一种新的基于目标和特征的图像分类框架

A NOVEL IMAGE CLASSIFICATION SCHEME BASED ON OBJECT AND FEATURE

首先提出了一种图像分类的结构。根据子块技术和形状描述技术来获取图像中的目标,如果图像类属于目标图像类,则提取图像中目标图像的特征,否则提取整幅图像的底层特征。然后利用主分量分析方法(PCA)对提取的特征进行降维处理,降维处理后的数据由支持向量机进行分类。该方法在标准的Corel图像库上进行了测试,实验结果表明提出的方法有效地提高了图像分类的性能,图像分类的结果与图像的高层语义概念相一致。

In this paper,a novel image classification scheme is presented. Object image classes are obtained by using sub-block technique and image sharp, if classification image belongs to object image classes, the extraction of feature adopts feature of object image, or adopts global low-level feature of image. Then we use Principal Component Analysis to reduce the dimensionality of feature and use support vector machines to learn to classify the images. The proposed method is tested on a standard Corel image databases, experimental results demonstrate the efficiency of the proposed image classification scheme and the consistency with high-level semantic concept.