基于模板识别的高精度自动定位系统

为满足机器人辅助 医疗手术中高精度自动识别物体的需要,利用Bumblebee 2 摄像头拍摄图像,把黑白交互的椭圆和圆作为一个模板,用改进的角点提取方法提取出图像中的角点;在对图像进行增强之后,利用模板黑白格的几何对称性特征提取模板中的角点,然后区分出椭圆中的角点和圆中的角点,最后利用向量垂直关系提取出椭圆中的目标角点,识别出模板.通过Bumblebee 2自带的深度函数,提取模板的深度,得到模板在摄像头空间中的坐标,达到高精度自动定位物体的目的,克服传统摄像机需要标定内外参数的步骤.实验表明,定位误差在0.42 mm之内,基本上可以满足手术的要求.

High-accuracy automatic locating system based on templates

In order to meet the requirement of robot assisted surgery which identified the object automatically with high precision,the Bumblebee 2 camera systems were used,and the ellipse and the circle were regarded as a template.The corners in the image were detected by the proposed algorithm.After the image enhanced for improving the contrast,using the symmetric properties of the black block and white block in the templates,the corners in the ellipse and in the circle were detected.At last using the vector vertical relationship,the goal vertex in the ellipse was extracted,and then the template was identified.Through the depth function of the Bumblebee 2,the world position of the templates was obtained automatically with the high accuracy.The system overcame the problem that the traditional cameras calibrated the inside and outside parameters.Finally,experiments show that the positioning accuracy will reach to 0.42 mm,which is satisfactory to the requirement of the surgery.