Chinese Journal of Lasers, Volume. 49, Issue 13, 1309001(2022)
Distortion-Invariant Target Recognition Based on Multichannel Joint Transform Correlator
Fig. 4. SDF images synthesized with 5 training images under different scaling sizes when initial angle is 0° and angle interval is 70°
Fig. 6. Relationship among SPCP with rotation angles and number of channels (preferred phases are chosen based on L-PCM, the first channels are marked with red margins). (a) 3D histogram; (b) 2D unfold chart
Fig. 7. Correlation outputs without and with preferred phases. (a) Correlation output without preferred phase; (b) correlation output with public preferred phase (original background); (c) correlation output with public preferred phase (simple background)
Fig. 8. Relationship among SPCP with rotation angles and number of channels (preferred phases are chosen based on variation of position of correlation peak, the right ones are marked with dotted lines and solid margins). (a) 1st4th channels (scaling size is 0.8); (b) 5th~8th channels (scaling size is 1.0); (c) 9th channel (2D unfold chart, and scaling size is 0.6)
Fig. 10. Variation of position of correlation peak and correlation output at scaling size of 0.4. (a) Relationship between variation of pixel position of CPI with running times; (b) correlation output without preferred phase; (c) correlation output with public preferred phase
Fig. 11. Variation of position of correlation peak and correlation output at scaling size of 0.6. (a) Relationship between variation of pixel position of CPI and running times; (b) correlation output without preferred phase; (c) correlation output with public preferred phase
Fig. 12. Relationship among SPCP with rotation angles and number of channels (SDF synthesized with 7 training images and
Fig. 13. SPCP diagram and variation of position of correlation peak (SDF synthesized with 9 training images and scaling size of 1.0). (a) Relationship among SPCP with rotation angles and number of channels; (b) relationship between variation of pixel position of CPI with running times
Fig. 14. Variation of position of correlation peak with running times (SDF synthesized with 11 training images and scaling size of 1.0)
Fig. 15. Scene images with varied background. (a) Original background; (b) background with slight variation; (c) background with large variation; (d) complex background
Fig. 16. Relationship among SPCP with rotation angles and number of channels (background with slight variation, SDF synthesized with 5 training images, and scaling size of 1.0)
Fig. 17. Relationship among SPCP with rotation angles and number of channels (background with large variation, SDF synthesized with 5 training images, and scaling size of 1.0)
Fig. 18. Relationship among SPCP with rotation angles and number of channels (complex background with recalculated phases, SDF synthesized with 5 training images, and scaling size of 1.0)
|
Get Citation
Copy Citation Text
Chao Lin, Yanli Han, Shuli Lou, Pei Liu, Wenlong Zhang, Zikang Yang. Distortion-Invariant Target Recognition Based on Multichannel Joint Transform Correlator[J]. Chinese Journal of Lasers, 2022, 49(13): 1309001
Category: holography and information processing
Received: Sep. 26, 2021
Accepted: Dec. 14, 2021
Published Online: Jun. 27, 2022
The Author Email: Lin Chao (vestigelinchao@163.com)