Exercise 5 2009. Classification using Bayes rule class channel 3 02 243 240 pixel 1 2 3 4 MEAN VAR 5 6 co2 $ channel 2 244 243 244 245 244 0.67 164 169 170 189 173 120.67 46 68 003 channel 1 248 247 248 249 248 0.67 174 177 184 195 182.5 87 170 136 142 118 146.3 491.67 0.0.6. 6. 8. SSBBSBPeeeee 243 249 243.75 14.25 49 36 65 91 60.25 560.92 48 46 c2 8 MEAN VAR 9 10 11 12 MEAN VAR 001 B&B 19 38.75 526.25 39 41.75 41.58 Figur 1: Training image For the multispectral image (3 channels) above, we have the training data picked from the indicated regions as given in the table Your task is to classify the pixels in the image below into class 01 (background), 02 (leaf). and 03 (apple ) based on the information from the image in Figur 1. You are only allowed to use two channels. a) Evaluate by visual inspection (using 2D feature spaces) which two channels are the most suited. 250 250 250 data1 A2 data 200 dalat dan dala 200 200 A2 dated 150 150 Febre2 Fenone 3 E BURE 100 100 100) 50 120 140 160 180 200 220 240 Feature 1 120 140 160 180 200 220 240 Feature 1 50 100 150 Feature 2 200 250 b) Based on this you are designing a classifier using Bayes rule. The data is modeled with gaussian distributions having the same covariance matrix E=l6 Write the generic form of the discriminant function