摘要
针对基于内容的商品图像检索,提出了以颜色和区域形状为特征的检索技术方案.以非等间隔量化方法提取颜色直方图,同时设置显著水平阈值提取视觉主色,从而获取颜色特征.利用二维函数矩构造出的Hu矩特征值,并对矩特征值进行权重设置,提取了比较符合人类视觉的区域形状特征.两种特征组合检索,实现了商品图像的较高相似性度量的匹配.
For product image retrieval, the method of retrieval based on color feature and shape feature is proposed. To obtain color feaWue, Non-interval quantitative measure is used to extract color histogram, and setting a value number is to find the primary color. Implementing the algorithm of two-demensional function moment that is weighted to get the shape feature. As these two features are combined in retrieval, the results of math in product image achieve a highly level.
出处
《计算机系统应用》
2012年第9期152-155,共4页
Computer Systems & Applications
关键词
图像检索
颜色特征
形状特征
颜色直方图
二维函数矩
相似性度量
image retrieval
color feature
shape feature
color histogram
two-demensional function moment
similaritymeasure
