摘要
针对图像理解中所需的图像检索,提出了一种新的图像检索方法。该方法将非负稀疏编码引入到ScSPM算法中进行图像的特征提取和表示,计算特征表示后图像之间的欧氏距离并排序。实验结果表明该方法在图像理解中能够有效地检索相关图像。
This paper proposes a novel method of scene retrieval--scene retrieval based on non-negative sparse coding--in the field of image understanding. It adopts the algorithm of ScSPM, combines the algorithm of non-negative sparse coding to learn feature extraction and representation, and retrieves a set of K-nearest neighbors in database, then rank the Euclidean distance from the query image to each image. The experimental results show the proposed method of scene retrieval has good performances in the field of image understanding.
出处
《信息技术》
2013年第1期39-42,共4页
Information Technology
基金
教育部新世纪优秀人才支持计划(NCET-10-0327)
关键词
图像检索
非负稀疏编码
图像理解
特征提取
特征表示
scene retrieval
non-negative sparse coding
image understanding
feature extraction
feature representation
