You are on page 1of 1

Computer Vision Group

Houston, we have a problem! No, make it four!


The Easy One:
1. Write a program to take an image as input and produce the output image with the following format :

The output image should be the same size as the input image.
Divide the output image into four quarters. Each quarter should be as mentioned below.
The top left quarter should contain only the red channel value and it should appear red as
well(recall the first presentation!). The top right should contain the blue channel and it should
appear blue. Similarly, the bottom left should contain green and appear green.
The bottom right should contain grayscale values of corresponding pixels.

The more challenging ones:


2. Write a program to detect the corners of an oblique white rectangle in a black background. This
rectangle would be brighter than the background. However, the exact pixel value may not be 255. Take
the image as input and print the coordinates of the rectangle vertices.
3. Write a program to detect all the circles in an image using the Hough Circle Transform. You may have
to play around with the various parameters to get good detection. Take an image as input and draw the
detected circles on the output image in the respective locations.
4. An edge can be roughly thought of as a location where there is a sharp change in the pixel value.
Using this as the criterion for edges, write a program to take an image as input and identify the edges in
the image. Don't use Canny/Sobel/Laplace or other edge detection functions.
Hint : The strength of an edge is dependent on the difference between the pixel to the right and the
pixel to the left(or top and bottom)
Try your code on different images, and compare it with the Canny edge detector.

You might also like