Classification Algorithm of Ground Moving Target Based on Decision Tree

Abstract

The invention relates to a ground moving target classification algorithm based on a decision tree, which is used for accurate classification and identification of wheeled vehicles, tracked vehicles, single persons and small groups by ground battlefield reconnaissance radar in battlefield environment. First, perform clutter suppression and feature extraction, use Bayesian classifier to train training samples based on each feature, determine the classifier threshold, then calculate entropy impurity, and find the optimal feature based on the maximum reduction of entropy impurity , for hierarchical classification. The problem of optimal feature selection in the process of hierarchical classification of four types of targets is solved: the algorithm selects features as root nodes and layer nodes based on the maximum criterion of entropy impurity reduction. Compared with feature selection algorithms based on experience, feature selection based on decision trees can realize feature selection in the process of stratification and ensure the performance of hierarchical classification algorithms.

Description

technical field [0001] The invention belongs to the technical field of radar, and in particular relates to a ground moving target classification algorithm based on a decision tree, which is used for the effective classification of wheeled vehicles, crawler vehicles, individual soldiers and squads in a battlefield environment. Background technique [0002] Ground moving target classification technology is of great significance in military affairs. Ground moving target classification can be used to judge threats and obtain detailed situational conditions on the battlefield for rapid response and decision-making. For the classification of ground moving targets, the Doppler and micro-Doppler features of moving targets can be used for feature extraction to realize target classification and recognition. The invention classifies four types of ground moving targets that are focused on by battlefield reconnaissance radars, and the target types include wheeled vehicles, tracked vehicl...

AI Technical Summary

Problems solved by technology

Therefore, the traditional four-type ground moving target classification algorithm has no theoretical support for feature selection in hierarchical classification, and relies on experience, and feature selection is sometimes not optimal

Images