Sobel Edge Detector 

  For our second implementation of logic locking was sobel edge detection. For our edge detection we first created a MATLAB code to run real time digital signal processing over a live video feed from a camera. Edge detection utilizes convolution over the data feed from the camera to identify where the changes in color value lie within an image and assigns the magnitude of this operation as an eight-bit color value, a gradient of grayscale.

​

  For the final design, a new edge detection algorithm written by our graduate student assistant Tyler was used that calculated edges at more angles and a binary color scheme (black or white with no gray).

​

  Once we got the final design, we ran the edge detector through the obfuscation algorithm. After running it through the algorithm, we were able to get a key. We used the key to get the edge detected image. Without the key, the image would not show or it was distorted. 

​

The video to the left, demonstrates what happens when the key is incorrect. 

​

​