OpenCV 3 Computer Vision with Python Cookbook

How it works

In this recipe, we open a video, detect the initial keypoints using the cv2.goodFeaturesToTrack function that we used earlier, and start tracking points using the sparse Lucas-Kanade optical flow algorithm, which has been implemented in OpenCV with the cv2.calcOpticalFlowPyrLK function. OpenCV implements a pyramidal version of the algorithm, meaning that the optical flow is first calculated in an image of a smaller size, and then refined in a bigger image. The pyramid size is controlled with the maxLevel parameter. The function also takes parameters of the Lucas-Kanade algorithm, such as window size (winSize) and termination criteria. The other parameters are previous and current frames, and keypoints from the previous frame. ...