Infrared weak and small target identification method based on shape prior segmentation and multi-scale feature aggregation

Abstract

The invention discloses an infrared weak and small target identification method based on shape prior segmentation and multi-scale feature aggregation, and the method comprises the steps: carrying out the Gaussian filtering operation of an input infrared original image, so as to enhance a dim weak and small target; performing shape prior-based segmentation on the infrared image after Gaussian filtering to obtain a target candidate area; cutting the target candidate region and inputting the cut target candidate region into a multi-scale feature extraction module to obtain feature representation of the small target; inputting the feature representation of the small target into a feature aggregation network to obtain an image after tensor splicing; and carrying out batch normalization processing and nonlinear transformation on the image after tensor splicing, and outputting a target classification result through Softmax. The segmentation module based on shape prior makes full use of prior information of a weak target to obtain a suspicious target area, the global parameter quantity is reduced to improve the algorithm efficiency, the multi-scale feature extraction and aggregation module realizes a sufficient number of feature channels for the weak target, and then the detectability of the weak target is ensured.

Description

technical field [0001] The invention belongs to the technical field of infrared target detection, in particular to an infrared weak and small target recognition method based on shape prior segmentation and multi-scale feature aggregation. Background technique [0002] The use of infrared imaging detection technology can realize the detection and tracking of UAVs, which is a technical means for effective monitoring of UAVs. However, in actual scenes, due to the influence of long-distance imaging and atmospheric radiation interference, the target has a low signal-to-noise ratio, few pixels, no shape, texture and structure information, and is easily interfered by complex background clutter and random noise, making Conventional target detection and recognition algorithms cannot balance detection accuracy and detection efficiency. [0003] In order to solve the problem of infrared weak and small target detection, there are mainly two types of methods: single-frame-based and mult...

AI Technical Summary

Problems solved by technology

However, most of the current deep learning detection methods regard the problem of infrared weak and small target detection as a problem of binary classification or saliency detection, and the detection accuracy of weak and small targets is still not ideal.

Images