基于改进粒子群的虹膜定位算法

提出了一种改进的粒子群优化算法（IPSO）来解决虹膜定位问题。该算法采用两种速度更新策略来增强种群多样性并提高算法自身的收敛速度，并提出一种变异操作以阻止IPSO陷入局部最优。对虹膜内进行边缘定位时，通过搜索6条直线与虹膜内边缘的交点来获得12个边缘点；另外建立了与这12个点有关的目标函数，并用IPSO来优化该函数。根据IPSO在该函数上的应用，找到一个最合适的圆来拟合虹膜内边缘。进行虹膜外定位时，设计了一个模板来提取虹膜外边界，然后从外边界中选择12个边缘点，并同样使用IPSO找到一个最合适的圆来拟合虹膜外边缘。为了验证基于改进粒子群优化算法的虹膜定位方法（ILA-IPSO）的性能，从中国科学院自动化研究所的数据库中选择了不同个体的108幅虹膜图像。实验结果表明，ILA-IPSO算法要好于其它两种方法，该算法利用最少的定位时间获得了最高的成功率。

Iris location algorithm based on improved particle swarm optimization

An Improved Particle Swarm Optimization (IPSO) algorithm was proposed to solve iris localization. The IPSO adopted two velocity updating strategies to diversify the swarm and to improve its convergence rate. In addition, a mutation operation was used to prevent the IPSO from being trapped into the local optimums. For inner location, twelve edge points were determined by finding the intersection points of six straight lines with the inner boundary. A function related to these twelve points was constructed, and this function was optimized by IPSO. By applying IPSO to this function, a most suitable circle could be found to fit the inner boundary of the iris. With respect to outer location, a template was designed to extract the outer boundary of the iris, and twelve edge points were equally selected from the outer boundary of the iris. By applying IPSO to a function related to these twelve points, a most appropriate circle could be obtained to fit the outer boundary of the iris. 108 iris images of different individuals were chosen from the eye image database of Institute of Automation of Chinese Academy of Sciences (CASIA) to test the performance of our proposed Iris Location Algorithm based on Improved Particle Swarm Optimization (ILA-IPSO) algorithm. Experimental results reveal that the ILA-IPSO performs better than the other two approaches for iris localization. This method uses the least time of iris localization, and obtains the highest success rate.