Skip to main content

Table 4 Classification rates for the test data set, using the proposed features

From: Efficient and accurate document image classification algorithms for low-end copy pipelines

Ground truth

Classification rates, %

 
 

color-text

color-mix

clor-picture

color-photo

mono-text

mono-mix

mono-picture

mono-photo

color-text

58/58

42/42

-/-

-/-

-/-

-/-

-/-

-/-

color-mix

-/-

96/98

2/2

2/-

-/-

-/-

-/-

-/-

color-picture

-/-

61/61

39/39

-/-

-/-

-/-

1/1

-/-

color-photo

-/-

36/42

-/-

64/58

-/-

-/-

-/-

-/-

mono-text

13/13

9/9

-/-

-/-

62/56

16/23

-/-

-/-

mono-mix

-/-

9/9

-/-

-/-

1/3

86/86

3/1

-/-

mono-picture

-/-

5/5

6/6

-/-

-/-

40/40

49/49

-/-

mono-photo

-/-

4/4

-/-

2/2

-/-

14/42

-/-

80/58

  1. Each entry in the table is “A/B” where A and B are the classification percentages, respectively, for the proposed classifier of Fig. 3 a and for the hard-decision tree classifier of Fig. 1 a, both used with the feature set proposed in the present paper