中文摘要：

传统的ROI（region of interest）检测方法忽略了图像中的上下文信息,为了解决这个问题,本文利用概率潜在语义分析（probability latent semantic analysis,PLSA）来对图像中的每块区域周围的图像特征进行分析,并利用其作为上下文特征来辅助ROI的检测。实验表明,该方法与直接进行分类相比,能够取得更好的分类效果。

英文摘要：

In order to solve the problem that context information was often missed in region of interest（ROI） detection,probability latent semantic analysis（PLSA） was employed to extract the context information from surrounding regions,and the context information was used as context features to aid ROI detection.Experimental results showed that the context information can effectively improve the accuracy of ROI detection.