/isoCvDemo

feature point tracking

Primary LanguageC++

isoCvDemo

  • Feature point tracking

Demo

Environment

  • Use Pod to install Masonry and OpenCV (default version).

How to use

  • Currently it only supports slow movement of camera or shoes.
  • Also error will be accumulated.

Main Interface

  • MainMenuViewController.m
  • The last Optical Flow Demo is current version.

Tracking Interface

  • ViewController.mm
  • Image resolution is 640x480
  • The brightness of image is fixed.
    • Clicking reset buttom can adjust image to current light.

Initialization

  • Put shoes into the retangle and click start button to start tracking.
  • Click reset buttom to back restart initialization.

Camera Pose Matrix (4x4)

  • cvOpticalFlow.cpp
  • function trackingFrontFromInitialNineGrid
  • avgHomo is the homography matrix of shoes plane in two frame (initial frame and current frame)
  • camera matrix is required to recover pose in function pose_estimation_2d2d
    • current camera matrix K is for 640x480 image by iphone 7
  • R and t is the rotation matrix and transformation vector of camera pose
  • Pose Matrix is saved in pose and printed

Note

The code is not up to date.