鉴于主成分分析方法忽略特征向量不同维度的差异,提出一种特征选择优化方法,即改进的PCA特征降维方法。该方法赋予不同维度的特征向量不同的权重,客观反映图像的颜色分布和边缘特性。将该降维方法应用到图像检索中,降低了图像检索系统的复杂度,检索效率提高了28.2%。
In view of the principal component analysis method ignoring the difference among different characteristic vectors,we put forward a kind of optimization method of feature selection,namely,the improved PCA feature dimension reduction method. This method entrusts with different weights to different dimensions of feature vectors,and objectively reflects the color distribution and the edge characteristic of images. Applied to image retrieval,this method reduces the complexity of image retrieval system,and improves the retrieval efficiency by 28. 2%.