Robotic and Computer Vision Projects with C++

---

This repository contains projects of computer vision tasks implemented with C++ libraries.

Kitti_Clouds_Viewer_and_Writer (leftCamRef): Visualize in 3D and write Kitti sequence point clouds in PCL format (.pcd). The output point clouds are in the left camera coordinate system. inputs: times_file_path: path of the directory where the times (.txt) file of the kitti sequences is saved clouds_sequence_path: path of the directory where the kitti point clouds (.bin) are saved max_frame: max mumber of frame in the sequence that we need visualize or write write_flag (optional): use only if you need save the point clouds into (.pcd) PCL format. how use: ./clouds_viewer_writer <times_file_path> <clouds_sequence_path> <max_frame> <write_flag> example: ./clouds_viewer_writer ~/data_odometry_gray/sequences/04/ ~/clouds/sequences/04/ 270 -w output: folder ~/build/sequence_clouds/ with the saved point cloud, only if you set write_flag with -w

Kitti_Clouds_Viewer_and_Writer (lidarRef): Visualize in 3D and write Kitti sequence point clouds in PCL format (.pcd). The output point clouds are in the Velodyne lidar coordinate system. inputs: times_file_path: path of the directory where the times (.txt) file of the kitti sequences is saved clouds_sequence_path: path of the directory where the kitti point clouds (.bin) are saved max_frame: max mumber of frame in the sequence that we need visualize or write write_flag (optional): use only if you need save the point clouds into (.pcd) PCL format. how use: ./clouds_viewer_writer <times_file_path> <clouds_sequence_path> <max_frame> <write_flag> example: ./clouds_viewer_writer ~/data_odometry_gray/sequences/04/ ~/clouds/sequences/04/ 270 -w output: folder ~/build/sequence_clouds/ with the saved point cloud, only if you set write_flag with -w

Kitti_Convert_StereoImages_to_PointClouds: Convert kitti sequence gray images into PCL point clouds (.pcd). inputs: sequence_path: path of the directory where the kitti images are saved num_frames: number of frames in the sequence that we need convert write_flag (optional): use only if you need save the point clouds into PCL format. how use: ./stereo2cloud <sequence_path> <num_frames> <write_flag (optional)> example: ./stereo2cloud ~/data_odometry_gray/sequences/04/ 270 -w output: folder ~/build/sequence_clouds/ with the saved point cloud, only if you set write_flag with -w

Kitti_Odometry_Viewer: Visualization 3D of the Kitti cameras odometry inputs: kitti_poses_file: poses file (.txt) of the kitti sequence how use: ./kitti_odometry_viewer <kitti_poses_file> example: ./kitti_odometry_viewer ~/data_odometry_poses/poses/04.txt output: file ~/build/poses_quaternion.txt with the kitti sequence poses (tx ty tz qw qx qy qz)

Kitti_Save_Scales: Compute the scales for a Kitti sequence. The scale is computed using the ground truth poses. inputs: kitti_poses_file: poses file (.txt) of the kitti sequence to compute the scales out_file_name: name of the output file (.txt) where the code save the kitti sequence scales how use: ./kitti_scales <kitti_poses_file> <out_file_name> example: ./kitti_scales ~/data_odometry_poses/poses/04.txt ~/Desktop/scales_kitti_seq_04.txt output: (.txt) file with the saved scales

Kitti_Sttiching: Demonstrates how to use the Stitching module of OpenCV to create a panorama using left and right images of the Kitti sequence. inputs: sequence_path: path of the directory where the kitti images are saved how use: ./panorama <sequence_path> example: ./panorama ~/data_odometry_gray/sequences/04/ output: file ~/build/panorama.png with the result of the sttiching beeteen left and right images.

Plot_3D_Arrow: Test PCL visualizer plotting 3D arrows. how use: ./plot_3D

Real_Time_Chessboard_Corners_Detection: Detect and plot the corners of a chessboard in real-time using OpenCV and a webcam. how use: ./livecorners

Real_Time_Chessboard_Corners_Detection_SVO: Detect and plot the corners of a chessboard in (.svo*) video using OpenCV. how use: ./zedcorners

Record_SVO_from_ZED: Record an (.svo) video using ZED camera. inputs: cam_resolution: ZED camera resolutions VGA or HD720 or HD1080 or HD2K output_dir: directory to save the (.svo) video and the camera intrinsic parameters how use:./record_svo_zed <cam_resolution> <output_dir> example: ./record_svo_zed HD720 ~/Desktop output: a folder with the (.svo) video and a (.txt) file with the camera intrinsic parameters

Record_SVO_ZED+P3AT_Automatic: Record a video using a ZED camera mounted on a robot Pioneer 3AT. The robot moves into a straight line for some distance. When the robot starts, use the arrow keys to drive it and the spacebarto stop. Assume you are ready to record a video, then press keyato switch to automatic mode. PressEsc` to exit. inputs: cam_resolution: ZED camera resolutions VGA or HD720 or HD1080 or HD2K output_dir: directory to save the (.svo) video distance_in_mm: distance that the robot moves fordward in straight line how use: ./record_svo_automatic <cam_resolution> <output_dir> <distance_in_mm> -rp /dev/ttyUSB0 example: ./record_svo_automatic HD720 ~/Desktop/ 17000 -rp /dev/ttyUSB0 output: a folder with the (.svo) video.

Record_SVO_ZED+P3AT_Manual: Record a video using a ZED camera mounted on a robot Pioneer 3AT. You can drive the robot with the keyboard (arrows). Then when you are ready to record, press s and press q to exit when you finish. inputs: cam_resolution: ZED camera resolutions VGA or HD720 or HD1080 or HD2K output_dir: directory to save the (.svo) video and camera parameters file. how use: ./record_svo_manual <cam_resolution> <output_dir> -rp /dev/ttyUSB0 example: ./record_svo_manual HD720 ~/Desktop/ -rp /dev/ttyUSB0 output: a folder with the (.svo) video and camera parameters file .yml

Registration_3D_using_Features: Demonstrates how use the registration PCL module. This code computes the relative transform between two point clouds. inputs: source_cloud : (.pcd) PCL cloud target_cloud : (.pcd) PCL cloud how use:./registration3D <source_cloud> <target_cloud>

Registration_Point_Clouds_using_SVD: Compute the transformation between two point clouds using the SVD algorithm. Use this transformation to register the two-point clouds. Finally, compute the RMSE between the two-point clouds. how use:./svd

Save_ZED_All_Data: Capture left, right images, depth, and a point cloud from the ZED. how use:./save_all <VGA or HD720 or HD1080 or HD2K> outputs: /left/: folder with (.png) images /right/: folder with (.png) images /depth/: folder with (.png) depth map (in milieters) /cloud/: folder with (.pcd) clouds /poses.txt: camera poses referenced to the world coordinate system in Kitti format /confidence.txt: depth map confidence /times.txt: timestamps in milliseconds /calibration.txt: intrinsics parameters of the camera

Save_ZED_Data_Manual(keyboard): Capture left, right images, depth, and a point cloud from the ZED using the keyboard. When you press s or S the code saves data only of the current frame. how use:./save_all <VGA or HD720 or HD1080 or HD2K> outputs: /left/: folder with (.png) images /right/: folder with (.png) images /depth/: folder with (.png) depth map (in milieters) /cloud/: folder with (.pcd) clouds /poses.txt: camera poses referenced to the world coordinate system in Kitti format /confidence.txt: depth map confidence /times.txt: timestamps in milliseconds /calibration.txt: intrinsics parameters of the camera

Save_ZED_Processing_Times(Images_Depth_Clouds): Carried out and analyzed of processing times of each capture step in the ZED camera, i.e., the time that takes in capture an image, a depth map, and a point cloud. inputs: zed_resolution : VGA or HD720 or HD1080 or HD2K frames_per_second : Remember that VGA mode have 15, 30, 60 and 100 fps. HD720 mode have 15, 30 and 60 fps. HD1080 mode have 15 and 30 fps and HD2K mode have only 15 fps. how use: ./save_processing_times <zed_resolution> <frames_per_second> outputs: depth_var_xx: (.png) image that reflects the variance in each pixel of the all taked images depth_mean_xx: (.png) image that reflects the mean in each pixel of all taked images for the analisys depth_var_xx: (.txt) file with the variance in each pixel of all taked images for the analisys depth_mean_xx: (.txt) file with the mean in each pixel of all taked images for the analisys

Save_ZED_Undistorted_and_Distorted_Images: inputs: zed_resolution : VGA or HD720 or HD1080 or HD2K how use: ./save_zed <zed_resolution> outputs: /left/: folder with (.png) left rectifiqued (undistorted) images /right/: folder with (.png) right rectifiqued (undistorted) images /left_unrec/: folder with (.png) left unrectifiqued (distorted images /right_unrec/: folder with (.png) right unrectifiqued (distorted) images /calibration.txt: (.txt) file with intrinsic parameters of left and right cameras

Save_ZED(SVO)_Images_and_Poses_in_Kitti_Format: Save left and right images form (.svo) video. Also, save camera poses in Kitti format. inputs: svo_video : (.svo) file path how use: ./save_zed_data <svo_video> example: ./save_zed_data ../Data/test.svo outputs: ../Data/left/: folder with saved left rgb images ../Data/right/: folder with saved right rgb images ../Data/poses_world.txt: file with ZED poses referenced to world coordinate system ../Data/poses_camera.txt: file with ZED poses referenced to camera coordinate system

Save_ZED(SVO)_Poses_in_TUM_Format: Save ZED camera poses form (.svo) video. The poses are in TUM format. inputs: svo_video : (.svo) file path how use:./save_zed_data <svo_video> example: ./save_zed_data ../Data/test.svo outputs: ../Data/poses_zed.txt: file with ZED poses referenced to camera coordinate system

Test_Kv1_using_OpenNI: Test the Microsoft Kinect version 1 using OpenNI. how use:./test_kv1

Test_Kv2_using_Libfreenect2: Test the Microsoft Kinect version 2 using Libfreenect2. how use:./test_kv1

Test_Kv2_using_OpenNI: Test the Microsoft Kinect version 2 using OpenNI. how use:./test_kv1

Visual_Monocular_SLAM_using_UcoSlam: Compute visual monocular SLAM using UcoSlam (MODE_SLAM). inputs: video : (.avi) video recorded with a monocular camera camera_parameters : .yml file with the intrinsics parameters of the camera vocabulary : .fbow vocabulary for the ORB keypoint descriptor output_map : .map result map with all data how use: ./ucoslam_monocular <video> <camera_parameters> <vocabulary> <output_map> example: ./ucoslam_monocular ../Data/test.avi ../Data/mono_params.yml ../Data/orb.fbow ../Data/out.map outputs: ../Data/poses_Kitti.txt: file with camera poses referenced to camera coordinate system in Kitti format.

Visual_Stereo_SLAM_using_UcoSlam: Compute visual stereo SLAM using UcoSlam (MODE_SLAM) in a Kitti sequence. MODE_SLAM is used to create and update maps. In that mode, the MapManager thread is activated to decide when keyframes are added, etc. inputs: video_left : .mp4 video recorded with left camera video_right : .mp4 video recorded with right camera stereo_calibration_file : .yml file with the intrinsics and extrinsics parameters of the left and right cameras kitti_times_file : (.txt) kitti times file of the sequence output_poses : (.txt) file with the saved camera poses in TUM format voc_path : .fbow vocabulary for the ORB keypoint descriptor ucoslam_params : .yml parameters to setup UcoSlam how use:./slam_stereo <video_left> <video_right> <stereo_calibration_file> <kitti_times_file> <output_poses> <voc_path> <ucoslam_params> example: ./slam_stereo ../Data/cam0.mp4 ../Data/cam1.mp4 ../Data/stereo.yml ../Data/times.txt ../Data/poses_TUM.txt ../Data/orb.fbow ../Data/myparams.yml outputs: ../Data/poses_TUM.txt: file with camera poses referenced to camera coordinate system

Installation:

• Dependences (mandatory):

  • ZED SDK 3.0
  • CUDA 10.0
  • OpenCV 3.4.1
  • Aruco 3.0.12
  • PCL 1.8
  • ARIA or ARIACODA
  • UcoSLAM
  • Boost
  • Freenect2
  • Libviso2
  • Open-NI
  • Libusb 1.0

• Download any project and open a terminal ctrl+t:

  $ cd path 
  $ mkdir build & cd build 
  $ cmake .. 
  $ make
  

NOTE:

If you find any of these codes helpful, please share my GitHub and STAR ⭐ this repository to help other enthusiasts to find these tools. Remember, the knowledge must be shared. Otherwise, it is useless and lost in time.