本文提出一种基于卷积神经网络的船舶红外图像边缘检测方法。首先，介绍船舶红外探测技术的基本原理，针对船舶红外图像的预处理进行研究，包括灰度的均衡化、红外图像的背景抑制、图像分割等。设计了一个基于卷积神经网络的红外图像边缘检测模型，该模型采用多层卷积和池化操作，以及非线性激活函数，能够有效地捕捉图像中的边缘信息。最后，通过对模型进行训练和优化，得到了准确度较高的船舶红外图像探测算法，为后续船舶的目标识别和跟踪提供了有效的基础。
In this paper, a ship infrared image edge detection method based on convolutional neural network is proposed. Firstly, the basic principles of ship infrared detection technology are introduced, and the preprocessing of ship infrared images is studied, including grayscale equalization, background suppression of infrared images, image segmentation, etc. An infrared image edge detection model based on convolutional neural network is designed, which adopts multi-layer convolution and pooling operations and nonlinear activation function, which can effectively capture edge information in images. Finally, through the training and optimization of the model, a ship infrared image detection algorithm with high accuracy is obtained, which provides an effective basis for the target recognition and tracking of subsequent ships.
2023,45(17): 154-157 收稿日期：2023-04-11
DOI：10.3404/j.issn.1672-7649.2023.17.030
分类号：U623.15
作者简介：赵明庆(1969-),男,硕士,高级实验师,研究方向为机器学习及计算机应用
参考文献：
[1] 刘志东. 改进神经网络的船舶红外图像边缘检测方法[J]. 舰船科学技术, 2023, 45(7): 166-169.
[2] 王璐烽. 船舶监控红外图像处理中多通道SDN控制器改进设计[J]. 舰船科学技术, 2021, 43(20): 196-198.
[3] 贾晓霞. 基于船舶红外图像的耦合Sine混沌映射加密方法[J]. 舰船科学技术, 2021, 43(16): 175-177.
[4] 余永武. 红外图像和神经网络在船舶目标识别中的应用[J]. 舰船科学技术, 2021, 43(12): 175-177.
[5] 喻会. 基于红外图像的海上船舶特征识别算法研究[J]. 舰船科学技术, 2020, 42(4): 40-42.YU Hui. Research on feature recognition algorithm of maritime ships based on infrared image[J]. Ship Science and Technology, 2020, 42(4): 40-42.
[6] 方玲玲, 赵婉彤, 王相海. 船舶红外成像目标跟踪研究进展[J]. 计算机辅助设计与图形学学报, 2017, 29(10): 1776-1785.
