Weak target detection method based on deep neural network and time-frequency image sequence

Abstract

The invention discloses a weak target detection method based on a deep neural network and a time-frequency image sequence. The weak target detection method comprises the steps of obtaining a to-be-detected time-frequency image sequence; performing convolution feature sequence extraction on the to-be-detected time-frequency image sequence by using a deep convolution neural network model to obtain aconvolution feature map sequence; performing time sequence feature extraction on the convolution feature map sequence by using a recurrent neural network to obtain a single-frame time-frequency feature map; and calling a regional suggestion network to perform point-by-point target and background discrimination and target frame adjustment on the time-frequency feature map.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and in particular relates to a weak target detection method based on a deep neural network and a time-frequency image sequence. Background technique [0002] Affected by factors such as noise and clutter, when the radar detects low RCS targets, the low signal-to-noise ratio may cause the target to be completely submerged in background noise or clutter. In order to ensure a certain detection probability, a low threshold value must be set, and a low threshold value will lead to a significant increase in the probability of false alarms. Therefore, traditional CFAR object detection algorithms are limited. [0003] The deep neural network brings new ideas to target detection. It relies on the powerful feature expression ability brought by the deep network to abstract the low-level features of the original input data into high-level features, which is more conducive to tasks such as target...

AI Technical Summary

Problems solved by technology

[0004] The technology of the present invention solves the problem: overcomes the deficiencies of the prior art, provides a weak target detection method based on deep neural network and time-frequency image sequence, overcomes the shortcomings of traditional target detection algorithms that need to pre-fit the background noise distribution, and enhances the The adaptive ability and robustness of the target detection algorithm improve the detection accuracy while reducing false alarms

Images