基于词频统计特征和 GVP 的大规模图像检索算法研究

针对传统的GVP(Geometry-Preserving Visual Phrases)图像检索算法计算量大、时间复杂度高且不适合处理大规模图像检索等缺点,文章提出了FSF-GVP(Frequency Statistics Feature-Geometry-Preserving Visual Phrases)算法,该方法将词频统计特征和GVP算法相结合,使用GVP排序算法对词频特征统计后的相似结果集进行排序,忽略不相似结果集,极大地提高了检索效率。实验结果表明,FSF-GVP在保证检索准确性的前提下,提高了检索效率,适用于实时大规模图像检索。

Image Retrieval Using Feature Word Frequency Statistics of Geometry-Preserving Visual Phrases

Traditional GVP (geometry-preserving visual phrases) image retrieval algorithm is not suitable for handling the large-scale image retrieval because of its high time complexity. In this paper, FSF-GVP (frequency statistics featuregeometry- preserving visual phrases) algorithm, which combined word frequency statistic characteristics and GVP algorithm, was proposed. FSF-GVP algorithm counts visual word frequency characteristics of an image to be searched and image database to get similar result set and dissimilar result set. Then FSF-GVP algorithm uses the GVP algorithm to sort the similar result set, which improves the retrieval efficiency. The experiment results on Oxford 5K dataset show that FSFGVP is suitable for the large-scale real-time image retrieval on the premise of ensuring the accuracy of retrieving result and improving the retrieval efficiency.