MATLAB多传感器融合--核心步骤

MATLAB的多传感器融合的核心步骤在stepImpl函数中实现，该函数的输入的跟踪目标和测量的目标的信息，输出为证实的真目标信息和处于试探的跟踪目标信息。

[confirmedTracks, tentativeTracks, allTracks] = stepImpl(tracker, detections, varargin)

该函数涉及到的核心步骤如下所示

(1)通过代价矩阵计算出测量的目标那些能够与跟踪目标关联即Assignments 目标，哪些是属于没有匹配成功的跟踪目标UnassignedTracks，哪些是没有匹配成功的测量目标UnassignedDetections。

[Assignments, UnassignedTracks, UnassignedDetections] =...

                                   tracker.associateDetectionsToTracksByUserCost(varargin{end});

 

(2) 将未匹配成功的测量目标UnassignedDetections初始化成新的跟踪目标

% Initiate tracks based on unassigned detections

initiateTracks(tracker, UnassignedDetections);

 

(3) 将匹配成功的目标Assignments的数据在跟踪目标中进行更新

% Update tracks based on assigned detections.

updateAssignedTracks(tracker, Assignments);

 

(4) 将未匹配成功的跟踪目标UnassignedTracks根据时间决定是否删除

% Delete tracks that are not assigned

deleteOldTracks(tracker, UnassignedTracks, time);

 

(5)预测所有跟踪目标的状态

% Predict all the tracks to the end of the time step.

predictTracks(tracker, time);