模板匹配的最小包容平面内直线度评定法

Straightness rating method of template matching base on minimum contain plane

基于传统的旋转法直线度评定方法提出了模板匹配评定法。首先,针对视觉的直线度评定架构了相应的硬件方案,并定义和分析了像素当量、放大倍数和视场3个关键参数。其次,借助音乐学中的拍子强弱分析了"待评定直线"的走势分为2拍子、3拍子、4拍子,实际存在共6种走势;再次,分析了"像素卡尺"的图像模板匹配方法,通过"像素卡尺"的模板匹配和仿射变换等算法逼近"待评定直线";最后,推导了"像素卡尺"模板匹配的算法和直线度误差。仿真表明该方法有一定的应用价值。

Straightness assessment by tradition - based rotation proposes the rating method of template matching. First, corresponding hardware solutions are constructed focused on the visual straightness visual assess- ment, and three key parameters including the pixel equivalent, magnification times and field of view are defined and analyzed. Secondly, relying on the beat strength in music science, the ＂assessment linear＂ is evaluated and divided into six kinds of trends ： 2 beats, 3 beats, 4 beats. Again, the ＂pixel caliper＂ image template matching method is analyzed and the ＂assessment linear＂ is approached by ＂pixel cali- per＂ template matching and affine transformation algorithms. Eventually the ＂pixel caliper＂ template matching algorithm and straightness errors are deduced. Simulation results show this method has some application value.