Probability prediction type target tracking method

Abstract

The present invention relates to a probability prediction type target tracking method and is mainly applied to an intelligent machine. However, according to an existing target tracking algorithm, only a track after target filtering is given, so that the target tracking has a delay, and because a complete model in a complex environment can not be obtained, the conventional target tracking algorithm has a limitation in practical application. In order to change the existing state and enable the target tracking method to be suitable for a changeable and complex environment, a new thought framework is proposed, wherein the intelligent machine can detect whether a target object is performing a non-motorized motion, and once the target enters a non-motorized motion stage, the intelligent machine starts Kalman filtration and is capable of giving out results of prediction and tracking 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