基于空间加权Bag of Features模型的图像检索
首发时间:2012-04-11
摘要:基于图像的视觉特征从差异很大的图像数据库上检索图像是一项具有挑战性的任务。近年来,基于Bag of Features(BoF)的方法在基于内容的图像检索上取得了显著的效果。然而,由于BoF方法将图像表示成局部描述子的无序组合,只考虑了各个视觉特征的密度信息,而却没有提供图像特征间的空间信息。在本文中,我们提出了两种全新的空间加权方法,一种方法对图像的子块计算局部方差作为子块内的特征点的权重,另一种方法通过计算图像子块之间的相邻关系来获得图像子块内特征点的权重。最后根据图像背景特征,本文还提出了级联检索策略。本文提出的方法简单,执行效率高,而且相比无序的'BoF'方法,检索准确率有了显著的提高。
关键词: 图像检索 Bag of Features(BoF) 空间加权 级联检索
For information in English, please click here
Image Retrieval Based on Spatial Weighted Bag of Features
Abstract:Visual Features based image retrieval on large database is challenging task. Recent years, methods based on Bag of Features (BoF) model gain attractive result. However, since the 'bag of features' approach represents an image as an unordered collection of local descriptors which only use the intensity information, the resulting model provides little insight about the spatial constitution information of the image. In this paper, two novel spatial weighting approaches were proposed. One method based on local variance of blocks of an image, the other based on the adjacent relation of the blocks. After analyze the background properties of images, we develop a cascaded Retrieval approach. Methods developed in this paper are simple and effective, and compare to the traditional orderless Bag of Features model, proposed methods significantly improve the performance.
Keywords: image retrieval Bag of Features spatial weighting cascaded Retrieval
基金:
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
基于空间加权Bag of Features模型的图像检索
评论
全部评论0/1000