Abstract

Remote sensing is a rich topic due to its utility in gathering detailed accurate information from locations that are not economically feasible traveling destinations or are physically inaccessible. However, poor visibility over long path lengths is problematic for a variety of reasons. Haze induced by light scatter is one cause for poor visibility and is the focus of this article. Image haze comes about as a result of light scattering off particles and into the imaging path causing a haziness to appear on the image. Image processing using polarimetric information of light scatter can be used to mitigate image haze. An imaging polarimeter which provides the Stokes values in real time combined with a “dehazing” algorithm can automate image haze removal for instant applications. Example uses are to improve visual display providing on-the-spot detection or imbedding in an active control loop to improve viewing and tracking while on a moving platform. In addition, removing haze in this manner allows the trade space for a system operational waveband to be opened up to bands which are object matched and not necessarily restricted by scatter effects.

© 2013 Optical Society of America

Full Article  |  PDF Article

References

  • View by:
  • |
  • |
  • |

  1. Y. Y. Schechner, S. G. Narasimhan, and S. K. Nayar, “Polarization-based vision through haze,” Appl. Opt. 42, 511–525 (2003).
    [CrossRef]
  2. S. Shwartz, E. Namer, and Y. Schechner, “Blind haze separation,” in 2006 Proceedings of IEEE CVPR (IEEE, 2006), p. 0-7695-2597-0/06.
  3. J. Mudge and M. Virgen, “Near-infrared simultaneous Stokes imaging polarimeter: integration, field acquisitions, and instrument error estimation,” Proc. SPIE 8160, 81600B (2011).
    [CrossRef]
  4. D. Goldstein, Polarized Light (Marcel Dekker, 2003).
  5. G. D. Boreman, Modulation Transfer Function in Optical and Electro-Optical Systems (SPIE, 2001).
  6. R. D. Fiete, Modeling the Imaging Chain of the Digital Cameras (SPIE, 2010), pp. 127–167.

2011

J. Mudge and M. Virgen, “Near-infrared simultaneous Stokes imaging polarimeter: integration, field acquisitions, and instrument error estimation,” Proc. SPIE 8160, 81600B (2011).
[CrossRef]

2003

Boreman, G. D.

G. D. Boreman, Modulation Transfer Function in Optical and Electro-Optical Systems (SPIE, 2001).

Fiete, R. D.

R. D. Fiete, Modeling the Imaging Chain of the Digital Cameras (SPIE, 2010), pp. 127–167.

Goldstein, D.

D. Goldstein, Polarized Light (Marcel Dekker, 2003).

Mudge, J.

J. Mudge and M. Virgen, “Near-infrared simultaneous Stokes imaging polarimeter: integration, field acquisitions, and instrument error estimation,” Proc. SPIE 8160, 81600B (2011).
[CrossRef]

Namer, E.

S. Shwartz, E. Namer, and Y. Schechner, “Blind haze separation,” in 2006 Proceedings of IEEE CVPR (IEEE, 2006), p. 0-7695-2597-0/06.

Narasimhan, S. G.

Nayar, S. K.

Schechner, Y.

S. Shwartz, E. Namer, and Y. Schechner, “Blind haze separation,” in 2006 Proceedings of IEEE CVPR (IEEE, 2006), p. 0-7695-2597-0/06.

Schechner, Y. Y.

Shwartz, S.

S. Shwartz, E. Namer, and Y. Schechner, “Blind haze separation,” in 2006 Proceedings of IEEE CVPR (IEEE, 2006), p. 0-7695-2597-0/06.

Virgen, M.

J. Mudge and M. Virgen, “Near-infrared simultaneous Stokes imaging polarimeter: integration, field acquisitions, and instrument error estimation,” Proc. SPIE 8160, 81600B (2011).
[CrossRef]

Appl. Opt.

Proc. SPIE

J. Mudge and M. Virgen, “Near-infrared simultaneous Stokes imaging polarimeter: integration, field acquisitions, and instrument error estimation,” Proc. SPIE 8160, 81600B (2011).
[CrossRef]

Other

D. Goldstein, Polarized Light (Marcel Dekker, 2003).

G. D. Boreman, Modulation Transfer Function in Optical and Electro-Optical Systems (SPIE, 2001).

R. D. Fiete, Modeling the Imaging Chain of the Digital Cameras (SPIE, 2010), pp. 127–167.

S. Shwartz, E. Namer, and Y. Schechner, “Blind haze separation,” in 2006 Proceedings of IEEE CVPR (IEEE, 2006), p. 0-7695-2597-0/06.

Cited By

OSA participates in CrossRef's Cited-By Linking service. Citing articles from OSA journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (11)

Fig. 1.
Fig. 1.

Light from the object is attenuated as it propagates to an imaging polarimeter. The light from the object that arrives at the sensor is direct transmission. Airlight or scatter increases as it makes it way to the sensor.

Fig. 2.
Fig. 2.

Airlight peak and valley angles of the polarizer define the two images ι˜ and ι˜ used in the dehazing algorithm. The airlight degree of linear polarization is approximately 0.32.

Fig. 3.
Fig. 3.

Hazed intensity image (ι˜) is the first Stokes value and is provided by the imaging polarimeter. Top part of the image is sky, Mount Umunhum is in the middle, and the bottom is a building in the foreground. “T” shaped object is an antenna attached to the building.

Fig. 4.
Fig. 4.

Dehazed image (r˜) provides enhanced clarity of the mountain as is the building in the foreground. The airlight degree of linear polarization is 0.32.

Fig. 5.
Fig. 5.

Horizontal line plot at the vertical position pixel 91 for the hazed (black) and dehazed (blue) images. The line passes through the decommissioned radar tower which is the large peak at the horizontal position pixel 45 and the NexRad weather radar which is a smaller peak at the horizontal position pixel 152.

Fig. 6.
Fig. 6.

Radial mean PSD of the hazed (black) and dehazed (blue) image. The dehazed image provides improved frequency response below the optical cutoff frequency.

Fig. 7.
Fig. 7.

Logarithm base 10 of the radial mean PSD of the hazed (black) and dehazed (blue) image showing the improvement at the higher spatial frequencies.

Fig. 8.
Fig. 8.

Hazed intensity image (ι˜) provided by the imaging polarimeter. The top right part of the image is sky, and Mount Hamilton is in the middle left. One Lick observatory dome is at horizontal position pixel 125 and vertical position pixel 70 sitting atop the mountain.

Fig. 9.
Fig. 9.

Dehazed image (r˜) provides enhanced clarity of the mountain not to the extent of the first sample. The airlight degree of linear polarization is 0.29.

Fig. 10.
Fig. 10.

Horizontal line plot at the vertical position pixel 70 for the hazed (black) and dehazed (blue) images. The line passes through one of Lick observatory’s domes, which is the large peak at the horizontal position pixel 125.

Fig. 11.
Fig. 11.

Logarithm base 10 of the radial mean PSD of the hazed (black) and dehazed (blue) image showing the improvement at the higher spatial frequencies.

Equations (6)

Equations on this page are rendered with MathJax. Learn more.

r˜=ι˜a˜eβ˜z˜,
t(z˜)=1a˜a˜,
r˜=ι˜(ι˜ι˜)/p˜1(ι˜ι˜)/p˜a˜.
θpeak=12tan1(s2s1)fors10
θpeak=π212tan1(s2s1)fors1<0
θvalley=θpeak+π2.

Metrics