Skip to main content

Table 6 The comparison of average RR results among the proposed, Chang and Lin's, and previous averaging schemes for Sets 2-5 images

From: On the pinned field image binarization for signature generation in image ownership verification method

Operation

Type

Specification

Proposed GA scheme

Previous averaging scheme[18]

Chang and Lin's scheme[1]

   

RR

RR

RR

     

t = 160

t = 180

t = 200

t = 230

t = 250

Blurring

Gaussian filter

9 × 9

99.68%

99.40%

99.76%

99.73%

99.97%

99.80%

99.86%

Surround cropping

 

26%

98.52%

86.98%

88.93%

89.06%

89.14%

89.17%

89.17%

Quarter cropping

 

25%

98.38%

86.68%

89.84%

90.60%

91.15%

91.89%

92.26%

Noise addition

Gaussian noise

Mean = 0

Variance = 0.01

95.78%

77.52%

91.12%

92.18%

92.95%

93.70%

94.11%

JPEG compression

 

Quality factor = 80

99.89%

99.17%

99.80%

99.87%

99.89%

99.91%

99.83%

Scaling

  

99.86%

98.75%

99.60%

99.70%

99.74%

99.71%

99.84%

Sharpening

linear mapping

 

99.84%

99.66%

91.49%

91.67%

92.02%

92.80%

92.76%

Median filtering

nonlinear filter

9 × 9

98.66%

92.12%

95.83%

96.30%

96.89%

97.22%

97.50%

Average filtering

linear filter

9 × 9

98.20%

92.59%

95.44%

95.88%

96.49%

96.86%

97.15%

Gamma correction

 

0.7

99.35%

98.43%

95.47%

95.91%

96.15%

96.21%

96.17%

Histogram equalization

 

Uniform

96.98%

95.40%

78.34%

78.88%

79.69%

81.10%

81.77%

Pixel shifting

 

60 × 60

91.73%

50.55%

53.98%

56.03%

57.52%

60.07%

61.49%

ARR

  

98.07%

89.77%

89.97%

90.48%

90.80%

91.53%

91.82%