Feature-Point-Tracking

The program contains implementation of Feature Point detection and tracking algorithm using Lucas-Kanade Algorithm. The users can either manually select the points(Cherry Picked) to track or let the program find optimal points to track. We utilize most of the functions from the earlier labs such as gaussian, gaussian derivative, convolve, etc. The program begins by taking in an input as the first image from a sequence (“flowergarden”, “statue_seq”). As an example, we’ll select the “flowergarden” image sequence. The program then gives us an option to choose from Cherry-picked or Haris corner. If the user chooses cherry-picked, the program prompts to select the first image from the sequence. Then it asks the user to mark 21 feature points. Preferably, these points should be corner points. The program then runs the Lucas-Kanade algorithm on the image with the selected feature points. The output of the lucas-kanade function generates the coordinates for the next frame of the feature point compared to the current one. If the user chooses the Haris-corner option, the program picks the top feature points and runs the algorithm similar to the previous option. The program also makes sure that the next feature point placed is atleast 2-city block distant from another ones.