大型网络下的特征高效匹配模型仿真分析  

The Characteristics Efficient Matching Model Simulation Analysis of Large Network

作　　者：王中华[1] 汪文彬[1] 

机构地区：[1]琼州学院电子信息工程学院,海南三亚572022

出　　处：《科技通报》2015年第7期200-203,共4页Bulletin of Science and Technology

基　　金：海南省自然科学基金(613170)

摘　　要：在对大型网络下的特征进行高效匹配的过程中,容易出现数据量大、计算过于复杂的情况,导致传统的基于多谱特征表示的大型网络下的特征匹配模型,由于需构造所有谱描述子,无法高效完成大型网络的特征匹配,提出一种基于加速鲁棒性特征算法的大型网络下特征高效匹配模型,对个体分类器的权值进行计算,依据特征出现的加权次数从大到小进行排序,从而完成大型网络所需特征的选择。通过固定的梯度方向矢量获取标准模板,利用标准模板得到一套适用于所有大型网络下的特征匹配算子,获取大型网络图像的二阶算子,依据积分获取所有固定梯度方向矢量,对Haar小波二阶响应进行计算,依据高斯函数对该小波响应进行加权处理,从而提高其对大型网络下特征高效匹配的鲁棒性。仿真实验结果表明,所提方法具有很高的准确性及高效性。Under the large network characteristics of high efficiency in the process of matching, it's easy to have a large amount of data and calculation is too complex, result in traditional based on spectral characteristics of the feature matching model in the case of large networks, due to the need to construct all spectrum descriptor, unable to efficiently complete the large network of feature matching, puts forward a kind of robustness based on acceleration features algorithm under the large network of efficient matching model, through the calculation of the weights of individual classifier based on the characteristics of the weighted number from big to small sort, which features needed to complete the large network of choice. Are acquired by using gradient direction vector of fixed standard template, using the standard template to get a set of applicable to all feature matching operator under large networks, access to large network image of second-order operator, based on the integral obtain all fixed direction of gradient vector, response calculation of Haar wavelet second order, on the basis of gaussian function weighted processing of the wavelet response, so as to improve the characteristics of large network efficient matching of robustness. The simulation results show that the proposed method has high accuracy and high efficiency.

关 键 词：大型网络 特征匹配 高效 

分 类 号：TP133[自动化与计算机技术—控制理论与控制工程]