基于模糊神经网络的舰船雷达图像弱小目标检测  

Dim and small target detection in ship radar images based on fuzzy neural network

作　　者：张勇飞[1] 陈涛[2] ZHANG Yong-fei;CHEN Tao(College of Science and Technology,Nanchang University,Jiujiang 332020,China;School of Mathematics and Computer Sciences,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学科学技术学院,江西九江332020 [2]南昌大学数学与计算机学院,江西南昌330031

出　　处：《舰船科学技术》2024年第9期147-150,共4页Ship Science and Technology

基　　金：国家自然科学基金资助项目(71363043);2021年度江西省教育厅科学技术研究资助项目(GJJ217813);2022年江西省高等学校教学改革研究省级课题资助项目(JXJG-22-30-6)。

摘　　要：舰船雷达图像信息的维度较高,导致弱小目标的关键特征难以被精准提取,降低了弱小目标检测的可靠性,因此提出一种基于模糊神经网络的舰船雷达图像弱小目标检测方法。该方法对舰船雷达图像进行背景校正,利用图像灰度值加性模型从图像中提取弱小目标。最后将提取的弱小目标输入到模糊神经网络中,输出的结果即为舰船雷达图像弱小目标检测结果。通过实验证明,在不同高斯噪声环境中,该方法能够准确地检测出雷达图像中的弱小目标,并具有较快的检测速度。The high dimensionality of ship radar image information makes it difficult to accurately extract key features of weak targets,reducing the reliability of weak target detection.Therefore,a ship radar image weak target detection method based on fuzzy neural network is proposed.This method performs background correction on ship radar images and extracts weak targets from the images using an image grayscale additive model.Finally,the extracted weak targets are input into the fuzzy neural network,and the output result is the weak target detection result of the ship radar image.Through experiments,it has been proven that this method can accurately detect weak targets in radar images in different Gaussian noise environments and has a fast detection speed.

关 键 词：舰船雷达图像 弱小目标检测 图像灰度值 高斯噪声 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]