基于SIFT,K-Means和LDA的图像检索算法

图像检索一直是信息检索领域的难题。提出了一种基于尺度不变特征变换（SIFT，Scale Invariant Feature Transform），K-Means和潜在狄利克雷分布（LDA，Latent Dirichlet Allocation）的图像检索算法。算法主要分为两个阶段。预备工作得到分类完成的图库、概率分配参数表和基本词库；实现检索是在预备工作的基础上归类测试图片，然后在该类下搜索最相似图片。对比传统的基于文本或内容的检索方法，该算法在检索之前将图片库中所有图片按其本身特征进行自动分类，取代人工标注图像信息的过程，同时由于整个算法完全基于图像特征，故此方法不会引入人工因素的干扰。实验结果表明，该算法能够较为准确地将要检索的图片归为图片库对应的类别中，有效地提高图像检索效率。

Image retrieval algorithm based on SIFT,K-means and LDA

Image retrieval is a problem in the field of information retrieval. An algorithm was developed for image retrieval based on scale invariant feature transform (SIFT), K-Means and latent dirichlet allocation (LDA). This algorithm was mainly divided into two stages. The preparations obtained the classified image library, the probability distribution of parameters table and the base vocabulary library; the retrieval classified the test image based on the preparations, and looked up the most similar image. Compared with the traditional methods based on text or content, the algorithm classifies automatically all the images in the library before the retrieval, which can replace the process of manual label. Meanwhile, the algorithm is based on image feature fully, which will not introduce artificial disturbances. Experimental results show that the algorithm can classify accurately the test image as the corresponding category, which can increase efficiency of the retrieval.