融合区域特征和相关反馈的图像检索

Image retrieval based on region features and relevance feedback

提出了一种新的基于内容的图像检索算法,该算法提取图像的颜色-空间特征,在HSV空间中将图像按照H分量进行区域划分,利用区域特征层面上的相似度对图像进行检索,并引入用户的相关反馈来调整并记录示例图像中各对象的特征权值。实验结果表明,该算法可以使计算机更加精确地理解用户的查询要求,提高查询的准确率。

The retrieval algorithm uses a method based on color-spatial features,translates the images into HSV color space,and cuts the images into regions with the H values.The algorithm uses the relevance feedback from the users into the retrieval algorithm,and proposes a method to change and save the information users provided during the retrieval progress.The experiment shows that the algorithm can greatly improve the retrieval precision.