本文已被：浏览 1077次 下载 553次
投稿时间：2010-07-12
投稿时间：2010-07-12
中文摘要: 探讨了相关向量机的分类原理及其在物体识别中的应用，其核函数无需满足mercer条件，且不需要误差参数的实验调整。提出了一种基于物体显著区域的特征描述方法，在有效提取物体特征的同时，大大减少了描述物体的特征量。实验结果表明，相关向量机不仅具有与支持向量机相同的性能，而且其相关向量较少，并取得了较好的识别效果。
Abstract:Both classification principle and application for object identification of the Relevance Vector Machine (RVM) are discussed.The RVM doesn't necessitate the estimate of a trade-off parameter and the satisfaction of the Mercer kernel functions. Anovel object descriptor based on salience field is proposed, which reduces the quantity of features for object discription.Experiment results demonstrate that the RVM achieves comparable recognition accuracy to SVM with substantially fewer vectors, and produces promising results for object identification.
keywords: object identification relevance vector machine support vector machine salience field features
文章编号：20110216 中图分类号： 文献标志码：
基金项目:
引用文本：
邵洁,董楠.基于相关向量机的物体识别[J].上海电力大学学报,2011,27(2):171-175.
SHAO Jie,DONG Nan.Object Identification Based on Relevance Vector Machine[J].Journal of Shanghai University of Electric Power,2011,27(2):171-175.
邵洁,董楠.基于相关向量机的物体识别[J].上海电力大学学报,2011,27(2):171-175.
SHAO Jie,DONG Nan.Object Identification Based on Relevance Vector Machine[J].Journal of Shanghai University of Electric Power,2011,27(2):171-175.