基于半监督学习的一种图像检索方法

为了提高图像检索的质量, 提出了一种基于半监督学习的图像检索方法。该方法提取图像的颜色、形状、纹理特征, 计算得到已知类别样本的中心图像, 检索过程中利用中心图像自适应调整相关度计算函数, 然后根据与查询图像相关度的大小对图像进行检索和排序。实验结果表明：该方法较已有的基于内容的图像检索方法有更高的查准率, 同时, 由查准率—查全率曲线可知该方法能够达到很好的检索质量。

Image retrieval method based on semi-supervised learning

In order to improve the quality of image retrieval, this paper put forward a semi-supervised learning-based image retrieval method. The method extracted the color, shape and textural features of image, calculated the central image of the classified sample, and then utilized the central image to adjust the correlation calculation function adaptively during the process of retrieval. Finally, it obtained the retrieval and reordered results by the correlation with the retrieval image. The experimental results show that this method has a higher precision than the existing content-based image retrieval method. At the same time, the precision-recall curve shows that this semi-supervised learning-based image retrieval method can achieve a nice retrieval quality.