Partial derivatives in x.
Partial derivatives in y.
Binarized edge detection.
Partial derivatives in x (after a Gaussian blur).
Partial derivatives in y (after a Gaussian blur).
Binarized edge detection. I noticed that the partials produced more defined results, and the resulting edge detection created edges that appeared continuous, rather than 1.1's edge detection that looked more like a discrete set of points that more or less fit to an edge. It appears to be an overall superior method of edge detection compared to 1.1
Original image
Straightened
Histogram
Original
Straigthened
Histogram
Original
Straigthened
Histogram
Original
Straigthened - Failure Case. Interesting that images with tons of detail and Fourier frequencies can straighten, but a simple plus sign can't with my algorithm.
Histogram
Original Image
Sharpened
Original Image Blurred
Blur resharpened - unfortunately, my filter only mildly effective at resharpening blurred images.
Original Image
Sharpened
Original Image Blurred
Blur resharpened
Original Image
Sharpened
Original Image Blurred
Blur resharpened
Starting Images
Hybridized
FFT Histogram - There is a small gap of unrepresented frequencies, and unfortunately I did not have sufficient time to tune the parameters and fix this.
Starting Images
Hybridized (imperfectly)
Starting Images
Hybridized (imperfectly)
Gaussian and Laplacian pyramids, respectively.
Gaussian and Laplacian pyramids, respectively.
Gaussian and Laplacian pyramids, respectively.
Gaussian and Laplacian pyramids, respectively.
Starting Images
Blended
Starting Images
Blended (imperfectly)