对网格中间区域矫正更优的图像矫正算法

常规放疗中模拟定位机影像增强器采集的图像常存在失真,为了保证放疗的精度,必须进行矫正,为此提出一种以点到直线距离为目标函数的图像矫正算法.在综合考虑矫正的速度、精度、平滑性的前提下,先采用基于一次线性基的移动最小二乘法对图像进行了矫正,取得了比较理想的结果;进一步,利用矫正后各线条应该为直线这一特征增加了参考点的数量,使交叉点中间区域也有了参考点.矫正结果显示,矫正之后各网格线呈现更理想的直线,交叉点之间的区域矫正比只考虑交叉点的矫正算法理想.当参考点增加到2 000多个点时,算法运行时间仍然较短.

The Optimized Image Correction Algorithm for the Middle Area of Grids

In conventional radiotherapy,the images obtained by the image intensifier of radiotherapy simulator are often distorted.In order to ensure the accuracy of radiotherapy,corrections must be conducted.Hence,an image correction algorithm is proposed with the objective function based on the distance of a point to a straight line.Given the correction speed,precision and smoothness,the moving least square method based on the primary linear base is adopted to correct images firstly and the results are satisfactory.Furthermore,on account of the characteristic that corrected grid lines should be straight,the number of reference points is increased so that the middle area of intersection points has reference points too.The corrected results present that after the correction,grid lines are more similar to straight line,and the correction effects of intersection areas are better than those of correction algorithm,which only take into account the intersection points.In addition,although more than 2 000 reference points are increased,the algorithm running time remains short.