摘要
为解决基于图像导航的机器人辅助外科手术过程中X光图像的畸变特性问题,提出了一种新型的基于BP神经网络的X光畸变图像校正方法。该方法首先从一个标准模板的X光畸变图像中提取标定样本位置信息作为神经网络输入,以模板的标定样本真实位置信息为神经网络输出,构建BP神经网络。该BP神经网络能够实现畸变图像与真实模板之间的映射关系,从而达到图像畸变校正的目的。最后通过机器人辅助髓内钉锁孔实验对该方法进行了实验验证,证明了该方法的有效性。
To solve the distortion problem of the X-ray image in the process of robot-assisted surgery, this paper proposed a new method of X-ray image distortion correction based on BP neural network. In this method, the BP neural network was built through using the position information of the calibration samples extracted from the distorted X-ray image in a standard template as the neural network input, and using the real position information of the calibration samples in the standard template as the neural network output. The mapping between the distorted X-ray image and the standard template could be achieved using this BP neural network, the distortion problem could be solved. The method was tested by the experiment of locking the distal screw of the medullary nail with the help of the robot-assisted surgery system, and it was proved to be effective.
出处
《计算机应用研究》
CSCD
北大核心
2008年第10期3050-3052,共3页
Application Research of Computers
基金
长江学者创新团队计划资助项目(IRT0423)
国家自然科学基金资助项目(60505016
60675037)
关键词
反馈神经网络
X光图像
畸变校正
BP neural network
X-ray image
distortion correction
