物体特征的快速检测与识别方法  被引量:3

A fast test and recognition method for object character

在线阅读下载全文

作  者:王欢欢[1,2] 杨清平 王向军[2] 

机构地区:[1]天津科技大学机械工程学院,天津300222 [2]微光机电系统技术教育部重点实验室(天津大学),天津300072 [3]School of Eng. & Design,Urunel University

出  处:《红外与激光工程》2014年第3期1002-1008,共7页Infrared and Laser Engineering

基  金:国家教育部支撑项目(625010110);微光机电系统技术教育部重点实验室(天津大学)开放基金(KF2K131001);天津科技大学引进人才科研启动基金(201304003)

摘  要:文中的目的是讨论一种新的物体特征快速检测与识别方法,该方法适于具有纹理特征的在运动中视角不断变化的物体快速检测与识别。该方法基于新发展的多级定向执行长度编码法(Multilevel Orientation Run Length Coding,MORLC)。由此产生了两种新的物体特征样本形式,即:MORLC坐标样本和MORLC长度样本。文中给出了MORLC方法的理论分析和匹配判据,以及利用MORLC坐标和长度样本对运动物体的检测和识别数据。实验结果显示MORLC编码属于字符数据样本,数据量小,占用存储空间少,构造过程简单,运算处理效率高,构建样本的灵活性强,可根据不同的应用需求选择不同级次的样本形式,匹配和识别的鲁棒性好,不易产生错误匹配等。该方法适用于变视角运动物体的快速检测与识别,以及产品特征检测和识别。A new method of fast test and recognition of moving object was presented in the article, that is suitable for fast inspection and recognition of object with complex texture and moving state in unceasing changing view angle. A new coding method based on Multilevel Orientation Run Length Coding (MORLC) was developed, and the two new templet models with MORLC coordinate and length coding were developed in the research. Systemic theory process and criterion were given in the the article, and experiment data of object character inspection and recognition by MORLC coordinates and length coding and templets were given. The MORLC coding and templets shows benefits with small data anount, small storage addresses, easy constructing, high processing speed, flexible coding form with different multilevels resting with different application situation, high robusticity, low mistakes, etc. The method can be used for fast inspection and recognition of moving objects at moving state in unceasing changing view angle, especially character test and recognition of a great lot products.

关 键 词:特征检测 特征识别 图像编码 产品分类 

分 类 号:TP391[自动化与计算机技术—计算机应用技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象