Lattice Boltzmann method for filtering and contour detection of the natural images

摘要

In this paper, the lattice Boltzmann method (LBM) is extended to study the filtering and contour detection of natural images, and a new lattice Boltzmann model is proposed for more complicated image processing model, like the Ambrosio and Tortorelli (A-T) model that contains two coupled nonlinear partial differential equations. The numerical results of image filtering and contour detection show that the noises in the image can be removed greatly, and simultaneously, important contours of the image are protected well. To improve the computational efficiency, we implement the developed lattice Boltzmann model on Graphic Processing Unit (GPU), and find that, compared to the CPU based algorithm, the GPU based LBM can gain more than 25 × speedup, which is very important in the further lattice Boltzmann study of large-scale image processing problems. And finally, these numerical results also show that the LBM is a feasible and efficient approach for filtering and contour detection of the natural images.