Author Affiliations
1School of Electronics and Information Engineering, Wuzhou University, Wuzhou 543001, Guangxi, China2School of Mechanical and Electrical Engineering, Guilin University of Electronic Technology, Guilin 531004, Guangxi, China3School of Information and Control Engineering, China University of Mining and Technology, Xuzhou 221116, Jiangsu, China4Department of Electrical and Computer Engineering, University of Alberta, Edmonton T6G 1H9, Alberta, Canada5School of Science and Information, Qingdao Agricultural University, Qingdao 266109, Shandong, Chinashow less
Fig. 1. Infrared maritime images under backlight condition
Fig. 2. Histograms of sub-images in Fig. 1
Fig. 3. Target detection results under different maritime conditions using classical LCM. (a) Target detection result under backlight condition with ω1=9 and k=3; (b) target detection result under backlight condition with ω1=27 and k=3; (c) target detection result under heavy wave interference with ω1=7 and k=3; (d) target detection result under heavy fog with ω1=9 and k=3
Fig. 4. Schematic diagram of original histogram modification
Fig. 5. Results of new histogram equalization
Fig. 6. Structural element in filter and edge information after amplification. (a) Structural element in filter; (b) result of edge information of #1 in Fig.1 amplified by 10 times; (c) result of edge information of #3 in Fig.1 amplified by 10 times
Fig. 7. Flowchart of HEPLEF algorithm
Fig. 8. Enhancement results of HEPLEF algorithm. (a) Entire result and local target region of #1 in Fig.1; (b) enhancement result and local target region of #1 in Fig.1; (c) entire result and local target region of #3 in Fig.1; (d) enhancement result and local target region of #3 in Fig.1
Fig. 9. Comparison of enhancement results before and after edge information fusion. (a) After edge information fusion; (b) before edge information fusion
Fig. 10. Enhancement result of #4 in Fig. 1 by HEPLEF algorithm. (a) Original image; (b) enhancement result
Fig. 11. Principle for target detection based on local contrast saliency and minimum target detection unit under single scale. (a) Principle for target detection based on local contrast saliency; (b) minimum target detection unit under single scale
Fig. 12. Schematic diagram of target detection unit with multiple scale
Fig. 13. Pseudocode of LCMMBC algorithm
Fig. 14. Schematic diagram of target detection result and local contrast saliency by LCMMBC algorithm. (a) Schematic diagram of target detection result; (b) schematic diagram of local contrast saliency
Fig. 15. Comparison of enhancement results for #1 in Fig. 1 obtained by different algorithms. (a) Classical histogram equalization algorithm; (b) MMBEBHE algorithm; (c) ETHE algorithm (Tthreshold=5); (d) Retinex algorithm (k=7)
Fig. 16. Comparison of enhancement results for #3 in Fig. 1 obtained by different algorithms. (a) Classical histogram equalization algorithm; (b) MMBEBHE algorithm; (c) ETHE algorithm (Tthreshold=5); (d) Retinex algorithm (k=7)
Fig. 17. Target recognition result and three-dimensional diagram of local contrast saliency for #1 in Fig. 1 with w1=9 and k=2.(a) Target recognition result; (b) three-dimensional diagram of local contrast saliency
Fig. 18. Target recognition result and three-dimensional diagram of local contrast saliency for enhancement result of #1 in Fig. 1 with w1=9 and k=2. (a) Target recognition result; (b) three-dimensional diagram of local contrast saliency
Image | #1 in Fig. 1 | Fig. 15(a) | Fig. 15(b) | Fig. 15(c) | Fig. 15(d) | Fig. 8(b) |
---|
AG | 1.11 | 1.67 | 1.46 | 1.33 | 0.99 | 2.34 | Image | #3 in Fig. 1 | Fig. 16(a) | Fig. 16(b) | Fig. 16(c) | Fig. 16(d) | Fig. 8(d) | AG | 1.07 | 2.71 | 1.56 | 3.51 | 1.55 | 4.72 |
|
Table 1. AG of original image and enhancement algorithms
Image | Fig. 15(a) | Fig. 15(b) | Fig. 15(c) | Fig. 15(d) | Fig. 8(b) |
---|
LCG | 3.36 | 1.14 | 0.30 | 2.78 | 3.46 | Image | Fig. 16(a) | Fig. 16(b) | Fig. 16(c) | Fig. 16(d) | Fig. 8(d) | LCG | 2.07 | 0.49 | 2.17 | 2.54 | 2.67 |
|
Table 2. Comparison of LCG of different enhancement algorithms
Indicator | MPCM | MLHM | DECM | LCMMBC |
---|
DR | 95.2 | 96.4 | 95.8 | 99.8 | FAR | 36.1 | 42.5 | 43.6 | 23.4 |
|
Table 3. Comparison of target detection rates and target false alarm rates between different algorithms unit: %