基于角点特征的目标识别方法  

Object recognition method based on corner feature

作　　者：曹健[1] 高春晓[1] 刘琼昕[1] 贺跃[1] 

机构地区：[1]北京理工大学计算机学院,北京100081

出　　处：《华中科技大学学报（自然科学版）》2010年第6期84-87,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：'十一五'国防预研项目

摘　　要：针对基于角点特征的目标识别存在的不足,提出了一种角点特征的构造方法,这种特征具有平移不变性、旋转不变性、尺度不变性以及对噪声的抗干扰能力.利用角点间的全局约束和局部约束得到类型可分离程度较高的模式向量,并根据目标角点的空间关系对模式向量进行适度的维数约简;根据目标的三维模型建立二维视面模型,从而提取出目标在不同姿态下的特征,解决目标姿态变化造成的难以识别的问题.结合反向传播网络的分类能力,将其应用到视点变化的目标识别领域.与其他三种形状特征进行实验对比,结果证明该方法在视点发生变化时对目标的识别更为有效.In allusion to the deficiencies of object recognition based on corners, a method for extracting corner feature from images was proposed. This feature is invariant to translation, rotation, scale change and is shown robust to addition of noise. The pattern vectors were obtained using global and local constraint conditions and the dimension of vectors in the method according to the spatial relationships between corners. To solve the problems caused by the attitude changes of the objects, the 3D modes were used to construct the 2D views. We presented a system to recognize the objects with changes in 3D viewpoint using this feature and BP network. The performance on the obtained experimental results demonstrates that the proposed method is more effective than the other three ones.

关 键 词：图像处理 目标识别 角点 特征提取 反向传播网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]