A Probabilistic Predictive Target Tracking Method

Abstract

The present invention relates to a probabilistic predictive target tracking method, which is mainly applied to intelligent machines. However, the existing target tracking algorithm only gives the track of the target after filtering, which has a certain delay, and because it cannot obtain Therefore, in practical applications, traditional target tracking algorithms have certain limitations. In order to change this situation and make it applicable to changing and complex environments, a new ideological framework is proposed, which enables intelligent machines to detect whether the target object is doing non-motorized motion. Once the target enters the stage of non-motorized motion, start Kalman filtering, and can give prediction and tracking results in the form of probability.

Description

technical field [0001] The invention relates to a target tracking method, in particular to a probabilistic prediction type target tracking method, which is applied in the field of target tracking. Background technique [0002] The existing target tracking algorithm only gives the track of the target after filtering, which has a certain delay, and because the complete model in the complex environment cannot be obtained, the traditional target tracking algorithm has certain limitations in the complex environment. Such as interactive multi-model and variable-structure interactive multi-model: the purpose of this method is to build a relatively complete target model, and then perform a series of tracking predictions on the target. [0003] Disadvantages of the traditional target tracking algorithm: When the characteristics of the target are well known, the corresponding model can be easily found by using the multi-model method, so that the accuracy of target tracking is more acc...

AI Technical Summary

Problems solved by technology

[0003] Disadvantages of the traditional target tracking algorithm: When the characteristics of the target are well understood, the corresponding model can be easily found by using the multi-model method, so that the accuracy of target tracking is more accurate

However, when the target characteristics are not known, this method can only use common models and use a series of different parameters to model, so it is difficult to accurately determine the target model; Competition is also more intense, making tracking data imprecise

Images