视觉传感器网络中基于RANSAC的顽健定位算法

视觉传感器网络由于节点故障或环境变化将导致节点对目标的观测数据出现错误，而基于最小二乘的多视觉信息融合定位方法将因此造成较大的定位误差。针对此问题提出一种基于集中式RANSAC的顽健定位算法，将错误数据进行筛选剔除，从而提高定位精度，进一步针对集中式 RANSAC 将会导致单个节点的计算复杂度过高而导致网络节点能耗不平衡问题，提出基于分布式 RANSAC 的顽健定位算法，从而将大量的迭代计算平均分布在各个节点中并行处理，在保证定位过程顽健性的同时保证了网络的计算能耗平衡性。最后通过实验对no-RANSAC、cen-RANSAC 和 dis-RANSAC算法的定位性能进行了比较，验证了该算法能够依照预定的概率获得良好的定位结果，并对算法的时间复杂度进行了分析。

RANSAC based robust localization algorithm for visual sensor network

Due to node failures or environmental changes,observed data on the target will be error in visual sensor network,so the least squares based multi-vision localization algorithm won’t be accurate.A centralized RANSAC based robust localization method was proposed,which can remove the error data and improve the positioning accuracy.Furthermore,to address this issue,energy imbalance of network nodes in centralized RANSAC where all computing load is executed in one single node,a robust localization algorithm based on distributed RANSAC was also proposed,which can distribute a large number of iterative calculations into each node averagely to ensure the network balance on calculation and energy without affecting the robustness.Finally,through comparing experiments on no-RANSAC,cen-RANSAC and dis-RANSAC,it’s verified that this algorithm can obtain robust and good positioning results in a predetermined probability,and the time complexity was analyzed via experiment.