[机器视觉]复现Cog3DAlignTool

发表于
复现Cognex的VisionPro9.1中的Cog3DAlignTool。

坐标空间树

3DAlignLego.cdb上附加的坐标系
数据类型：CogImage16Range(500*600)

2D/3D 浮动显示 映射点的函数
2D * 无
2D *\# #
2D *\#\@ @
2D *\#\@\Sensor2D @\Sensor2D
3D *\# #
3D *\#\@ @
3D *\#\@\Sensor3D @\Sensor3D

一般情况下，#与@的坐标值相同，Sensor2D与Sensor3D的x、y相同

注意：

坐标系 含义 是否在坐标空间树中 是否能坐标映射
* 显示pixel space，只能在绘图时使用 没有 没有
# Pixel space 没有 有
@ Root space 有 有

示例代码（C#）：

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
CogImage16Range InputImage = mToolBlock.Inputs[0].Value as CogImage16Range; 

// 坐标空间树
CogCoordinateSpaceTree Tree2D = InputImage.CoordinateSpaceTree;
Cog3DCoordinateSpaceTree Tree3D = InputImage.CoordinateSpaceTree3D;

// 2D坐标空间树的节点
string tree2D_root = Tree2D.RootName; //即@
CogStringCollection tree2D_childs_root = Tree2D.AllChildNames(tree2D_root, CogSpaceTreePathnameConstants.Full, false);

// 3D坐标空间树的节点
string tree3D_root = Tree3D.RootName; //即@
List<string> tree3D_childs_root = Tree3D.AllChildNames(tree3D_root, CogSpaceTreePathnameConstants.Full, false);

Cog3DVect2 point2D;
Cog3DVect3 point3D;
bool pixelUsed;
// 在坐标系空间树中映射坐标：2D点->3D点
point2D = new Cog3DVect2(0, 0);
point3D = new Cog3DVect3();
InputImage.MapPoint3DFrom2D(dst3DspaceName, src2DspaceName, point2D, out pixelUsed, out point3D);

// 在坐标系空间树中映射坐标：3D点->2D点
point2D = new Cog3DVect2();
point3D = new Cog3DVect3(0, 0, 0);
InputImage.MapPoint2DFrom3D(dst2DspaceName, src3DspaceName, point3D, out point2D);

// 验证#是否在坐标空间树中
string res_LookUpName = Tree2D.LookUpName("#", CogSpaceTreePathnameConstants.Full);
CogStringCollection res_FindNames = Tree2D.FindNames("#");

// 获取两个坐标系之间的转换关系
ICog3DTransform trans_3d_tree = Tree3D.GetTransform("Sensor3D", "@"); // ToSpaceName、FromSpaceName
ICog3DTransform trans_3d_img = InputImage.GetTransform3D("Sensor3D", "@");
ICogTransform2D trans_2d_tree = Tree2D.GetTransform("Sensor2D", "@");
ICogTransform2D trans_2d_img = InputImage.GetTransform("Sensor2D", "@");

另：导出点云数据的示例代码（C#）：

1
2
3
4
5
6
string dst3DspaceName = "Sensor3D";
string src2DspaceName = "#";
Cog3DVect2 point2D = new Cog3DVect2(0, 0);
Cog3DVect3 point3D = new Cog3DVect3();
bool pixelUsed;
InputImage.MapPoint3DFrom2D(dst3DspaceName, src2DspaceName, point2D, out pixelUsed, out point3D);

参考

参考内容 参考方面
Cognex VisionPro 9.1的示例作业：3D Align 主体思路