gusucode.com > 扩展卡尔曼滤波,粒子滤波,去偏卡尔曼滤波和循环增益尔曼滤波的源程序 > 扩展卡尔曼滤波/new_err_count.m
%由各次的误差计算出目标的平均误差,作为跟踪的均值估计,然后根据均值估计得出方差估计 function [ME_final_view_err_x,ME_final_filter_err_x,ME_final_view_err_y,ME_final_filter_err_y,RMSE_final_view_err_x,RMSE_final_filt_err_x,RMSE_final_view_err_y,RMSE_final_filt_err_y]=new_err_count(ME_temp_view_err_x,ME_temp_filter_err_x,ME_temp_view_err_y,ME_temp_filter_err_y); [ROW,COL]=size(ME_temp_view_err_x); for i=1:COL ME_final_view_err_x(i)=sum(ME_temp_view_err_x(:,i))/ROW; ME_final_filter_err_x(i)=sum(ME_temp_filter_err_x(:,i))/ROW; ME_final_view_err_y(i)=sum(ME_temp_view_err_y(:,i))/ROW; ME_final_filter_err_y(i)=sum(ME_temp_filter_err_y(:,i))/ROW; end for i=1:COL RMSE_final_view_err_x(i)=sum((ME_temp_view_err_x(:,i)-ME_final_view_err_x(i)*ones(ROW,1)).^2)/ROW; RMSE_final_filt_err_x(i)=sum((ME_temp_filt_err_x(:,i)-ME_final_filt_err_x(i)*ones(ROW,1)).^2)/ROW; RMSE_final_view_err_y(i)=sum((ME_temp_view_err_y(:,i)-ME_final_view_err_y(i)*ones(ROW,1)).^2)/ROW; RMSE_final_filt_err_y(i)=sum((ME_temp_filt_err_y(:,i)-ME_final_filt_err_y(i)*ones(ROW,1)).^2)/ROW; end