基于模板匹配的改进型目标识别算法  

Modified target recognition algorithm based on template matching

作　　者：丁筱玲 赵强[1,4] 李贻斌 马昕[1] DING Xiaoling;ZHAO Qiang;LI Yibin;MA Xin(School of Control Science and Engineering, Shandong University, Jinan 250061, Shandong, China;Mechanical ＆ Electronic Engineering College, Shandong Agricultural University, Tai＇an 271018, Shandong, China;Shandong Provincial Key Laboratory of Horticultural Machineries and Equipments, Tai＇an 271018, Shandong, China;Shandong Power Equipment Company, LTD, Jinan 250022, Shandong, China)

机构地区：[1]山东大学控制科学与工程学院,山东济南250061 [2]山东农业大学机械与电子工程学院,山东泰安271018 [3]山东省园艺机械与装备重点实验室,山东泰安271018 [4]山东电力设备有限公司,山东济南250022

出　　处：《山东大学学报（工学版）》2018年第2期1-7,共7页Journal of Shandong University（Engineering Science）

基　　金：山东省自然基金资助项目(ZR2012CM040);山东省重点研发资助项目(2015GGB01311);山东农业大学现代农业智能化装备研发资助项目(SA24137)

摘　　要：针对图像处理过程中采用局部特征提取与特征匹配的目标识别算法对纹理不丰富物体识别精度差、在同一次学习过程中不能多视角识别同一个物体的缺点,提出采用基于模板匹配的改进型目标识别算法,提高对纹理不丰富物体的识别速度及准确率。利用梯度作为特征量完成模板匹配,结合DOT算法去除次要的梯度特征,只采用幅值大的主导梯度方向作为特征量进行模板匹配,融入仿射投影变换算法、将模板特征二进制化来提高在线同时识别多个不同物体、多视角识别同一个三维物体的速度及准确率。试验证明,该目标识别算法对复杂背景中纹理较少的物体,发生微小变形、微小平移和光照变换的物体识别效果鲁棒性强。Because the local feature extraction algorithm got lowrecognition rate for poor texture object and it could not recognize the same object from different perspective. A modified target recognition algorithm was proposed. The algorithm used gradient characteristics as features to complete the template matching,which only used the main direction of the gradient and the minor gradient feature was removed by DOT（ dominant orientation templates） algorithm. The template feature,which was made binary by fusing affine projection transformation algorithm,could improve the recognition rate of identifying multiple objects at the same time or the same object from different perspective. The experiments demonstrated that the proposed algorithm could get better recognition rate and was robust for the object with poor texture,small deformation,small translation and light transformation.

关 键 词：目标识别算法 模板匹配 主导梯度方向 仿射投影变换 模板特征二进制化 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]