dq621 / cvtracking

Python scripts for tracking fiducials via overhead mounted cameras using the OpenCV API

CVTracking v3.07

Python scripts for tracking fiducials via overhead mounted cameras using the OpenCV API

References: A lot of the code written has been pulled from or inspired by various amazing resources from great content creators. A few shout outs include: Adrian Rosebrock from https://www.pyimagesearch.com. OpenCV Team and Contributors for tutorials, demonstrations, and source code. ...

Camera.py:

Two classes are introduced which help to initialize a usb webcamera and to initialize a webcamera stream.

Class: camObjThreaded: A class which takes in the video source (a number from 0 - n) which represents the dev/video# used to initialize an OpenCV capture device. The other intialization variable is the width. This defines the width of the stream. This is to allow for smaller resolutions which will increase processing speeds.

def __init__(): takes in the values mentioned above and then initializes the camera with a user-guided prompt. This checks to make sure the camera is calibrated for intended operations
  (CALIBRATION IS NOT AN OPTION): This design choice was made because all aruco tracking needs the mtx and dist provided from calibration. 
    In the works: This step will be expidited to prompt the user to calibrate if it doesn't find the necessary calibration files.
  Once the camera is initialized, the stream begins in a thread which handles all i/o, and undistortion of the frames being recieved by the camera.

  It is necessary that the folder calibCache is set up as follows:
  This is to provide a storage point for the mtx and dists from calibration for future use (I.E. not needing to calibrate every time)
  badImg and goodImg are directories which contain the good and bad calibration images.

  \ProjectFolder
  ---\calibCache
  ------\cam"ID"
  ---------\badImg
  ---------\goodImg
  ---------\rotationMatrices
  ---------\translationMatrices
  ---------\resolution_mtx.txt
  ---------\resolution_dist.txt
  ---\trialCache

def calibrate(): runs through an automated process of calibrating which takes images until 20 good images are captured based on the checkerboard
  calibration.

def showCalibFeed() outputs the thread stream. When closed, the FPS average will be outputted.

def start(), update() called within the class and shouldnt be called by the user

def read() returns the undistorted frame from the thread

def stop() ends the thread and stops the cv.VideoCapture() stream. Class object is rendered useless at this point and should be deleted.

class FPS: Tracks the fps of the stream and can also be used to track the average number of loop iterations per second.

class WebcamVideoStream: The base threading class that was used to help implement camObjThreaded. Thanks to pyimagesearch.com for the amazing article on threading: source URL: https://www.pyimagesearch.com/2015/12/21/increasing-webcam-fps-with-python-and-opencv/

class camObj = previous iteration of the camera object. This is implemented without the threading and as such was slow and cumbersome in application.

Tracker.py: README in progress