| 摘要: |
| 图像分类作为图像处理和计算机视觉的重要组成部分，能够快速准确地对数字图像进行分析和管理.对基于bag of word（BOW）模型的分类问题进行了研究，针对图像理解中的图像相似度之间的关系，提出了一种最大间隔最近邻居分类算法，通过对成对约束的度量学习算法，在优化目标中增加原空间数据分类的约束，学习到了一个可以反映当前样本数据的距离函数，并且在k-NearestNeighbor（KNN）分类器上使用该学习到的距离函数来构建分类器，并在多个国际标准图像数据集上进行实验，结果表明：该算法相比传统的基于欧式距离的算法具备更高的正确率. |
| 关键词: 图像分类 词袋模型 大裕度最近邻分类算法 |
| DOI：10.3969/J.ISSN.1000-5137.2017.04.020 |
| 分类号:TP391 |
| 基金项目:国家自然科学基金（61503251） |
|
| An image recognition method based on bag-of-word model and large margin nearest neighbor classification algorithm |
|
Yang Yibo, Wang Bin, Wang Jianfeng
|
|
The College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 200234, China
|
| Abstract: |
| As an important part of image processing and computer vision,image classification can analyze and manage digital images quickly and accurately.This paper studies the classification problem based on bag of word (BOW)model and learns the relationship between image similarities in image comprehension.Then,this paper proposes a maximum-interval Nearest Neighbor classification algorithm.By learning the pair-wise constraint metric and adding the constraint of the original spatial data classification to the optimal target.This paper learns a distance function which can reflect the current sample data and uses this function to construct the classifier according to the k-NearestNeighbor(KNN) classifier.Compared with the traditional Euclidean distance algorithm,the classifier based on metric learning has higher correct rate than the traditional one based on the experiments on several international standard image datasets. |
| Key words: image classification bag of word model large margin nearest neighbor classification algorithm |
Editorial Board