摘要
基于词包模型的图像表示方法是目前应用最广泛的特征表示方法之一,特征编码是该模型中非常重要的环节。针对已有编码方法未考虑语义信息的缺点,提出了基于局部性约束和视觉显著性的特征编码方法,并用于图像分类。在5个标准图像库进行实验和分析,结果表明融入显著性语义信息的图像编码方法能够提升分类性能。
Bag of feature(BoF)model is a most popular image representation method by now.Feature coding is a very important phase in BoF model construction.Aiming at the weaknesses of lacking semantics of current coding methods,the locality constraint and saliency based feature coding method is proposed.The new image representation can be used for image classification.Experiments were carried on five bench-mark datasets.The experimental results show semantics of saliency can assist feature coding and prove the efficiency of image classification.
作者
梁晔
马楠
许亮
桂雨晗
Liang Ye;Ma Nan;Xu Liang;Gui Yuhan(Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China;College of Robotics,Beijing Union University,Beijing 100101,China)
出处
《北京联合大学学报》
CAS
2020年第1期57-62,共6页
Journal of Beijing Union University
基金
北京市自然科学基金项目(4182022)
国家自然科学基金项目(61871038,61871039)
关键词
视觉显著性
特征编码
图像分类
局部性约束
Visual saliency
Feature coding
Image classification
Locality constraint
