Extended-range imaging in underwater environments can be regarded as an image blurring process. The final image can be modeled as a convolution of the point spread function (PSF) and the original image as corrupted by ambient noise. In the underwater laser line scan system, the noise distribution generally follows Poissonian distribution.
The EODES model was developed by Metron (Reston, VA) for the simulation and performance prediction of electro-optic imaging systems in ocean environments. EODES can be used to predict the system PSF. The EODES-predicted PSF is then used to improve the image quality obtained using pulsed or CW laser line scan system. With the EODES-predicted PSF available, it is now possible to enhance the image quality via deconvolution. The Richardson-Lucy filter is one of the most effective tools in recovering images polluted by Poissonian noise. The following images demonstrate the image enhancement results using EODES-predicted PSF against images obtained in high turbidity (C=0.9) at extended range (D=7M). This, in turn, validated the accuracy of the EODES model.
(a) Original Image using pulsed LLS at high turbidity (b) Enhanced Image using RL filter
For more information, please contact Bing Ouyang