Feature Selection for SAR Target Discrimination and Efficient Two-Stage Detection Method

Feature-based target detection in synthetic aperture radar (SAR) images is required for monitoring situations where it is difficult to obtain a large amount of data, such as in tactical regions. Although many features have been studied for target detection in SAR images, their performance depends on the characteristics of the images, and both efficiency and performance deteriorate when the features are used indiscriminately. In this study, we propose a two-stage detection framework to ensure efficient and superior detection performance in TSX images, using previously studied features. The proposed method consists of two stages. The first stage uses simple features to eliminate misdetections. Next, the discrimination performance for the target and clutter of each feature is evaluated and those features suitable for the image are selected. In addition, the Karhunen-Loeve (KL) transform reduces the redundancy of the selected features and maximizes discrimination performance. By applying the proposed method to actual TerraSAR-X (TSX) images, the majority of the identified clusters of false detections were excluded, and the target of interest could be distinguished.

Automated summary

In this study, a two-stage detection framework is proposed for feature-based target detection in synthetic aperture radar (SAR) images. The framework ensures efficient and superior detection performance in TerraSAR-X (TSX) images by using previously studied features. The first stage eliminates misdetections using simple features, and the second stage evaluates the discrimination performance of each feature and selects the suitable features for the image. The proposed method also incorporates the Karhunen-Loeve (KL) transform to reduce redundancy and maximize discrimination performance.