HELP--How to read RANSAC output?

[Santhi]

Hi Team,

I am using RANSAC plugin to identify the cylinders through the command line, it is working fine and the output is in the form of .bin
I want to read that binary data to understand the cylinder height , radius and also the base point of the cylinder
Can any one help me what is the bin file structure and how to read it from python?

If we can convert the .bin to ASC it is easy for me to read calculate above parameter.
I have tried below command but unfortunate I am not able to find the output file in the folder. Below is the command I have tried

Code: Select all

 CloudCompare -O "E:\Cloud_CYLINDER_0060_pair_2020-08-27_10h54_54_842.bin" -C_EXPORT_FMT ASC

If I get each file information like below, then I can not read all the binaries (if i get 100 i can eliminate some of the cylinders)

Code: Select all

Name;Points;meanX;meanY;meanZ;
Cylinder (r=387.517517/h=41.628941);46097;17.9512;42.1412;59.8026;
vertices;50;342.193;254.535;63.4394;

Thanks in advance,
Santhi

[daniel]

The BIN format is CC's own binary format which is a serialization of the memory (whit all possible entities managed by CC, the DB hierarchy, etc.). It's quite complex to read and changes often as the software is improved.

And sadly there might not be another format that can store the result in a readable format...

[Santhi]

Thank you Daniel for the Quick reply.
I have used parameters as shown below

Code: Select all

 
 min support points for primitive = 500
 e = 0.04 
 b = 0.07
 a = 25.00 deg
 overlooking probability = 0.01
 

and the command line parameters I have set like below.

Code: Select all

cloudcompare -O "PATH-TO-LAS"  -RANSAC ENABLE_PRIMITIVE CYLINDER EPSILON_ABSOLUTE 0.04 BITMAP_EPSILON_ABSOLUTE 0.07 SUPPORT_POINTS 500 MAX_NORMAL_DEV 25 PROBABILITY 0.01  OUT_PAIR_DIR "E:\\"

I am trying to get only the cylinders of radius 0.03 to 0.05 and the height is nearly 15mts, could you please suggest me what could be the above values to get my required features.
I am getting all the points clouds are classified like buildings cars & trees etc as cylinders. How to avoid those ? any suggestions please?

[daniel]

Sadly I don't think this is how this plugin works... I will really find all the cylinders it can. Then you have to clean the result...

[WargodHernandez]

Actually in the Ransac new features branch I made it allows you to specify limits for primitives your searching for. It's not finished but I am planning on finishing the work when I can. It definitely already can limit to min max radii for cylinders, I don't remember if I limit by length or not yet though

[WargodHernandez]

Actually the current version has just max radii and max length adjustments, I could add minimum parameters also but I've found that more issues crop up from the excessive sized primitives (Sphere or cylinders can be used in place of planes if you have a large enough primitive)

[Santhi]

Thank you Daniel, I am doing the same :)
And I know I am asking some silly and foolish questions sorry for that.

Hi WargodHernandez,

Thank you for the reply.

I have observed another difference in the output generated by cmd and GUI(tool)
Is there any reason behind that? The count of cylinders are not same and the radius and height is also not matching :(
Your reply is more helpful to me

Thanks once again,
Santhi

[daniel]

Isn't it because the RANSAC detection process is RANdom?

[Santhi]

Thank you for the reply. this helps me alot

[ros_gala]

Hi,
I am using Ransac shape detection plugin in the latest CC version (2.12.alpha - 11/27/2020). Is there a way to maintain the RGB values of each point in the output detected shapes clouds?
Thank you in advance