基于视觉词袋与颜色矩的图像检索方法  

Image Retrieval Method Based on Visual Bag-of-Words and Color Moments

作　　者：林勇强 洪茂雄 Lin Yong-qiang;Hong Mao-xiong(School of Computer Science and Mathematics,Fujian University of Technology,Fuzhou 350118,Fujian Province,China)

机构地区：[1]福建理工大学计算机科学与数学学院,福建福州350118

出　　处：《科学与信息化》2025年第2期104-106,共3页Technology and Information

摘　　要：针对视觉词袋(Bag of words,BoW)图像检索模型中SIFT特征只提供局部梯度描述,其编码在图像中的其他鉴别信息仍然没有被利用,以及局部特征之间缺乏几何约束等问题,提出一种基于视觉词袋与颜色矩的图像检索方法。该方法在BoW模型基础上,通过后期融合图像颜色矩特征增加图像的颜色信息,以及对得到的检索列表使用RANSAC算法来对查询图像与检索列表靠前的图像特征位置间拟合单应性增加位置信息。In view of the issues in the Bag of Words(BoW)image retrieval model that SIFT features only provide local gradient descriptions,and other discriminative information encoded in the image has not been utilized,as well as the lack of geometric constraints among local features,a novel image retrieval method based on visual bag-of-words and color moments is proposed.This method,built upon the BoW model,incorporates image color moment features in the post-processing stage to enrich the color information of the image.Moreover,it applies the RANSAC algorithm to the retrieved image list to fit homography between the query image and the top-ranked images in the list,thereby enhancing the positional information.

关 键 词：图像检索 视觉词袋 特征融合 颜色矩 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]