《一种滤波反投影算法的加速方式》由会员分享，可在线阅读，更多相关《一种滤波反投影算法的加速方式（4页珍藏版）》请在金锄头文库上搜索。
1、 32 null 2 d 0 / Vol. 32 null No. 22010M 2 Systems Engineering and Electronics February 2010 c I | : 1001null506X(2010) 02null0401null04l : 2008null12null14; : 2009null10null26b :S E 1 S ( 60532080) ; 0 k / S S / L i ( 9140C1204040908) ; 8 1 S ( 2007012003) 5T e :D (1977null) ,3 , q ,p V , 1 Z _ | a
2、 y a) M Y bEnullmail: zhiweiqiaook nuc.edu. cnB r o Q g E F Z T乔志伟1, 2, 韩null 焱1, 2, 魏学业1( 1. 北京交通大学电子信息工程学院, 北京 100044;2. 中北大学电子测试技术国防科技重点实验室, 山西太原 030051)null null K null 1 : 提出了滤波反投影( filtered back project ion, FBP)算法的基于被动式射线驱动和预插值的反投影策略b首先对滤波后的投影采用线性插值预插值到更加精细的间隔上,然后对每个角度下的滤波投影用被动式射线驱动法反投影,得到重建图
3、像b反投影时为了减少求取投影地址时三角函数的运算,将三角函数值事先求得并存到查找表中,需要时调用b反投影时采用简单的最邻近插值法插值,以节省插值时间b仿真实验表明,这种方法可以有效提高反投影速度b1 o M : 信息处理技术; 滤波反投影; 加速; 图像重建; 射线驱动 m s | : T N 917null null null null D S : AAcceleration method of the filtered back projection algorithmQIAO Zhinullw ei1, 2, H AN Yan1, 2 , WEI Xuenullye1(1. School
4、of Electronics and information engineering, Beijing Jiaotong Univ., Beijing100044, China;2. Key Lab. of national defense technology onelectronic measurement, Taiyuan030051, China)null null Abstract: A back projection method of the filtered back projecion algorithm is proposed based on pass iveraynul
5、ldriven technique and prenullint erpolation technique. Firstly, the f iltered project ions are interpolated to moredense point s by linear interpolation. T hen, every anglenull s project ion is back project ed using passive raynulldrivent echnique and t he reconst ruct ed image is made. For decreasi
6、ng t he calculation amount of angle functions forobtaining projection address, the trigonometric functionsnull values are calculated in advance and stored into a t ablewhich can be indexed if needed. The s imple neighbor interpolat ion method is used during backprojecion so as toreduce time for int
7、erpolation. T he simulat ion experiments illustrate that t he method eff ectively improve t he backproject ion speed.Keywords: information processing technology; f iltered back projection; acceleration; image reconstrucnullt ion; raynulldriven0null null null null 9 / ( computed t omography, CT ) -K
8、z _ / B b B , Y , FBP E Q g F V bFBP E = M $ ? Z b M $ V w r o Q g E Tf(x,y) = !null0g(t,null)dnull (1)T g(t,null) = p(t,null)* h(t) (2) null null null null null null null t= xcos null+ ysin nullh(t) = !+ - | f | ej2nullft df (3)p(t,null) / g ;g(t,null) $ Y h(t) r o r o r o g bC L ! V 180# S = W M /
9、 g , N , y m NN,5 Q g / :f = zeros(N,N); S m 0for i= 1%Nfor j= 1%Nfor k= 1%Mt= xcos K - 1M null + ysin K - 1M nullY V p g t,K - 1M nullf(i,j)= f (i,j)+ g t,K- 1M nullendendend ,x y y m W #m U S i j bs V ,Q g V K n H K= Q g t p | bQ g p | 1 2MN 2 Q 2MN2Q E # MN2 Q b! 180 g , y m v l 256 256,5 1 23 59
10、2 960Q , E 23 592 960Q , 11 796 480Q b - PC X B , y N A B b2null Z E F # s V 3 Z 4 Q g :( 1) Y V L 7 Q g H e # Z E h ;( 2) Y V $ T L z Z T 4 r q ;( 3) Y V P s V , h f b2. 1null + e H W s - ,1 FBP E Q g V / X 1 , v V L / # Z # H z Z E bL V B X | L , L 9 W b! C X (a,f(a) (b,f (b), W B c f (c) = f (a) + f (b) - f (a)b- a (c- a) (4)null null . d Z E , 1 MN2 Q E , 4MN 2 QF E , 9 M V 4 b T Q g - ,| g L Z E W ,Q g H e # E ,V F b!
