基于模拟退火粒子群优化算法的ECT图像重建方法研究

Research on ECT Image Reconstruction Based on Simulated Annealing Particle Swarm Optimization Algorithm

电容层析成像系统中,其反问题的“病态性”特点,导致传统图像重建算法的重建结果伪影现象严重。粒子群算法(Particles Swarm Optimization,PSO)作为智能算法的一种,具有易实现,收敛快等显著优点,缺点也很明显,即粒子易收敛于局部最优解。将模拟退火算法与粒子群搜索算法相结合,利用模拟退火算法中的概率突变能力,能够有一定概率跳出算法的局部最优解。设计了五种不同的典型流型并进行了仿真实验,利用所提算法与LBP算法、Tikhonov算法、Landweber迭代算法以及标准PSO算法分别对其进行了图像重建。仿真实验所得到的主观结果和客观数据均表明,所提算法可以有效减少重建图像中的伪影,明确图像形状,提高重建图像质量,使得图像重建结果更加接近原始流型。

The ill-posed inverse-problem of ECT(Electrical Capacitance Tomography)has exposed the result of classic reconstruction algorithms to the risks of serious artifacts.As one of the intelligence algorithms,PSO(Particles Swarm Optimization)is effortless to implement and fast to converge,however,it may lead to local optimum.Both the Simulated Annealing algorithm(Simulated Annealing algorithm,SA)and PSO are taken into account to resolve this drawback.SA has probability mutation ability so that the result can probably get rid of the limitation of local optimum.To verify the effectiveness of the proposed algorithm,five different flow patterns are set as reconstruction targets in which The LBP algorithm,Tikhonov algorithm,Landweber algorithm and standard PSO are compared groups.As a result,the experiment shows that the proposed algorithm plays a significant role in cutting down the artifact,upgrading the quality of reconstructed images and helping to approach to the original flow pattern.