本文提出了一种以(Ф,θ,ρ,α)四参数表示的空间直线为基础的三维Hough变换,用来检测三维超声图像中的针状物体。我们用实验室研发的三维超声系统采集插入水中的钢针的三维超声图像,然后用本文提出的方法对其进行检测。实验结果表明,本文提出的方法能比较准确地检测出三维超声图像中的针状物体,方向误差在1°以内,位置误差在1mm以内。
Based on the representation (Ф,θ,ρ,α) of straight lines in the 3D space, we propose a 3D Hough transformarion algorithm to segment needles in 3D ultrasound (US). The needle segmentation techniques are evaluated using the 3D US images acquired by the equipment developed in our laboratory. The experimental results show that the segmentation techniques in this paper can accurately detect the needle in the 3D US with the orientation deviation of less than 1° and the position deviation of less than 1 mm.