Image Reconstruction Algorithm for Electrical Capacitance Tomography Based on Constrained Particle Swarm Optimization

Abstract

When the particle swarm is active, the robustness of the system is quite great, which is very useful for solving ill-conditioned problems like picture reconstruction. In the reconstructed image, however, the large number of pixels results in the large dimension of particles, making it harder for particles to attain the ideal solution during the optimization process. A constraint condition is applied to the position of particles to solve this problem. The image reconstruction algorithm regularized by Tikhonov is used as the particle position reference to constrain the particles to search within a certain range of the algorithm, and the penalty function is used to solve the problem to improve the particle search speed. For the inertia weight of particle swarm, this paper uses linear decreasing weight to realize its adaptive dynamic adjustment, which improves the flexibility of the algorithm. When chaos operator is added to the particle swarm location search process, if the particle is trapped in a local optimum, the chaos variable will fluctuate within a certain range, thus reducing the error rate of the optimal solution. The experimental comparison results illustrate that the modified particle swarm optimization technique for the image reconstruction of Electrical Capacitance Tomography outperforms the classic LBP and Tikhonov algorithms in terms of accuracy and efficiency.