摘要随着互联网技术的迅速普及和发展,互联网上的图像数量迅速增加,如何能够在海量图像中寻找到用户满意的结果已成为一项十分有挑战性的任务。“语义鸿沟”即底层视觉特征和高层语义之间存在着不一致性,需要对图像内容进行分析与理解变得十分困难。为此,需要对图像进行语义标注,自动地理解图像的视觉内容。本论文重点研究了基于图模型的语义标注问题,利用信息传播的思想对图像标注相关语义标签。本文重点研究了三种图模型方法,即K近邻(K-Nearest Neighbors,KNN)算法、稀疏图方法、以及K近邻稀疏图方法。最后,本文在当前流行的数据集NUS-WIDE上对上述三种图构建方法进行了实验验证和分析。25434
关键字:语义鸿沟、图像标注、图像检索、图模型 毕业论文设计说明书外文摘要
Title Graph-based Image Annotation
Abstract
With the rapid popularization and development of Internet technology,the number of images on the Internet is increasing rapidly,so the question that how can we find the satisfying results in the mass of images has become a very challenging task.Semantic gap that exists inconsistencies between low-level visual and high-level semantic features makes image content analysis and understanding very difficult.To this end,we need to annotate images,and automatically understand the visual content of images.This paper focuses on the issue that graph-based image annotation,using the ideology of information dissemination to annotate semantic tags.This paper concentrates on three diagram model approach,namely K-Nearest Neighbors algorithm,sparse graph methods,and KNN sparse graph method.Finally,in the current popular data sets NUS-WIDE,this paper carries out experiment verification and analysis for the above three graph construction methods.
Keywords Semantic gap image annotation image retrieval graph model
目 次