明所提算法的有效性。...
Saved in:
| Published in | 计算机应用研究 Vol. 32; no. 12; pp. 3821 - 3824 |
|---|---|
| Main Author | |
| Format | Journal Article |
| Language | Chinese |
| Published |
安徽理工大学计算机科学与工程学院,安徽淮南,232001%安徽理工大学机械工程学院,安徽淮南,232001
2015
|
| Subjects | |
| Online Access | Get full text |
| ISSN | 1001-3695 |
| DOI | 10.3969/j.issn.1001-3695.2015.12.070 |
Cover
| Summary: | 非相度保持投影算法是一种有效的特征提取算法,该算法无须参数设置且识别性能稳定,但算法的运算量大,并且存在着小样本问题,因此提出了一种基于PCA空间的自适应相似近邻保持投影算法。该算法为了减少权值的计算量,提出直接利用PCA处理过的样本进行近邻计算,以归一化的样本平方欧氏距离来构造相似权值;而为了解决小样本问题提出了最大化差异形式的目标函数,从而有效地解决了小样本问题,同时也提高了算法的识别精度。最后人脸库上的实验结果表明所提算法的有效性。 |
|---|---|
| Bibliography: | Lin Yu＇ e, Guo Yongcun, Li Jingzhao （ a. School of Computer Science ＆ Engineering, b. College of Mechanical Engineering, Anhui University of Science ＆ Technology, Huainan Anhui 232001, China） 51-1196/TP Dissimilarity preserving projection is an effective feature extraction algorithm, which does not need to set parameters and has stable recognition performance. However, this method not only needs a lot of computation time but also encounters the small sample size problem. Therefore, this paper proposed an adaptive similarity neighborhood preserving projection algorithm based on PCA space. In order to reduce the amount of calculation weights between pairs of samples, the algorithm directly used the normalized squared Euclidean distance, which was calculated using samples processed by PCA algorithm, to construct the weight matrix. In such a way, the difference form of criterion function not only effectively solved the small sample size problem but also improved the recognition performance of the algorithm. Experim |
| ISSN: | 1001-3695 |
| DOI: | 10.3969/j.issn.1001-3695.2015.12.070 |