基于matlab语言提取图像PHOG特征的检索方法
周美丽;白宗文
【期刊名称】《电子测试》
【年(卷),期】2014(000)006
【摘要】随互联网的普及和发展,电子商务网站的数量急剧增长,电子商务网站需要通过对在线销售商品进行标注以方便用户进行搜索。本文是通过提取类别图像和测试图像的PHOG全部特征,然后计算两者之间的距离,测试图像与类别图像距离比较近的就属于同一类图像。我们利用MATLAB语言设计出了能够实现这一性能的分类并计算的方法。实验证明本文所用的方法灵活性好,准确性高。%The development and popularization of the Internet,the rapid growth of the number of e-commerce web site,sales of commodity on needs the annotation to facilitate users to search for in e-commerce site. In this paper,through the extraction category image and test image PHOG feature,then calculate the distance between the test image and categories.if the distance is in the range,it belongs to the same class of image.We use MATLAB language to design the classification to achieve this performance and calculation method.Experiments show that this method is good flexibility,high accuracy.
【总页数】3页(29-30,9)
【关键词】电子商务;图像检索;matlab;PHOG
【作者】周美丽;白宗文
【作者单位】延安大学,陕西延安,716000;延安大学,陕西延安,716000