Multi-band color image deblurring using contourlets for fluid lens camera systems

Fluidic lens camera systems present a new field of exploration for both the optics and image processing communities. Developed for surgical applications, these cameras do not have moving parts while zooming and they have better miniaturization possibilities. However, the lens causes non-uniform color blur between color planes which creates an image processing problem. We propose the use of a contourlet filter bank system to deblur color images without estimating a point spread function. This multi-band deblurring method uses sharper color planes to improve blurred ones. Compared to the conventional Lucy-Richardson andWiener filtering, our previous wavelet-based method significantly improves sharpness and ghosting artifacts. The proposed contourlet-based system better adjusts to the natural image contours. This effect produces an image with a similar level of sharpness, but fewer ghosting artifacts. Furthermore, we analyze conditions for when this algorithm will reduce the mean squared error. We also use an Ant Colony Optimization algorithm to detect sharp edges. This algorithm naturally extends to many systems for multi-band deblurring that have high edge correlation.