Optical Flow Edge Detection Alexander Darino Optical Flow

Optical Flow & Edge Detection Alexander Darino

Optical Flow a. Derive change in intensity across x and y gradients a. Obtained via convolution of below masks xmask = [ -1, 1 ymask = [ ] -1, -1 1, 1 ] • Compute net change in intensity • Plot vectors indicating direction and magnitude of change in intensity

Optical Flow: Results

Canny Edge Detection a. Computer x/y intensity gradients as before b. Determine direction of net changes of intensity a. Assign each direction to one of 8 -directional neighbors c. Determine if a pixel's change in magnitude is a local maximum in relation to it's two directional neighbors

Canny Edge Detection a. Assign pixels as edges if their change in magnitude exceeds a hi threshhold b. Disqualify pixels as edges if their change in magnitude is below a low threshhold c. For all remaining candidate pixels: if they are adjacent to a confirmed edge, then they too become an edge

Canny Edge Detection: Live Demo

Acknowledgements a. Berkan a. Clarified how convultions worked and are implemented in MATLAB b. Corrected computation error in optical flow b. Corey a. Shed light on implementation of optical flow algorithm in MATLAB c. Nancy & Rye a. Team effort in implementing optical flow algorithm in MATLAB

Thank You https: //github. com/alexanderdarino/Canny