为提高显微操作精度和效率,本文利用实物标记上的两个不同大小的圆,通过霍夫(Hough)圆算法配合人工识别辅助校正,自动建立实物坐标系。再结合图像识别技术、计算机控制的自动平台系统和自主开发的坐标转换算法,提出了一个可实现显微镜下CCD(电荷耦合元件)偏角的自动检测、微观影像的自动扫描和显微操作的再次(重复)定位的方法。实验结果显示,前后两次位置定位的误差可控制在实验所用CCD的取屏范围(640 pix×480 pix)的中间1/4(320 pix×240 pix)范围内,定位成功率可达90%。此方法只需将印有两个圆的透明标签贴于实物底部,安装配套的软件和硬件,即可实现显微镜下实物的精确重复定位。
In order to improve the precision and efficiency of micromanipulation, a physical coordinate system is estab-lished automatically by two different circles printed on the physical marker. The circles are recognized by Hough circle algorithm and artificial auxiliary. Then, by combining with the physical coordinate, automatic platform system and the coor-dinate conversion algorithm developed by ourselves, a method is proposed to automatically realize the detection of the CCD (charge coupled device) deflection angle, the scanning of microscopic image and the repeat positioning of micromanipulation under a microscope. As the experimental results show, the error of two successive positioning can be controlled in a range of 320 pix×240 pix in the center of CCD screen (640 pix×480 pix), that is a quarter of the whole screen, and the success rate can be up to 90%. The repeat positioning of the physical object will be realized precisely under the microscope, while it is only needed to affix the transparent label with two circles to the bottom of the material object, and install the relevant software and hardware.