将 Java 中的 3D 仿射变换构建导入 R

Question

我有一个与仿射变换（<affine>标签之间）相关的问题。我正在使用AffineTransform3D函数从使用 Java 创建的 BigDataViewer Fiji 插件创建的 xml 文件（此处为完整 xml 文件）中提取以下两个仿射转换：<ViewRegistrations><ViewRegistration timepoint="0" setup="0">  <ViewTransform type="affine">    <Name>Fast 3d geometric hashing (rotation invariant), AffineModel3D on [beads (c=0)]</Name>    <affine>1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 0.0 1.0 0.0</affine>  </ViewTransform>  <ViewTransform type="affine">    <Name>calibration</Name>    <affine>1.9998662296836334 0.0 0.0 0.0 0.0 1.9998662296836334 0.0 0.0 0.0 0.0 1.9998662296836334 0.0</affine>  </ViewTransform></ViewRegistration>我想在 R 中导入两个仿射变换，使用buildAffine()R 包 {RNiftyReg} 中的函数，然后使用composeTransforms(){RNiftyReg}计算它们的组成。buildAffine(translation = c(0, 0, 0), scales = c(1, 1, 1), skews = c(0, 0,0),  angles = c(0, 0, 0), source = NULL, target = NULL,  anchor = c("none", "origin", "centre", "center"))我的问题：上面的仿射变换存储在一个包含 12 个索引的向量中。buildAffine()需要将平移、比例、倾斜和角度的值作为输入参数。我想知道哪个值对应什么。

Answer 1

我主要是 R 用户，但这里是： hte Java 调用中的变量名称是：

 dat <- scan (text="double xx, double yx, double zx, double tx, double xy, double yy, double zy, double ty, double xz, double yz, double zz, double tz", what="")

dat <- dat[dat != "double"]

matrix(dat,4)

      [,1]  [,2]  [,3] 

[1,] "xx," "xy," "xz,"

[2,] "yx," "yy," "yz,"

[3,] "zx," "zy," "zz,"

[4,] "tx," "ty," "tz" 

此页面记录了如何对仿射变换进行编码，唯一的区别是转置了 4x3 矩阵：https : //o7planning.org/en/11157/javafx-transformations-tutorial

 t( matrix(dat,4) )

     [,1]  [,2]  [,3]  [,4] 

[1,] "xx," "yx," "zx," "tx,"

[2,] "xy," "yy," "zy," "ty,"

[3,] "xz," "yz," "zz," "tz" 

所以矩阵将应用于 c(x,y,z,1) 的向量以获得输出：