基于新投影模型的利用图像均匀性的RayCasting加速方法
Improved RayCasting Method Based on Image Homogeneity
-
摘要: 计算机仿真投影图像(digitally reconstructed radiograph,DRR)在计算机辅助医学领域有着广泛的应用,特别是二维三维图像配准对仿真投影图像生成的效率有着苛刻的要求.提出了一种新的投影模型,该模型可有效地分离投影图像中物体的尺度变换和投影失真的耦合度,提高仿真投影图像尺度的可控性.同时提出基于图像内容均匀性的RayCasting改进算法,加速针对光线的积分过程.实验表明:针对均匀性较好的数据,改进算法的效率较传统RayCasting算法有一定提高,同时改进算法生成的仿真投影图像的边缘具有较高的精确度.Abstract: Digitally reconstructed radiograph(DRR) has been widely used in computer assisted diagnostics and 2D-3D medical image registration applications.This paper designs a new projection model that can release the interconnection of image scale and projection distortion.Therefore,this model can create projection image with a higher accuracy in image scale.Based on this model,an improved RayCasting method,which employs the image region homogeneity to accelerate the integration process along the X-ray beam,is proposed.According to the experimental result,this method is faster than the traditional RayCasting method when it comes to the data with high homogeneity.This method can preserve the accuracy of the object edges.