基于SURF特征的汽车轮毂型号识别  

Vehicle Hub Type Recognition Based on SURF Features

作　　者：侯永涛[1] 黎良臣 顾寄南[1] 冒文彦 HOU Yong-tao;LI Liang-chen;GU Ji-nan;MAO Wen-yan(School of Mechanical Engineering,Jiangsu University,Jiangsu Zhenjiang 212000,China)

机构地区：[1]江苏大学机械工程学院,江苏镇江212000

出　　处：《机械设计与制造》2021年第8期5-7,12,共4页Machinery Design & Manufacture

基　　金：国家自然科学基金(51875266)。

摘　　要：提出了一种基于SURF特征与神经网络相结合的方法,实现了对多种型号轮毂的识别。首先,将所有采集的样本图像分成训练集、验证集和测试集;然后,将所有样本图像缩放至合理大小并进行转灰操作;提取所有样本图像的SURF特征并导入到搭建好的神经网络中进行训练、验证和测试,通过神经网络强大的学习能力,从对大量特征数据的学习过程中获得一个最佳的识别模型。该方法可以以最少的样本图像获得一个能准确识别轮毂型号的识别模型。另外,该方法鲁棒性好、抗干扰能力强,能满足自动化生产线实时性的要求。The proposes a method based on the combination of SURF features and neural network to realize the recognition of various types of hubs.First,all collected sample images are divided into training set,verification set and test set;then,all sample images are scaled to a reasonable size and grayed out;the SURF features of all sample images are extracted and imported into the built neural network to train,verify and test.Through the powerful learning ability of the neural network,an optimal recognition model is obtained from the learning process of a large number of feature data.The method can obtain a recognition model that accurately identifies the hub type with a minimum sample images.In addition,the method has good robustness and strong anti-interference ability,and can meet the real-time requirements of automated production lines.

关 键 词：机器视觉 汽车轮毂 特征 识别 实时性 

分 类 号：TH16[机械工程—机械制造及自动化] TP391[自动化与计算机技术—计算机应用技术]