PeTrack’s execution flow

This site contains a rough sketch of the program flow of PeTrack. This may help understanding the architecture a little bit better when figuring out how to develop new features.

The “main-loop” of PeTrack is the player. In the Player::playVideo method, we move from image to image - Either frames of a video or literal images of a sequence - process it and then display it. New images are retrieved here and then forwarded to methods for processing and displaying.

The heavy lifting of processing is done in the Petrack::updateImage method. First we apply different filters to the image.

• Swapping - mSwapFilter

  • Swapping the image, if demanded by the user

• Brightness & Contrast - mBrightContrastFilter

  • Improving brightness and contrast of the image

• Border

  • Adding a border to the image

  • Important: Changes pixel coordinates

  • border-width is user-given

  • must be activated by the user

• Undistortion - mCalibFilter

  • Undistorts the image

  • If distortion parameters change or the calib filter is toggled on/off, all trajectories get deleted

  • must be activated by the user

• Background Substraction - mBackgroundFilter

  • Performs background substraction

  • must be activated by the user - mostly off

  • was used in stereo-videos

After these filters were applied, tracking is executed. We execute tracking first, so new points from the recognition step can be evaluated immediately. Tracking happens in the Tracker::track method.

After tracking recognition is performed. The method of recognition is decided by the user. These new points then are added to existing trajectories or start a new one. See How does a TrackPoint get added for more details.

After all of this, the filtered Image is copied to the ImageItem’s QImage and the GraphicsView orchestrating the GUI gets updated/redrawn.