Microscope Image Processing

11.4.2 A Numerical Example

To illustrate the operation of the three-class Bayes classifier, we include

a numerical example having six pixels from each class in the training set.

While this is hopelessly inadequate for any real case, it serves to illustrate the

calculations. This example will permit readers who choose to implement a Bayes

classifier to check their implementation for numerical accuracy.

Assume the training set is

33 31 46 57 18

42 10 24 38 56

62 50 34 21 33

2

6

4

3

7

5

628472158

96 116 126 84 73

70 82 96 117 90

2

6

4

3

7

5

78 115 122 76 134

12 52 34 70 22

81 100 146 78 70

2

6

4

3

7

5

(11:21)

for the RGB color values of six pixels each of background, normal cells, and

abnormal cells, respectively. The mean vectors for the three classes are then

m

1

¼

37

34

40

2

4

3

5

m

2

¼

32 ...