摘要
提出一种基于颜色和形状特征的图像检索方法,该法采用改进的颜色直方图表示图像的颜色特征,利用小波矩提取图像的形状特征和颜色特片,与候选图像的特征进行相似度计算。为进一步提高图像检索的质量,提出一种基于支持向量机的相关反馈方法。实验结果表明,该方法具有较好的检索性能。
A new method of combining color, shape feature for efficient image retrieval is proposed. Color feature is extracted by the histogram of colors; shape analysis is done using wavelet moment. Then this paper implements similarity measurement of query image and image database by taking into consideration the similarity of both color and shape feature. In order to improve retrieval performance, a relevance feedback mechanism, based on support vector machines is proposed. Experiments show that the proposed system is not only efficient but also effective.
出处
《太原理工大学学报》
CAS
北大核心
2008年第2期127-131,共5页
Journal of Taiyuan University of Technology
基金
国家863基金资助项目(2005AA414010)
浙江省自然科学基金资助项目(M603034)
绍兴市基金资助项目(xk06009)
关键词
颜色直方图
小波矩
图像检索
支持向量机
相关反馈
color histogram
wavelet moment
image retrieval
support vector machine
relevance feedback
