Siddhartha Ghosh, Dirk Froebrich, and Alex Freitas, "Robust autonomous detection of the defective pixels in detectors using a probabilistic technique," Appl. Opt. 47, 6904-6924 (2008)
Detection of defective pixels in solid-state detectors/sensor arrays has received limited research attention. Few approaches currently exist for detecting the defective pixels using real images captured with cameras equipped with such detectors, and they are ad hoc and limited in their applicability. In this paper, we present a probabilistic novel integrated technique for autonomously detecting the defective pixels in image sensor arrays. It can be applied to images containing rich scene information, captured with any digital camera equipped with a solid-state detector, to detect different kinds of defective pixels in the detector. We apply our technique to the detection of various defective pixels in an experimental camera equipped with a charge coupled device (CCD) array and two out of the four HgCdTe detectors of the UKIRT’s wide field camera (WFCAM) used for infrared (IR) astronomy [Astron. Astrophys. 467, 777–784 (2007)].
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Blocks of 25 pixels containing the defective pixels were analyzed. For every p total number of images chosen was such that the number of image data sets constructed was seven in all cases. For , for instance, images were analyzed.
Table 7
Performance at the Detection of Different Faulty Pixels Detected by the Ground-Truth Techniquea
Ground-Truth
Parameters
Result
Pixel Type
Exposure
Temperature
Detection Rate
100
100
65
56.67
100
100
82.3
90
100
100
100
93.34
100
100
100
76.67
100
100
100
90
41.17
100
100
100
63.15
96.67
78.5
56.67
In total 72 images per trial, with , resulting in six image data sets were used and blocks of 25 pixels containing the defective pixels were analyzed.
Table 8
Performance of our Technique at Detecting Different Pixels Subject to Different Values of Gain (m)a
Detection Rate for Different Values of Gain (m)
Exposure
0.94
0.95
0.96
0.97
0.98
0.99
1.01
1.02
1.03
1.04
1.05
1.06
1.07
1.08
100
86.7
63.3
60
33.3
23.3
10
6
26.7
53.3
73.3
93.3
100
100
100
100
100
90
70
56.7
40
23.3
13.3
13.3
30
43.3
73.3
83.3
86.7
93.3
96.6
100
In total 60 images per trial, with , resulting in six image data sets (one used for GMM step and the remaining five for the Bayes step). Windows containing 25 pixels are analyzed.
Table 9
Performance of our Technique at Detecting Clusters of Defective Pixels, Subject to Different Values of Gain (m)a
Gain (m)
Dead
Stuck
0.91
0.94
1.04
1.08
Cluster Size
DR
2
100
100
100
75
70
100
3
96.7
99.6
94.4
78.9
71.1
96.7
4
96.7
99.4
88.3
76.7
76.7
93.3
Total 60 images per trial and , resulting in six image data sets (one used for GMM step and remaining five for Bayes step). Windows containing 25 pixels are analyzed.
Tables (9)
Table 1
Posterior Probabilities and Classification of Pairwise Difference Estimates
Pixel a
Pixel b
1
2
0.1
0.7
0.2
2
1
3
0.15
0.8
0.05
2
Table 2
Detection Rate for Different Numbers of Image Data Sets
Data Sets
Ground-Truth
1
2
3
4
Pixel Type
Detection Rate
100
100
100
100
100
78.5
56.7
50
36.67
26.67
Table 3
Numbers of Selections of Different Numbers of GMM Components by the BIC Criterion (K) in 30 Trials
Blocks of 25 pixels containing the defective pixels were analyzed. For every p total number of images chosen was such that the number of image data sets constructed was seven in all cases. For , for instance, images were analyzed.
Table 7
Performance at the Detection of Different Faulty Pixels Detected by the Ground-Truth Techniquea
Ground-Truth
Parameters
Result
Pixel Type
Exposure
Temperature
Detection Rate
100
100
65
56.67
100
100
82.3
90
100
100
100
93.34
100
100
100
76.67
100
100
100
90
41.17
100
100
100
63.15
96.67
78.5
56.67
In total 72 images per trial, with , resulting in six image data sets were used and blocks of 25 pixels containing the defective pixels were analyzed.
Table 8
Performance of our Technique at Detecting Different Pixels Subject to Different Values of Gain (m)a
Detection Rate for Different Values of Gain (m)
Exposure
0.94
0.95
0.96
0.97
0.98
0.99
1.01
1.02
1.03
1.04
1.05
1.06
1.07
1.08
100
86.7
63.3
60
33.3
23.3
10
6
26.7
53.3
73.3
93.3
100
100
100
100
100
90
70
56.7
40
23.3
13.3
13.3
30
43.3
73.3
83.3
86.7
93.3
96.6
100
In total 60 images per trial, with , resulting in six image data sets (one used for GMM step and the remaining five for the Bayes step). Windows containing 25 pixels are analyzed.
Table 9
Performance of our Technique at Detecting Clusters of Defective Pixels, Subject to Different Values of Gain (m)a
Gain (m)
Dead
Stuck
0.91
0.94
1.04
1.08
Cluster Size
DR
2
100
100
100
75
70
100
3
96.7
99.6
94.4
78.9
71.1
96.7
4
96.7
99.4
88.3
76.7
76.7
93.3
Total 60 images per trial and , resulting in six image data sets (one used for GMM step and remaining five for Bayes step). Windows containing 25 pixels are analyzed.