Problem 1: MORPHOLOGICAL PROCESSING
A binary image, sample1.png, is given in Figure 1. Please implement several morphological operations and provide discussions about the results. (Note that the white pixels represent foreground objects and the black pixels are background.)
(a) (15 pt) Perform boundary extraction on sample1.png to extract the objects’ boundaries and output the result as result1.png.
(b) (15 pt) Perform hole filling on sample1.png and output the result as result2.png.
(c) (30 pt) Please design an algorithm to count the number of objects in Figure 1. Describe the steps in detail and specify the corresponding parameters.
(a) sample1.png
Figure 1: The test image for morphological processing.
Problem 2: TEXTURE ANALYSIS
In this problem, there is an image sample2.png composed of several different textures.
(a) (10 pt) Perform Law’s method on sample2.png to obtain the feature vector of each pixel and discuss the feature vectors in your report.
(b) (10 pt) Use k-means algorithm to classify each pixel with the feature vectors you obtained from (a). Label same kind of texture with the same color and output it as result3.png.
(c) (20 pt) Based on result3.png, design a method to improve the classification result and output the updated result as result4.png. Describe the modifications in detail and explain the reason why.
(d) (Bonus) Try to replace the flowers in color or gray-scale sample2.png with sample3.png or other texture you prefer by using the result from (c), and output it as result5.png.
It’s allowed to utilize external libraries to help you accomplish it, but you should specify the implementation detail and functions you used in the report.
(a) sample2.png (color) (b) sample2.png
(c) sample3.png Figure 2: Images for texture analysis.
Appendix
Problem 1: MORPHOLOGICAL PROCESSING
sample1.png: 400 × 400
Problem 2: TEXTURE ANALYSIS
gray-scale
sample2.png: 400 × 600
gray-scale
sample2 color.png: 400 × 600
color
sample3.png: 150 × 200
gray-scale
