摘要
相关反馈技术和主动学习的提出,缩小了图像低层特征与高层特征的语义鸿沟,进一步提高了检索准确性。支持向量机应用在图像检索中,以所需图像特征作为训练样本数据,通过选取合适的核函数以及相关参数设置构建分类器,对测试样本进行分类,从而检索出相似度更接近的图像。本文融合了图像颜色与纹理综合特征,通过支持向量机学习机制构造分类器,并引入主动学习,实现图像检索。实验表明,引入主动学习可提高检索准确性。
Relevance feedback technology and active learning proposed reduce the semantic gap between the image low-level features and high- level features,further improve the retrieval accuracy. Support vector machine in image retrieval,image feature desired as training data,by selecting the appropriate kernel and related parameters set to build a classifier,to classify the test samples,thereby to retrieve an image closer similarity. This article combines the image color and texture synthesis features,constructs the classifier using support vector machine learning mechanism,and by the introduction of active learning,realizes image retrieval. Experimental results show that the introduction of active learning can improve the retrieval accuracy.
出处
《智能计算机与应用》
2015年第6期24-26,共3页
Intelligent Computer and Applications
