Point-Cloud-Renconstruction-and-Camera-Calibration

Implementation of using structured light for 3D depth reconstruction, based on the techniques discussed in:

High-Accuracy Stereo Depth Maps Using Structured Light
Projector calibration for 3D scanning using virtual target images

The project folder can be divided into two parts: calibration and reconstruction.

Each folder contains necessary functions in order to meet project's specification.

Reconstruction

NOTE: Before doing reconstruction, please make sure you have obtained your camera and projector's extrinsic and intrinsic matrices.

Running reconstruction is simple, just calling function reconstruction(path,prefix,first,digits,suffix,outputName,use_calibration).

If I'm calling reconstruction('head_own','_H8A0',065,3,'JPG','head_own',0);

The program will do reconstruction from images in 'head_own' folder. And output a .ply file with name 'head_own'.

Details of first five parameters can be found in load_sequence_color.m

The last parameter must be either 0 or 1.

If input is 1 then the program will use your calibrated matrices.

All image sets must be placed at the same directory as the function is located. For using calibrated matrices place four matrices in this directory as well but must be named as: intrinsic_matrix_cam.mat intrinsic_matrix_proj.mat extrinsic_matrix_cam.mat extrinsic_matrix_proj.mat

Calibration

Performing calibration can be done by calling function calibration(isSynthetic,isReal,isOur).

Each of these paramenters are flags telling program what data sets to use. All data sets must be placed at the same directory as the function is located.

After running above function, make sure to use calib_gui from TOOLBOX_calib to finish the calibration process. This tool box is introduced by lab2/caltech.

Output

For reconstruction, there are several subfolders named by the data set. Each subfolder I put the u v and depth outcome in it.

For calibration, the subfolders are named by the calibration data sets. Each subfolder contains homographed images and the computed extrinsic/ intrinsic matrices for camera and projector.

Feel free to leave your comments on issue section

Please refer to file report.pdf for further project details

pleaseredo / point-cloud-renconstruction-and-camera-calibration Goto Github PK