CN104794345B

CN104794345B - Platypelloid type assumes association process method more in Track In Track

Info

Publication number: CN104794345B
Application number: CN201510189952.7A
Authority: CN
Inventors: 王运锋; 吴义忠
Original assignee: Sichuan University
Current assignee: Sichuan University
Priority date: 2015-04-21
Filing date: 2015-04-21
Publication date: 2017-12-01
Anticipated expiration: 2035-04-21
Also published as: CN104794345A

Abstract

The invention discloses a kind of a kind of effective processing scheme when complex environment lower sensor measurement report carries out assuming to associate more with the system flight path safeguarded, the program by being converted into platypelloid type structure more by the conventional tree structures for assuming association, assign each relevance assumption different weights simultaneously, the complexities assumed association and safeguarded can be effectively reduced more, while the accuracy of association can also be effectively increased.

Description

Platypelloid type assumes association process method more in Track In Track

Technical field

The present invention relates generally to the processing of complex scene target tracking domain data correlation, is especially carrying out the more mesh of complex scene During mark tracking, insurmountable situation is handled by correlating methods such as arest neighbors, universe arest neighbors, the probability interconnections of routine Under, using assume track association when.When generally assuming track association more, establish flight path and assume relevance tree, in complicated feelings Relevance tree occurs that geometry speed increases under condition, safeguards that implementation process brings dyscalculia to association, The present invention gives one kind It is effectively to assume that track association handles improved method more, under the premise of effectively solving the accuracy of track association, reduce more hypothesis and close Join the complexity of processing, avoid the occurrence of the situation of relevance tree geometric growth.

Background technology

In existing all kinds of Information Handling Systems, aerial, ground mobile target is detected by various kinds of sensors, passed through Information processing system is that target establishes effective flight path maintenance system, in processing procedure, due to by all kinds of interference and sensing , there are a variety of interrelational forms in the reasons such as device detection, between realizing sensor target measurement report and safeguarding flight path for observation Scene is responsible for the difference of degree, and common measurement report associates with the correlating method safeguarded between flight path including arest neighbors, universe Arest neighbors association, more hypothesis associations, probabilistic data association etc., every kind of association algorithm has its scene and advantage and disadvantage for adapting to.Often The processing of rule hypothesis association more is will be possible as the flight path positioned at the multiple measurement reports for predicting association door for safeguarding flight path Affiliated partner, establish one-to-many relevance assumption, and in the follow-up measurement report cycle, each relevance assumption occurs again Multiple relevance assumptions, under complex scene, more hypothesis relevance trees of geometry expansion occur quickly, as shown in Figure 1.

3 measurement reports are carved with during T2 to be located in the correlation threshold of the flight path, are assumed correlation rule according to more, are established 3 Association is built, and dotted line mark is connected to the situation of measurement report 1,2,3 in figure, during to the T3 moment, assumes association to above-mentioned 3, Occur measurement report 4-10 multiple relevance assumptions respectively again, i.e., by two moment of T3, T4, there have been 7 to assume association, Complexity is very big in routes planning and maintenance.And in actual treatment, usual somebody is using the limitation too fast increasing of hypothesis tree Long situation, i.e. 3 moment of restricted T, three hypothesis spread scenarios, only allow at the T3 moment to retain original it is assumed that this limit only The complexity of maintenance is reduction of, but to the accuracy aspect of track association, causes certain loss, it is possible to occurs closing by mistake Connection situation occurs.

The present invention is improved for the tree-like relational structure of more hypothesis of this geometric growth shown in Fig. 1, proposes one More hypothesis correlating methods that kind platypelloid type is safeguarded, and assume, according to follow-up hypothesis spread scenarios, to set different weights to each, The selection that can be assumed in subsequent treatment according to the size of weights, in the case where ensureing to associate accuracy, effectively drop The low complexities for assuming association more.

The content of the invention

The invention mainly relates to complex scene lower sensor measurement report and safeguard that flight path is carried out in more hypothesis association process, Assume to assign different processing weights by the improvement processing methods assumed relevance tree and be adjusted to platypelloid type maintenance, and to each more, So as to realize the method for determining final association results, following four step is mainly reflected in.

（1）By the tree structures for assuming association more, platypelloid type enclosed structure is converted to, is entered with the examples reference actually encountered Row analysis, process is as shown in Fig. 2 effectively reduce the complexity of tree structure maintenance.

（2）Processing weights are assigned for the hypothesis in each platypelloid type structure, assignment method is as follows：

Each initial weight for assuming association is arranged to 0；

Next detection cycle, when there is the measurement report being associated（No matter single or multiple more hypothesis associate）, power Again plus a（Such as value a=2）, and more hypothesis associated weights that its direct subsequent cycle derives from are added into 2a（Indirect derivation does not change Become）, by that analogy；When it does not have measurement report to be associated, its weight is subtracted into 5a, by that analogy.

（3）In platypelloid type structure, only retain 2 of maximum weight, remaining deletion, because these hypothesis include two Class, one is already belonging to out-of-date it is assumed that the second is the interruption for belonging to no subsequent association is assumed.

（4）Selected according to the weights of platypelloid type structure, i.e. the great hypothesis of right to choose is non-as output, maximum weights In the case of unique, a conduct can be selected to export in real time wherein, in platypelloid type classification, only maintain the two of maximum weight It is individual it is assumed that treating the follow-up multicycle to continue to verify.

Brief description of the drawings

Fig. 1 is more hypothesis association tree structures；

Fig. 2 is more hypothesis association platypelloid type structure weights assignment and selection schematic diagram.

Embodiment

（1）Software design and development environment

In complex environment targetpath of the present invention tracking measurement report point mark assume with flight path more relating design and Development environment is as follows：

Operating system：Windows XP and above version, Linux, VxWorks etc.；

Software translating environment：C/C++ compilers.

（2）Measurement report-flight path assumes association implementing procedure more

The software includes following four step in implementation process.

The first step：

The sensor measurement report in the range of the system maintenance Trajectory Prediction correlation threshold is calculated, when multiple measurements being present During report, establish multiple hypotheis tracking and safeguard list（The t2 moment in figure）, and one is selected as output, and to all hypothesis in real time Flight path in linked list is safeguarded, is assumed distribution initial weight to each, is calculated its predicted position, be easy to next measurement Cycle is associated detection.

Second step：

When next measurement report cycle arrives, according to the predicted position of all hypothesis of background maintenance, detection is located at Measurement report in each prediction correlation threshold, the weight computing designed according to the present invention is regular, calculates the weights of each hypothesis.

3rd step：

Being exported in real time it is assumed that being used as maximum weight is selected, when there are multiple maximum weights, wherein optionally；Only simultaneously Two kinds of maximum weight are it is assumed that delete the less hypothesis of other weights in maintained hypothesis list；And the relevance assumption of maintenance is entered Row subsequent time status predication.

4th step：

Second and third above-mentioned step is repeated, more hypothesis association tracking can be repaired automatically.

Claims

1. a kind of complex scene lower sensor measurement report and flight path carry out the design method that platypelloid type is assumed to associate more, based on C/ C++ compiler languages, in any platform applications of Windows or Linux or Vxworks, mainly include：

（1）By the conventional tree structures for assuming association more, platypelloid type enclosed structure is converted to, reduces answering for tree structure maintenance Miscellaneous degree；

（2）Processing weights are assigned for the hypothesis association in each platypelloid type structure, assignment method is as follows：

Each initial weight for assuming association is arranged to 0；

Next detection cycle, when there is the measurement report being associated, weight increase a, and its subsequent cycle is directly derived from More hypothesis associated weights increase 2a, more hypothesis associated weights of indirect derivation keep constant, by that analogy；When it is not surveyed When amount report is associated, its weight is subtracted into 5a, by that analogy；

（3）In platypelloid type structure, only retain 2 of maximum weight, remaining deletion, because these hypothesis include two classes, its First, already belong to out-of-date it is assumed that the second is the interruption for belonging to no subsequent association is assumed；

（4）Selected according to the weights of platypelloid type structure, i.e. the great hypothesis of right to choose is as output, maximum weights identical In the case of, select one of as output in real time, in platypelloid type classification, only maintain two of maximum weight it is assumed that after The continuous multicycle continues to verify.