Sea surface small target feature detection method based on time-frequency domain depth network

Abstract

The invention discloses a sea surface small target feature detection method based on a time-frequency domain deep network. The method comprises the following steps: obtaining an observation vector of a to-be-detected unit, and building a ternary hypothesis testing problem; converting the observation vector into a two-dimensional time-frequency domain to obtain a two-dimensional time-frequency graph; performing normalization processing on the two-dimensional time-frequency graph to obtain a normalized time-frequency graph; constructing a deep network model, and extracting probability values belonging to the three types of hypotheses as feature values; constructing a 2D feature vector as a final test statistic; under a given false alarm rate, determining a judgment area with controllable false alarm in combination with a cubic spline curve algorithm; and calculating whether the test statistics are in the judgment region. The neural network and the feature detection technology are combined, so the method has the advantages of autonomous learning feature extraction and multi-dimensional feature combination, and improves the detection performance of the radar on the small sea surface target under the low signal-to-clutter ratio.

Description

technical field [0001] The invention belongs to the technical field of radar signal processing, and in particular relates to a feature detection method of a small target on the sea surface based on a time-frequency domain deep network, which is suitable for a sea detection radar with a long-term observation system. Background technique [0002] At present, small sea targets such as boats, speedboats, and aircraft wreckage have become the focus and difficulty of marine radar detection. Typically, these small targets are physically small and have stealthy materials such that the signal-to-clutter (SCR) is often borderline detectable. Compared with air targets, low-speed moving sea targets usually have weaker maneuverability, which makes their spectrum easily fall in the main clutter area of ​​sea clutter, which increases the difficulty of detection. [0003] In radar signal processing, long-term observation accumulation is an effective way to improve the detection performance...

AI Technical Summary

Problems solved by technology

This kind of intelligent detection method is an effective way to improve the detection performance. The difficulty lies in the design of the false alarm controllable judgment area.

Images