最小二乘法的未知控制点检测

Detection of the invisible control points by least square method

背景:数字减影血管造影系统使用X射线影像增强器作为图像检测器,获取的投影图像与真实图像之间存在几何失真。目的:为实现脑血管的三维重建,对数字减影血管造影图像的几何失真进行校正。方法:基于数学形态学检测可视控制点,在此基础上采用最小二乘法曲线拟合进行非可视控制点检测,利用获取的672个控制点对原始图像进行非线性局部校正。结果与结论:实验结果表明在边缘校正的结果较好,圆弧区域基本不失真。使用最小二乘法曲线拟合得到的控制点坐标准确性高,图像校正结果好,可用于脑血管数字减影血管造影图像的校正,为后续的三维重建提供保证。

Abstract BACKGROUND： Digital subtraction angiography system uses the X-ray image intensifier as an image detector and the geometric distortion exists between the real image and the obtained image. OBJECTIVE：To correct the geometric distortion of digital subtraction angiography image in order to achieve three-dimensional reconstruction of cerebral blood vessels. METHODS： Based on mathematical morphology to detect visual control point, a method of least squares was used in this article to process the images in order to gain the coordinates of the invisible control points. The obtained 672 control points were used to do the nonlinear partial correction on the original image. RESULTS AND CONCLUSION： Experiment results showed that there were better results in the edge correction without distortion in arc region basically. The method of least squares curve achieved high accuracy control points and good image correction results, it could be used in distortion correction of digital subtraction angiography images and provided guarantee for a subsequent three-dimensional reconstruction.