Part 1 - start with: greyscale image manipulations. Wow factor: sharpened/unsharp mask, and edges (a Sobel operator)
-
def correlate(image, kernel): Image Filtering via Correlation
-
def blurred(image, n): Blurring with Box blur
-
def sharpened(image, n):
- A sharpen filter, aka an unsharp mask because it results from subtracting an "unsharp" (blurred) version of the image from a scaled version of the original image. If we have an image (IM) and a blurred version of that same image (B), the value of the sharpened image (S) at a particular location is:
- S_{x,y} = 2IM_{x,y} - B_{x,y}
-
def edges(image): Sobel operator, a neat (super cool!) filter, uses for detecting edges in images.
Part 2 - from greyscale to color image manipulatios, making use of functional programming. Wow factor: filter_cascade, seam_carving, vignette.
-
def color_filter_from_greyscale_filter - convert greyscale_filters implemented so far into color_filters that work on color images
-
def filter_cascade (work of wonder) - cascading filters into one super-powered filter: functional chaining, 9 levels high
-
def seam_carving - content-aware resizing i.e. retargeting (such a dope technique)
-
def vignette: apply Gaussian Kernel (similar to cv2.getGaussianKernel(), and compute Frobenius matrix norm, similar to numpy.linalg.norm()).
-
//TODO seam filling - smart resizing to increase the size of an image by inserting appropriate rows at low-energy regions in the image.
