Marine Ship Target Segmentation Method Based on Improved DeepLabV3+

doi:10.12382/bgxb.2024.0912

Abstract

Abstract:

Ship target segmentation plays a crucial role in enhancing the automation level of marine monitoring systems.However,the existing target segmentation methods often suffers from the issues such as mis-segmentation,susceptibility to noise interference,and low processing efficiency,making them difficult to adapt flexibly and efficiently to real-world scenarios.To address these challenges in maritime ship target segmentation tasks,a segmentation algorithm based on an improved DeepLabV3+ network,termed MET-DeepLabV3+,is proposed.During the image preprocessing stage,the bilateral filtering,multi-scale Retinex algorithm,and linear transformation are employed to mitigate the effects of noise,weather and other adverse factors.In terms of model design,the lightweight MobileNetV2 is used as the backbone network to reduce model complexity,and the ECA-Net attention mechanism is introduced to enhance the model's ability to capture the multi-scale features.Additionally,a transfer learning approach is adopted,and the feature weights pre-trained on the SeaShips dataset are applied to model training to further optimize the segmentation performance.Experimental results demonstrate that the improved algorithm achieves an average interaction-over-union (IoU) of 91.62% and an average pixel accuracy (PA) of 92.87%,which are improved by 2.13% and 1.11%,respectively,compared with those of the baseline DeepLabV3+ model.Moreover,it outperforms the segmentation models such as HRNet,PSPNet,and UNet,which effectively meets the practical demands of ship target segmentation tasks and offers significant application value.

Key words: ship target segmentation, DeepLabV3+, lightweight network, attention mechanism, transfer learning

CLC Number:

TP391

TAO Weihao, LUO Yasong, LIN Fajun, QU Jianjing, LIU Yiping. Marine Ship Target Segmentation Method Based on Improved DeepLabV3+[J]. Acta Armamentarii, 2025, 46(8): 240912-.

Figures/Tables 20

References 39

[14]	CHEN L C, ZHU Y K, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation: arXiv:1802.02611[R]. Ithaca,NY, US: Cornell University, 2018:1802.02611.
[15]	郭开武. 基于空间特征相关的图像去噪及分割算法研究[D]. 西安: 西安电子科技大学, 2014.
	GUO K W. Research on image denoising and segmentation algorithms based on spatial feature correlation[D]. Xi'an: Xidian University, 2014. (in Chinese)
[16]	钱康亮. 基于图像局部滤波去噪增强算法的图像语义分割模型[J]. 机械设计与制造工程, 2023, 52(10):109-113.
	QIAN K L. Image semantic segmentation model based on local filtering denoising and enhancement algorithm[J]. Journal of Mechanical Design and Manufacturing Engineering, 2023, 52(10):109-113. (in Chinese)
[17]	TOMASI C, MANDUCHI R. Bilateral filtering for gray and color images[C]// Proceedings of the Sixth International Conference on Computer Vision. Bombay,India: IEEE, 1998:839-846.
[18]	RAHMAN Z, JOBSON D J, WOODELL G A. Multi-scale retinex for color image enhancement[C]// Proceedings of the 3rd IEEE International Conference on Image Processing. Lausanne,Switzerland: IEEE, 1996, 3:1003-1006.
[19]	MUNTEANU C, ROSA A. Color image enhancement using evolutionary principles and the retinex theory of color constancy[C]// Neural Networks for Signal Processing XI:Proceedings of the 2001 IEEE Signal Processing Society Workshop. North Falmouth,MA,US: IEEE, 2001:393-402.
[20]	CHEN L C. Rethinking atrous convolution for semantic image segmentation:arxiv:1706.05587[R]. Ithaca,NY,US: Cornell University, 2017:1706.05587.
[21]	CHOLLET F. Xception:deep learning with depthwise separable convolutions[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu,HI,US: IEEE, 2017:1251-1258.
[22]	刘文祥, 舒远仲, 唐小敏, 等. 采用双注意力机制DeepLabv3+算法的遥感影像语义分割[J]. 热带地理, 2020, 40(2):303-313. doi: 10.13284/j.cnki.rddl.003229
	LIU W X, SHU Y Z, TANG X M, et al. Semantic segmentation of remote sensing images using dual attention mechanism DeepLabV3+ algorithm[J]. Tropical Geography, 2020, 40(2):303-313. (in Chinese)
[23]	FU H X, GU Z Q, WANG B Y, et al. Marine target segmentation based on improved DeepLabV3+[C]// Proceedings of the 2022 41st Chinese Control Conference. Hefei,China: IEEE, 2022:7314-7319.
[24]	CHENG L, XIONG R, WU J R, et al. Fast segmentation algorithm of USV accessible area based on attention fast DeepLabV3[J]. IEEE Sensors Journal, 2024, 24(15):24168-24177.
[25]	陈昌川, 郝晓严, 龙虹毓, 等. 基于改进DeepLabv3+网络的沥青道路裂缝检测方法[J]. 半导体光电, 2024, 45(3):493-500.
	CHEN C C, HAO X Y, LONG H Y, et al. Asphalt road crack detection method based on improved DeepLabV3+ network[J]. Semiconductor Optoelectronics, 2024, 45(3):493-500. (in Chinese)
[26]	SANDLER M, HOWARD A, ZHU M L, et al. MobileNetV2:Inverted residuals and linear bottlenecks[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City,UT,US: IEEE, 2018:4510-4520.
[27]	LU H X, HONG Z Y, ZENG Z Q, et al. Improving semantic segmentation of switch connector images in DeepLabV3+[C]// Proceedings of the 2023 6th International Conference on Information Communication and Signal Processing. Xi'an,China: IEEE, 2023:90-96.
[28]	ZHANG C Y, LI M L, WEI D Z, et al. Enhanced DeepLabV3+ for urban land use classification based on UAV-borne images[C]// Proceedings of the 2022 7th International Conference on Image,Vision and Computing. Xi'an,China: IEEE, 2022:449-454.
[29]	SIFRE L, MALLAT S. Rigid-motion scattering for texture classification:arXiv:1403.1687[R]. Ithaca,NY,US: Cornell University, 2014:1403.1687.
[30]	WANG Y, YANG L, LIU X Z, et al. An improved semantic segmentation algorithm for high-resolution remote sensing images based on DeepLabV3+[J]. Scientific Reports, 2024, 14(1):9716. doi: 10.1038/s41598-024-60375-1 pmid: 38678060
[31]	NIU Z Y, ZHONG G Q, YU H. A review on the attention mechanism of deep learning[J]. Neurocomputing, 2021, 452:48-62.
[1]	余扬. 基于二维Otsu的舰船目标图像分割方法[J]. 舰船电子工程, 2022, 42(1):36-39.
	YU Y. Ship target image segmentation method based on two-dimensional Otsu[J]. Ship Electronic Engineering, 2022, 42(1):36-39. (in Chinese)
[2]	薛丽霞, 孙伟, 汪荣贵, 等. 基于边缘概率的层次交互式图像分割算法[J]. 合肥工业大学学报(自然科学版), 2020, 43(3):342-348.
	XUE L X, SUN W, WANG R G, et al. Hierarchical interactive image segmentation algorithm based on edge probability[J]. Journal of Hefei University of Technology (Natural Science Edition), 2020, 43(3):342-348. (in Chinese)
[3]	何翠萍. 应用区域生长的无人船红外图像精确分割方法[J]. 信息与电脑(理论版), 2022, 34(24):81-83.
	HE C P. Precise segmentation method for infrared images of unmanned ships using region growing[J]. Information & Computer(Theoretical Edition), 2022, 34(24):81-83. (in Chinese)
[4]	黄泽贤, 吴凡路, 傅瑶, 等. 基于深度学习的遥感图像舰船目标检测算法综述[J]. 光学精密工程, 2023, 31(15):2295-2318.
	HUANG Z X, WU F L, FU Y, et al. A review of ship target detection algorithms in remote sensing images based on deep learning[J]. Optics and Precision Engineering, 2023, 31(15):2295-2318. (in Chinese)
[5]	ARBELÁEZ P, MAIRE M, FOWLKES C, et al. Contour detection and hierarchical image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(5):898-916. doi: 10.1109/TPAMI.2010.161 pmid: 20733228
[6]	吕亚飞, 石敏, 江志浩, 等. 基于图像分割的海上舰船目标智能感知算法[J] 舰船科学技术, 2022, 44(24):91-95.
	LÜ Y F, SHI M, JIANG Z H, et al. Intelligent perception algorithm for maritime ship targets based on image segmentation[J]. Ship Science and Technology, 2022, 44(24):91-95. (in Chinese)
[7]	李琦, 王树良, 党迎旭, 等. 基于聚类分析的战场重要目标识别方法[C]// 第十届中国指挥控制大会论文集(上册). 北京: 中国指挥与控制学会, 2022:472-478.
	LI Q, WANG S L, DANG Y X, et al. Battlefield key target recognition method based on cluster analysis[C]// Proceedings of the 10th China Command and Control Conference (Volume I). Beijing: Chinese Institute of Command and Control, 2022:472-478. (in Chinese)
[8]	PARK H. α-MeanShift++:improving meanshift++ for image segmentation[J]. IEEE Access, 2021, 9:131430-131439.
[9]	GANESAN P, SATHISH B S, JOSEPH L M I L, et al. Investigation of fuzzy and SOM based segmentation of high-resolution satellite images in HSL color space for information retrieval[C]// Proceedings of the 2023 9th International Conference on Advanced Computing and Communication Systems. Coimbatore,India: IEEE, 2023, 1:420-425.
[10]	夏平, 任强, 吴涛, 等. 融合多尺度统计信息模糊C均值聚类与Markov随机场的小波域声纳图像分割[J]. 兵工学报, 2017, 38(5):940-948. doi: 10.3969/j.issn.1000-1093.2017.05.014
	XIA P, REN Q, WU T, et al. Sonar image segmentation fusion of multi-scale statistical information FCM clustering and MRF model in wavelet domain[J]. Acta Armamentarii, 2017, 38(5):940-948. (in Chinese) doi: 10.3969/j.issn.1000-1093.2017.05.014
[11]	陈松生. 基于模糊聚类的非监督学习研究[D]. 南京: 南京师范大学, 2007.
	CHEN S S. Research on unsupervised learning based on fuzzy clustering[D]. Nanjing: Nanjing Normal University, 2007. (in Chinese)
[12]	XU D X, ZHU X W, WANG Z H. Segmentation of marine targets instance based on improved mask R-CNN[C]// Proceedings of the 2023 IEEE International Conference on Mechatronics and Automation. Harbin,China: IEEE, 2023:1633-1637.
[13]	WEI C X, JI Z Y, WEI M S, et al. SCV-UNet:saliency-combined complex-valued U-Net for SAR ship target segmentation[C]// Proceedings of the 2023 IEEE International Geoscience and Remote Sensing Symposium. Washington,D.C.,US: IEEE, 2023:6948-6951.
[32]	WANG Q L, WU B G, ZHU P F, et al. ECA-net:efficient channel attention for deep convolutional neural networks:arXiv:1910.03151[R]. Ithaca,NY,US: Cornell University, 2019:1910.03151.
[33]	HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City,UT,US: IEEE, 2018:7132-7141.
[34]	PAN S J, YANG Q. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2009, 22(10):1345-1359.
[35]	LI T W, LEE G C. Performance analysis of fine-tune transferred deep learning[C]// Proceedings of the 2021 IEEE 3rd Eurasia Conference on IOT,Communication and Engineering. Yunlin,Taiwan,China: IEEE, 2021:315-319.
[36]	SHAO Z F, WU W J, WANG Z Y, et al. Seaships:a large-scale precisely annotated dataset for ship detection[J]. IEEE Transactions on Multimedia, 2018, 20(10):2593-2604.
[37]	赵为平, 陈雨, 项松, 等. 基于改进的DeepLabv3+图像语义分割算法研究[J]. 系统仿真学报, 2023, 35(11):2333-2344. doi: 10.16182/j.issn1004731x.joss.22-0690
	ZHAO W P, CHEN Y, XIANG S, et al. Research on an improved DeepLabV3+ image semantic segmentation algorithm[J]. Journal of System Simulation, 2023, 35(11):2333-2344. (in Chinese)
[38]	NI Z L, CHEN X H, ZHAI Y J, et al. Context-guided spatial feature reconstruction for efficient semantic segmentation[C]// Proceedings of the European Conference on Computer Vision. Cham,Germany: Springer, 2024:239-255.
[39]	XIE B, CAO J L, XIE J, et al. SED:a simple encoder-decoder for open-vocabulary semantic segmentation[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington,D.C.,US: IEEE, 2024:3426-3436.

层名称	输出尺寸 (B×M×H×W)	卷积核参数 (S/δ/P/D)
MobileNetV2	B×320×32×32	…
1×1卷积+ECA-Net	B×256×32×32	S=1×1/δ=1/P=0/D=1
3×3卷积+ECA-Net	B×256×32×32	S=3×3/δ=1/P=6/D=6
3×3卷积+ECA-Net	B×256×32×32	S=1×1/δ=1/P=12/D=12
3×3卷积+ECA-Net	B×256×32×32	S=1×1/δ=1/P=18/D=18
全局平均池化+ECA-Net	B×256×1×1
1×1卷积	B×256×32×32	S=1×1/δ=1/P=0/D=1
低层特征1×1卷积	B×48×128×128	S=1×1/δ=1/P=0/D=1
4倍线性上采样	B×256×128×128
特征融合	B×304×128×128
3×3卷积	B×256×128×128	S=3×3/δ=1/P=0/D=1
4倍最终线性上采样	B×2×512×512

层名称	输出尺寸 (B×M×H×W)	卷积核参数 (S/δ/P/D)
MobileNetV2	B×320×32×32	…
1×1卷积+ECA-Net	B×256×32×32	S=1×1/δ=1/P=0/D=1
3×3卷积+ECA-Net	B×256×32×32	S=3×3/δ=1/P=6/D=6
3×3卷积+ECA-Net	B×256×32×32	S=1×1/δ=1/P=12/D=12
3×3卷积+ECA-Net	B×256×32×32	S=1×1/δ=1/P=18/D=18
全局平均池化+ECA-Net	B×256×1×1
1×1卷积	B×256×32×32	S=1×1/δ=1/P=0/D=1
低层特征1×1卷积	B×48×128×128	S=1×1/δ=1/P=0/D=1
4倍线性上采样	B×256×128×128
特征融合	B×304×128×128
3×3卷积	B×256×128×128	S=3×3/δ=1/P=0/D=1
4倍最终线性上采样	B×2×512×512

Input	Operator	d	c	z	s
224×224×3	conv2d		32	1	2
112×112×32	bottleneck	1	16	1	1
112×112×16	bottleneck	6	24	2	2
56×56×24	bottleneck	6	32	3	2
28×28×32	bottleneck	6	64	4	2
14×14×64	bottleneck	6	96	3	1
14×14×96	bottleneck	6	160	3	2
7×7×160	bottleneck	6	320	1	1
7×7×320	conv2d 1×1		1280	1	1
7×7×1280	avgpool 7×7			1
1×1×v	conv2d 1×1		v

Input	Operator	d	c	z	s
224×224×3	conv2d		32	1	2
112×112×32	bottleneck	1	16	1	1
112×112×16	bottleneck	6	24	2	2
56×56×24	bottleneck	6	32	3	2
28×28×32	bottleneck	6	64	4	2
14×14×64	bottleneck	6	96	3	1
14×14×96	bottleneck	6	160	3	2
7×7×160	bottleneck	6	320	1	1
7×7×320	conv2d 1×1		1280	1	1
7×7×1280	avgpool 7×7			1
1×1×v	conv2d 1×1		v

项目	配置参数
操作系统	Windows 11 64bit
处理器	IntelⓇ Core^TM i9-14900KF@3.2GHz
显卡	NVIDIA GeForce RTX 4090 24GB
运行内存	DDR5 64G 6400MHz
开发语言	Python 3.9
深度学习框架	Pytorch 1.12.0
CUDA版本	12.8