Image Processing and Computer Vision Assignments
Assignment 1 -
1. Region grow Segmentations (in MATLAB):
1.1 Choose some image from our set of images and run the region grow on a 50x50 segment of the image where there are some interesting parts. Use the random map to display.
1.2 First apply a smoothing filter (with a boxcar or Gaussian filter) on the 50x50 and then re-run the region grow and compare the results. Use the random map to display.
2. K-means segmentation
2.1 Use the entire original image from 1 and generate the histogram of the image.
2.2 Run the k-means algorithm with k=5 and k=7. Use the random map to display.
2.3 Smooth the original image with a boxcar or Gaussian filter and then re-run the k-means with the two values from 2.2. Use the random map to display.
3. Discussion:
3.1 Compare and contrast the various results from 1 and 2.
4. Extra credit - I threw together the region grow pretty quickly. Can you rewrite it to run faster? Discuss what you did to improve the speed.
Assignment 2 -
1. K-means segmentation with Connected Components
1.1 Select an interesting image and generate the histogram of the image.
1.2 Run the k-means algorithm with k=5 and k=7. Use the random map to display.
1.3 Smooth the original image with a boxcar or Gaussian filter and then re-run the k-means with the two values from 2.2. Use the random map to display.
1.4 Run the connected components code I uploaded (get_all_blobs_fill_2.m) NOTE: you will need the other routines as well in that directory. Try with a minimum blob size of 50 and then with 100.
1.5 Discussion: Compare and contrast the various results.
2. Texture
2.1 Create three 2-D Law filters (you can pick any combination. Show the filter kernel.
2.2 Process two of the texture images and display the results.
2.3 Histogram the results for each. Discuss how similar or different the results are.
2.4 Run the filters on the segment2.bmp image.
2.5 Display the resulting images from the three filters. Discuss how different the amplitudes are in each of the quadrants of the image.
Attachment:- Assignment.rar