Design Method of Radar Ground Target Discriminator Based on Spatial Distribution

Abstract

The invention discloses a spatial distribution-based method for designing radar ground target discriminators, and solves the problem in the prior art that a ground target cannot be accurately discriminated in real time in a complex environment. The method comprises the steps of 1) constructing a training template library feature space of a radar target discriminator; 2) dividing the region of the training template library feature space; 3) generating the boundary and the support vector of the region based on a weighted optimization-based support vector domain model; 4) discriminating the type of a test target according to the characteristics of the test target and the location relationship between the boundary and the support vector of the region. According to the technical scheme of the invention, the region of the training template library feature space is divided, so that the operation speed is greatly improved. At the same time, the target discrimination accuracy is effectively ensured based on the refined region description method. In this way, the discrimination performance for ground targets in the complex environment is improved. Meanwhile, the target discriminator has the real-time processing capability. The above method can be used for the effective authentication of ground targets by a radar system in the complex environment.

Description

technical field [0001] The invention belongs to the field of radar target recognition, and relates to a design method of a radar ground target discriminator based on space distribution, which can be used in a radar system in a complex environment to realize ground target discrimination. Background technique [0002] Radar target recognition technology is based on the detection and positioning of the target by radar, extracting the characteristic information of the target, and realizing the judgment of the target attribute and type. For the specific application background of the radar system, the synthetic broadband radar system is used to generate a high-resolution one-dimensional range image HRRP of the ground scene. The scene includes the target to be attacked and the interference of false targets, such as iron towers, civilian vehicles, and road signs. The scattering characteristics of such false targets are related to the The target to be attacked is similar, and it cann...

AI Technical Summary

Problems solved by technology

SVDD is a typical OCC classifier. This classifier determines the boundary of the template library sample feature area by looking for support vectors, and uses the positional relationship between the boundary and the sample to be tested to determine the test sample category. Its disadvantage is that when the template library sample features When the area exhibits multi-area aggregation, the determined feature area boundary will contain a large area of ​​non-sample target feature space, resulting in a serious decline in performance

The KNN classifier is based on the template matching framework, which needs to traverse all the sample features of the template library, which has high computational complexity and cannot meet the requirements of real-time processing of radar signals.

Images