摘要
图像检索是信息挖掘领域的重要方法,针对单一特征图像检索方法鲁棒性不足的问题,文章提出了一种基于组合特征检索策略的图像检索方法。首先,求取空间颜色直方图特征、局部二值阵列特征、基于轮廓的Hu矩特征,分别作为颜色特征、纹理特征、形状特征的表达;再对3种特征分别配置相应的权重,进而形成组合检索特征。实验结果表明,基于组合特征检索策略的图像检索方法明显提高了检索性能,并且具备较好的灵活性。
Image retrieval is an important method of information mining. Aiming at the problem of the lack of the robustness in retrieval methods based on single feature, a new image retrieval method based on combined feature retrieval strategy is proposed. The spatial color histogram, local binary array and contour based Hu moment are computed first as color feature, texture feature and shape feature respec- tively. Then the combined retrieval feature is produced by arranging different weights to the three fea- tures. The experimental results show that the retrieval method based on combined feature retrieval strategy has better retrieval performance and good flexibility.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2015年第7期918-922,共5页
Journal of Hefei University of Technology：Natural Science
基金
国家863高技术研究发展计划基金资助项目(2013AA01A215)
河南省高等学校青年骨干教师计划资助项目(2011GGJS-198)
河南省教育厅科学技术重点研究资助项目(14A520045)
