CUDA, OpenMP, and regular serial C implementations of Shi Tomasi feature detection. Done as a comparison of architecture project.
usage: ./stfd [-v,-vv] <full path to the image> [sigma] [windowsize] [num_features]
flags:
-h: show this help menu
-v: output basic execution information
-vv: output all information... good for debugging
arguments:
sigma: the sigma value for the Gaussian distribution used to form the convolution mask.
windowsize: the size of a pixel 'neighborhood' in an image
num_features: how many features to extract
MIT