Adaptive superpixel segmentation of SAR images using an adaptive adjustment strategy for seeds

Zhao Teng; Du Xiaoping; Yan Zhenzhen; Zhu Junjie; Xu Chen; Fan Xiangtao

doi:10.11834/jrs.20233183

Views : 0 下载量: 8 CSCD: 0 更多指标

R-PDF
PDF
Export
Share
Collection
Album

Adaptive superpixel segmentation of SAR images using an adaptive adjustment strategy for seeds
Pages: 1-12(2023)
Published Online： 16 November 2023 ，
DOI： 10.11834/jrs.20233183

扫描看全文

赵腾，杜小平，严珍珍，朱俊杰，徐琛，范湘涛.XXXX.种子点自适应调整策略下的SAR影像超像素分割.遥感学报，XX（XX）： 1-12

Zhao Teng，Du Xiaoping，Yan Zhenzhen，Zhu Junjie，Xu Chen，Fan Xiangtao. XXXX. Adaptive superpixel segmentation of SAR images using an adaptive adjustment strategy for seeds. National Remote Sensing Bulletin， XX（XX）：1-12
赵腾，杜小平，严珍珍，朱俊杰，徐琛，范湘涛.XXXX.种子点自适应调整策略下的SAR影像超像素分割.遥感学报，XX（XX）： 1-12 DOI： 10.11834/jrs.20233183.

Zhao Teng，Du Xiaoping，Yan Zhenzhen，Zhu Junjie，Xu Chen，Fan Xiangtao. XXXX. Adaptive superpixel segmentation of SAR images using an adaptive adjustment strategy for seeds. National Remote Sensing Bulletin， XX（XX）：1-12 DOI： 10.11834/jrs.20233183.

摘要

SAR影像超像素分割，即将SAR影像中相似像素按照度量准则聚合为超像素的过程，超像素能一定程度体现图像的语义特征，可有效降低后续图像理解的难度，已成为影像分类、变化检测等算法重要的预处理步骤。然而，现有的SAR影像超像素分割算法多基于局部聚类方法实现，这类方法存在超像素种子点个数预定义、缺乏影像细节自适应性和多次迭代导致的耗时过多等不足。针对上述问题，本文提出了基于邻域特性的单次迭代超像素自适应分割算法ASSA，该算法基于高斯混合模型的种子点自适应调整策略，实现了超像素个数自适应确定，并确保了超像素内部的同质性；利用优先级队列和邻域特性，实现了单次迭代下的超像素分割；同时，ASSA算法使用高斯核函数和后处理两种策略进行了SAR影像噪声抑制。本文从可视化效果、定量指标和运行时间三方面对算法的有效性和高效性进行了评估。实验结果表明，相比于其他超像素分割算法，ASSA算法能够基于影像特性实现自适应超像素分割，提高分割效率的同时生成的超像素边界贴合度和内部同质性都较高。其中边界召回率较SLIC和ESOM分别提高11.3%和15.9%，修正的欠分割错误率较SLIC和ESOM分别降低33.3%和29.4%。

Abstract

objective Superpixel segmentation offers significant advantages for information extraction from SAR images. First

it effectively reduces data volume and enhances the efficiency of subsequent applications. Second

it effectively reduces noise interference in SAR images

thereby improving data quality. Third

superpixel segmentation preserves the edge features of images

which is beneficial to the SAR image post-processing stages

such as deep learning-based classification. Lastly

the results of superpixel segmentation can be directly used as inputs for graph convolutional networks to explore the application of superpixel-based graph convolutional networks. As a result

SAR image superpixel segmentation has found extensive application in ship monitoring

water body extraction

and various other fields. Existing superpixel segmentation algorithms for SAR images predominantly rely on local clustering methods; however

they exhibit certain shortcomings including a predefined number of superpixels

limited adaptability

and the necessity for multiple iterations. To overcome these limitations

this paper proposes a novel adaptive superpixel segmentation algorithm called ASSA. This algorithm maximizes the benefits derived from Gaussian mixture models

neighborhood properties

and priority queues.Method Firstly

this paper proposes an adaptive adjustment strategy for seeds to overcome the challenges associated with predefined number of superpixels and limited adaptability. The strategy is based on Gaussian mixture models

involving seed adjustment and generation using homogeneity discrimination criteria. Secondly

the algorithm solves the issue of multiple iterations by implementing single-iteration superpixel segmentation using neighborhood properties and priority queues under the neighborhood compulsory connection. Finally

the algorithm tackles severe speckle noise in SAR images employing a Gaussian kernel function to smooth the unmarked pixels and a post-processing algorithm to eliminate isolated superpixels.Result In this paper

we used 9 sentinel-1 images to evaluate the proposed ASSA in terms of visualization effect

quantitative accuracy and runtime efficiency. The results show that

compared to existing superpixel segmentation algorithms

the proposed ASSA achieves higher boundary adherence and internal homogeneity while improving segmentation efficiency. In particular

the average boundary recall rate is improved by 11.3% and 15.9% compared to SLIC and ESOM

respectively

while the average corrected under-segmentation error rate is reduced by 33.3% and 29.4%

respectively. We also conducted experiments on the parameters of the ASSA - the minimum number of superpixels and the number of Gaussian distributions

and concluded that these parameters need to be flexibly set according to the actual scenario to achieve the optimal effect. Furthermore

we experimentally demonstrated the advantages and robustness of the ASSA’s adaptive adjustment strategy for seeds. Additionally

our comparative experiments on noise suppression strategies revealed that combining filtering and post-processing yields the optimal noise suppression effect

resulting in the minimum number of superpixels.Conclusion To address the certain shortcomings of existing superpixel segmentation algorithms for SAR images

this paper proposes a single-iteration superpixel adaptive segmentation algorithm based on neighborhood characteristics and adaptive adjustment strategy for seeds. This algorithm combines Gaussian mixture models with superpixel homogeneity discrimination to achieve adaptive segmentation. The experimental results demonstrate that the proposed ASSA algorithm is an effective and efficient method for SAR image superpixel segmentation.

关键词

SAR超像素分割优先级队列种子点自适应调整策略高斯混合模型

Keywords

SARsuperpixel segmentationpriority queueadaptive adjustment strategy for seedsGaussian mixture model

references

Achanta R, Shaji A, Smith K, Lucchi A, Fua P and Süsstrunk S. 2012. SLIC superpixels compared to state-of-the-art superpixel methods. IEEE Transactions on Pattern Analysis and Machine Intelligence 34(11): 2274-2281 [DOI: 10.1109/TPAMI.2012.120http://dx.doi.org/10.1109/TPAMI.2012.120]

Achanta R and Susstrunk S. 2017. Superpixels and Polygons using Simple Non-Iterative Clustering. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA: IEEE: 4895-4904 [DOI: 10.1109/CVPR.2017.520http://dx.doi.org/10.1109/CVPR.2017.520]

Celik T and Tjahjadi T. 2012. Automatic image equalization and contrast enhancement using Gaussian mixture modeling. IEEE Transactions on Image Processing 21(1): 145-156 [DOI: 10.1109/TIP.2011.2162419http://dx.doi.org/10.1109/TIP.2011.2162419]

Jia S C, Xue D J, Li C R, Zheng J, Li W Q. 2019. Research on water information extraction method based on Sentinel-1 data. Yangtze River, 50(02): 213-217

贾诗超,薛东剑,李成绕,郑洁,李婉秋. 2019. 基于Sentinel-1数据的水体信息提取方法研究. 人民长江, 50(02): 213-217 [DOI: 10.16232/j.cnki.1001-4179.2019.02.038http://dx.doi.org/10.16232/j.cnki.1001-4179.2019.02.038]

Jing W, Jin W and Xiang D. 2021. Edge-Aware Superpixel Generation for SAR Imagery With One Iteration Merging. IEEE Geoscience and Remote Sensing Letters 18(9): 1600-1604 [DOI: 10.1109/LGRS.2020.3005973http://dx.doi.org/10.1109/LGRS.2020.3005973]

Li M D, Cui X C and Chen S W. 2022. Adaptive Superpixel-Level CFAR Detector for SAR Inshore Dense Ship Detection. IEEE Geoscience and Remote Sensing Letters 19: 1-5 [DOI: 10.1109/LGRS.2021.3059253http://dx.doi.org/10.1109/LGRS.2021.3059253]

Li Z L，Wang L Y，Jiang S，Wu Y H and Zhang Q J. 2021. On orbit extraction method of ship target in SAR images based on ultra-lightweight network. National Remote Sensing Bulletin. Journal of Remote Sensing 25(3): 765-775

李宗凌,汪路元,蒋帅,吴雨航,张庆君. 2021. 超轻量网络的SAR图像舰船目标在轨提取. 遥感学报，25(3):765-775 [DOI: 10.11834/jrs.20210160http://dx.doi.org/10.11834/jrs.20210160]

Liu Q C, Xiao L, Yang J X and Wei Z H. 2021. CNN-Enhanced Graph Convolutional Network With Pixel- and Superpixel-Level Feature Fusion for Hyperspectral Image Classification. IEEE Transactions on Geoscience and Remote Sensing 59(10): 8657-8671 [DOI: 10.1109/TGRS.2020.3037361http://dx.doi.org/10.1109/TGRS.2020.3037361]

Lu X G, Lin Z S, Han P and Zou C. 2019. Fast detection of airport runways in PolSAR images based on adaptive unsupervised classification. Journal of Remote Sensing, 23(6): 1186–1193

卢晓光,蔺泽山,韩萍,邹璨. 2019. 自适应无监督分类的PolSAR图像机场跑道区域快速检测. 遥感学报, 23(6): 1186–1193 [DOI： 10.11834/jrs.20198384http://dx.doi.org/10.11834/jrs.20198384]

Moon T K. 1996. The expectation-maximization algorithm. IEEE Signal Processing Magazine 13(6): 47-60 [DOI: 10.1109/79.543975http://dx.doi.org/10.1109/79.543975]

Pappas O A, Anantrasirichai N, Achim A M and Adams B A. 2021. River Planform Extraction From High-Resolution SAR Images via Generalized Gamma Distribution Superpixel Classification. IEEE Transactions on Geoscience and Remote Sensing 59(5): 3942-3955 [DOI: 10.1109/TGRS.2020.3011209http://dx.doi.org/10.1109/TGRS.2020.3011209]

Qin F C, Guo J M and Lang F K. 2015. Superpixel Segmentation for Polarimetric SAR Imagery Using Local Iterative Clustering. IEEE Geoscience and Remote Sensing Letters 12(1): 13-17 [DOI: 10.1109/LGRS.2014.2322960http://dx.doi.org/10.1109/LGRS.2014.2322960]

Wang X Q, He Y, Li G, and plaza A J. 2021. Adaptive Superpixel Segmentation of Marine SAR Images by Aggregating Fisher Vectors. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 14: 2058-2069 [DOI: 10.1109/jstars.2021.3051301http://dx.doi.org/10.1109/jstars.2021.3051301]

Wang X Q, Li G, Plaza A, and He Y. 2022. Revisiting SLIC: Fast Superpixel Segmentation of Marine SAR Images Using Density Features. IEEE Transactions on Geoscience and Remote Sensing 60: 2058-2069 [DOI: 10.1109/TGRS.2022.3142068http://dx.doi.org/10.1109/TGRS.2022.3142068]

Wu F, Li J J, Zhang B, Wang C, Zhang H, Chen F L, Li L and Xu L. 2021. Inundation monitoring of immovable cultural relics with time-series SAR images. Journal of Remote Sensing 25(12): 2431-2440

吴樊,李娟娟,张波,王超,张红, 陈富龙,李璐,许璐. 2021. 时间序列SAR图像不可移动文物水域淹没监测. 遥感学报， 25（12）： 2431-2440 [DOI: 10.11834/jrs.20211146http://dx.doi.org/10.11834/jrs.20211146]

Xiang D L, Tang T, Zhao L J and Su Y. 2013. Superpixel Generating Algorithm Based on Pixel Intensity and Location Similarity for SAR Image Classification. IEEE Geoscience and Remote Sensing Letters 10(6): 1414-1418 [DOI: 10.1109/LGRS.2013.2259214http://dx.doi.org/10.1109/LGRS.2013.2259214]

Xiang D L, Wang W, Tang T and Su Y. 2017. Multiple-component polarimetric decomposition with new volume scattering models for PolSAR urban areas. IET Radar, Sonar & Navigation 11(3): 410-419 [DOI: 10.1049/IET-RSN.2016.0105http://dx.doi.org/10.1049/IET-RSN.2016.0105]

Xiao X L, Zhou Y C and Gong Y J. 2018. Content-Adaptive Superpixel Segmentation. IEEE Transactions on Image Processing 27(6): 2883-2896 [DOI: 10.1109/TIP.2018.2810541http://dx.doi.org/10.1109/TIP.2018.2810541]

Xu Q, Zhang X, Yu S H, Chen Q H and Liu X G. 2019. Multi-feature-based classification method using random forest and superpixels for polarimetric SAR images. Journal of Remote Sensing, 23(4): 685–694

徐乔,张霄,余绍淮,陈启浩,刘修国. 2019. 综合多特征的极化SAR图像随机森林分类算法. 遥感学报, 23(4): 685–694 [DOI： 10.11834/jrs.20197475http://dx.doi.org/10.11834/jrs.20197475]

Song X Y, Zhou L L, Li Z G, Chen J, Zeng L and Yan B. 2015. Review on superpixel methods in image segmentation. Journal of Image and Graphics, 20(5):599-608

宋熙煜,周利莉,李中国,陈健,曾磊,闫镔. 图像分割中的超像素方法研究综述. 中国图象图形学报. 2015. 20(5):599-608 [DOI:10.11834/jig.20150502http://dx.doi.org/10.11834/jig.20150502]

Yin J J, Wang T, Du Y L, Liu X Y, Zhou L J and Yang J. 2022. SLIC Superpixel Segmentation for Polarimetric SAR Images. IEEE Transactions on Geoscience and Remote Sensing 60: 1-17 [DOI: 10.1109/TGRS.2020.3047126http://dx.doi.org/10.1109/TGRS.2020.3047126]

Zhang JL and Wang B. 2017. SAR image change detection method of DSSRM based on cascade segmentation. Journal of Remote Sensing 21(4): 614-621

张建龙, 王斌. 2017. DSSRM级联分割的SAR图像变化检测. 遥感学报, 21(4): 614–621 [DOI: 10.11834/jrs.20176330http://dx.doi.org/10.11834/jrs.20176330]

Alert me when the article has been cited

提交

Flood monitoring and analysis based on time-series SAR image for complex area

SAR ship detection through generative knowledge transfer

Hyperspectral unmixing network considering spectral information and superpixel segmentation

Hyperspectral anomaly detection via combining adaptive window saliency detection and improved superpixel segmentation

Research progress on radar remote sensing for rice growth monitoring