Question

Resampling: Do not use any in-built functions from opencv and numpy. In general, you can use function from math library. Functions allowed for this part are: np.array(), np.matrix(), np.zeros(), np.ones(), cv2.imread(), cv2.namedWindow(), cv2.waitKey().

Write code for zooming and shrinking an image using the nearest neighbor and bilinear interpolation. The input to your program is: (i) image, (ii) transformation parameters, and (iii) interpolation method.

(Can I get more of an explanation and puesdocode of how you use nearest neighbor and bilinear interpolation to do this without built-in functions)

Answer 1

Answer:
Any image is two dimensional array and with each element having a value .i.e. intensity of that pixel.
1 2 3
4 8 9
7 4 1
In above all the numbers are pixel values according to RGB scale.

Methods:
Nearest Neighbor method:
For any image, at any given co-ordinate or at any pixel set, finding the "Nearest pixel value" for zooming or shrinking operations.

Bilinear Interpolation:
Zoom or shrink, 2D array of the image is resized .i.e
Zoom: 3x3 image would be 9x9 or 12x12, new values of rows and colums can be calculated throught bilinear interpolation method.
Shrink: 10x10 image would be shrink to 5x5, then again finding the new values of that matrix using this method only.

Approach to solution:
For any given image, it's size can be known and pixel values can be calculated.
Above two are transformational parameters along with size to be achieved after conversion.
In Nearest neighbor method, each neighbors pixel value is calculated and image is created, whereas in bilinear interpolation, a random small size matrix is chosen and corresponding new matrix pixels are calculated.