为提升基于PSO算法的煤矿相机参数选取的准确性,提出了一种基于惯性权重调整对粒子群算法进行优化的方法。在优化中在结合e指数法的基础上,通过粒子适应度函数值对惯性权重进行动态调节,在迭代的前期阶段,此时的适应度值较大,根据迭代次数对惯性权重进行调整,在迭代的后期阶段,此时的适应度较小,则根据适应度值对惯性权重进行调整,从而达到减少迭代次数、提高精度的目的,通过优化后的算法对张氏标定法得到的相机内参进行调节。为验证算法优化后的效果,对优化参数后的相机进行标定试验,经标定试验可知,改进PSO算法对煤矿相机参数优化后,平均误差为0.4517piexl、标准差为0.2989piexl、迭代次数为213次,对比标准的PSO算法、e指数法,能够明显提高相机参数的精准度、降低迭代时间。

To improve the accuracy of coal mine camera parameter selection based on PSO algorithm, a particleswarm optimization method based on inertia weight adjustment was proposed. On the basis of combining the e-indexmethod in optimization, the inertia weight was dynamically adjusted through the particle fitness function value. Inthe early stage of iteration, when the fitness value was large, the inertia weight was adjusted according to thenumber of iterations. In the later stage of iteration, when the fitness was small, the inertia weight was adjustedaccording to the fitness value to reduce the number of iterations and improve accuracy. The camera internalparameters obtained by Zhang’ s calibration method through the optimized algorithm was adjusted. To verify theeffectiveness of algorithm optimization, calibration experiments were conducted on the camera with optimizedparameters. After calibration experiments, it was found that the improved PSO algorithm optimized the parametersof coal mine cameras, with an average error of 0. 451 7 piexl, a standard deviation of 0. 298 9 piexl, and 213iterations. Compared with the standard PSO algorithm and e-index method, it can significantly improve the accuracyof camera parameters and reduce iteration time.