Question 1 (55 marks). Index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 CONOUNP Red 93 89 103 115 135 153 162 188 218 142 48

[answerhappygod]

Question 1 55 Marks Index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Conounp Red 93 89 103 115 135 153 162 188 218 142 48 1 (324.9 KiB) Viewed 56 times

Question 1 (55 marks). Index 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 CONOUNP Red 93 89 103 115 135 153 162 188 218 142 48 66 33 125 12 20 Green 123 89 136 147 159 164 173 188 215 147 49 67 31 121 11 19 Blue 144 94 156 166 174 172 182 193 217 155 52 71 33 123 12 baru 20 K-means clustering is useful in computer and engineering fields. For instance, in computer graphics, K-means clustering could be used to reduce the number of different colors in a stored image. In this question, we'll consider the color image on the top, which initially contains 16 different colors. Each used color is represented as a tuple of 3 numbers (range 0..255), for its Red, Green and Blue components respectively. O represent the total absence of such color, while 255 represent 100% of such color. For instance, yellow, which is formed by mixing 100% red, 100% green and 0% blue is stored as {Red: 255, Green: 255, Blue: 0}. The 16 used colors of the image are tabulated (see above table).

Now we want to use K-means clustering to reduce the number of colors from 16 to 3. Derive the Red/Green/Blue values for the 3 colors with intermediate steps (0 marks if no steps are given). Apply Euclidean distance when deriving distances between colors and centroids. Also, use real numbers for calculation and answers (3 decimal places) despite that the original numbers are integers. a) Use color indexes 2, 5 and 8 from the table above as the initial centroids. b) Then rework on the clustering all over again (with intermediate steps), this time use color indexes 2, 3 and 4 respectively as the initial centroids. Do you get the same result?