Detect 2D feature points and generate corresponding 3D corresponding points in ground truth object
Detect 2D feature points and generate 3D corresponding points using ground truth object
This is a simple Matlab code to generate 2D - 3D corresponding points of pair. This work was inspired by the paper 2017 CSVT An integrated platform for live 3D human reconstruction and motion capturing. You can start right away by using a built-in sample image in the code. For the full implementation, this job requires the ground truth box (You should have it in the real-world) and of course, a camera.
This code was built in MATLAB R2020b. Also, you need to install the MatLab add-on “Computer Vision Toolbox” for SURF feature detection.
First, open Run.m and start running. You will see in the console,
$ Type filename = <from_current_folder_path> :
Type the image path you want to use. For a quick start, we will use a sample image in the folder.
$ Type filename = <from_current_folder_path> : images/Camera.jpg
Next, just type <enter> for the double size question. This improves accuracy.
$ Do you want to Double size? Y/N [Y]:
Now, here comes the image, you should crop AR markers by dragging your mouse.
This will generate SURF feature points inside the region you cropped.
And it will automatically find the best matching AR marker from the built-in dictionary. Just wait for the result.
All you need to do is to decide whether you want to save the detected feature points in the output data.
Type Y to save the detected feature points. Be careful! N (No) is the default.
$ Do you want to Save? Y/N [N]: Y
And The Console will ask if you want to continue. If you get all of your feature points detected, Type N to finish the process.
If you want to keep detecting feature points, type <enter> and re-do your crop & save the process.
$ Do you want to continue? Y/N [Y]:
Finally, you can see the console log like this.
You can see 2D&3DPoints.mat is saved in the directory.
This data includes the 2D feature points and the corresponding 3D points of Ground Truth Object.
2D <-> 3D Correspondence is saved Successfully!
To calibrate your own camera, you should have a real-world Ground Truth Object. These Objects were built with 4 boxes. The for the detailed manual, go to 2017 CSVT An integrated platform for live 3D human reconstruction and motion capturing. In this code, the box size can be customized. Change line 46 of Run.m. Beware the scale is millimeter.