Underwater video image enhancement

Poor visibility in underwater is a major problem for many applications of computer vision. Optically, poor visibility underwater is due to the substantial presence of particles that have significant size and distribution in the participating medium. Light from the water and light reflected from an object are absorbed and scattered by those particles, causing the visibility of a scene to be degraded [1].

Several algorithms on image dehazing and visibility improvement were proposed recently [1-6]. The objectives of this project consist of
(1) an evaluation and comparison of state-of-the-art methods for image dehazing, clarification, enhancing, and visibility improvement;
(2) extension/adaptation of image visibility improvement methods for underwater image restoration purposes [7];
(3) extensions to video processing taking into account spatio-temporal dependencies to improve the robustness of the methods.

[1] Tan, R.T., Visibility in bad weather from a single image. In CVPR 2008.
[2] Fattal, R., Single image dehazing. ACM Transactions on Graphics, 27(3), 2008.
[3] He, K., Sun, J. and Tang, X., Single image haze removal using dark channel prior. IEEE PAMI, 33(12):2341-2353, 2011.
[4] Galdran, A., Vazquez-Corral, J., Pardo, D. and Bertalmío, M., Enhanced variational image dehazing. SIAM Journal on Imaging Sciences,8(3):1519-1546, 2015.
[5] Morel, J.M., Petro, A.B. and Sbert, C., Screened Poisson equation for image contrast enhancement. Image Processing On Line, 4, pp.16-29, 2014.
[6] Bahat, Y. and Irani, M., Blind Dehazing Using Internal Patch Recurrence. ICCP, 2016.
[7] Qing, C., Yu, F., Xu, X., Huang, W. and Jin, J., Underwater video dehazing based on spatial–temporal information fusion. Multidimensional Systems and Signal Processing, pp.1-16, 2016
[8] Toda, M., Senzaki, K. and Tsukada, M., Image Clarification Method Based on Structure-Texture Decomposition with Texture Refinement. In International Conference on Image Analysis and Processing, pp. 352-362, Springer, 2015.

Supervisor name: 
Yvan Petillot
Supervisor and Deputy email addresses: 

Project Type:

Project location: 
The project will be carried out in EM building on the Riccarton Campus
Deputy name: 
Alexander Belyaev
Staff comments: 
This project requires a good background in signal and image processing.