SAR Image Classification Based on CRFs With Integration of Local Label Context and Pairwise Label Compatibility

Ding, Yongke; Li, Yuanxiang; Wang, Yu

doi:10.1109/jstars.2013.2262038

Cited by 22 publications

(8 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The SVM classifier takes the simple but efficient features (i.e., backscattering intensity, texture, and supertexture) proposed in [1] as its inputs and outputs the land cover types of the SAR images. Optimization of the training set from grid labeling is implemented by reweighting strategy of the boosting algorithm [17,18], which has been improved by us through introducing a new penalty term for label proportions.…”

Section: The Learning Model With Label Proportionsmentioning

confidence: 99%

“…For classification, we consider four classes for Tianjin and Rosenheim areas as in [1], i.e., UA, WL, OA, and WB. We use about 10% of all the cells coming from slicing the original SAR images for training the SVM model by random selection strategy and the whole images for classification using SVM, which is a common practice in remote sensing image classification [1,23]. The number of training samples for each label is proportional to the label distributions in the image to avoid class imbalance.…”

Section: Dataset Description and Experiments Settingmentioning

confidence: 99%

“…Synthetic aperture radar (SAR) image classification or land use/land cover (LULC) mapping plays an important role in many and diverse SAR applications [1]. Supervised learning is one of the most popular methods for SAR image classification.…”

Section: Introductionmentioning

confidence: 99%

“…Due to the SAR imaging mechanism, the complex backscattering effects of objects and scenes in such HR SAR images will lead to local noises, double reflection, and even object deformation and part missing, making sample labeling highly difficult. So, researchers need to label the training samples more precisely and more carefully and even need the support of professional interpreters, field surveys, or other relevant information, *Correspondence: yuanxli@sjtu.edu.cn 1 Center for Advanced Sensing Technology, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai 200240, China Full list of author information is available at the end of the article such as ancillary multispectral, geographic information system (GIS), or elevation data [3,4]. The cost and difficulty of sample labeling have highly restricted the applications of large amounts of available HR SAR data, even though the data contains rich details of objects and scenes.…”

Section: Introductionmentioning

confidence: 99%

“…Incorporation of label proportions and reweighting strategy helps to enhance the robustness of the SVM classifier with the existing uncertain (partially mislabeled) samples from grid labeling. To evaluate our method, we consider HR TerraSAR-X images of Tianjin and Rosenheim with four LULC classes [1]: urban area (UA), woodland (WL), open area (OA, such as farmland, grassland, bare soil, etc. ), and water body (WB).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Learning from label proportions for SAR image classification

Ding

Wang

2017

EURASIP J. Adv. Signal Process.

Self Cite

View full text Add to dashboard Cite

Synthetic aperture radar (SAR) image classification plays a key role in SAR interpretation. Due to the cost and difficulty of truth labeling for SAR images, the newly labeled samples available for image classification are very limited. This paper focuses on defining a new sample labeling method to solve the problem of truth acquisition for training data in SAR image classification. An efficient classification framework for high-resolution SAR images is presented in this paper, which is built on learning from uncertain labels. We use grid labeling for rapid training data acquisition by assigning a label to a group of neighboring pixels at a time. A novel SVM-based learning model is proposed to optimize the uncertain training data within the constraints of label proportions in each group and then predict the label of each sample for the test data based on the optimized training set. This work intends to explore a rapid labeling method called grid labeling for efficient training set definition and apply it to large-scale SAR image classification. The model demonstrates good performance in both accuracy and efficiency for scene interpretation of high-resolution SAR images.

show abstract

Section: The Learning Model With Label Proportionsmentioning

confidence: 99%

Section: Dataset Description and Experiments Settingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Learning from label proportions for SAR image classification

Ding

Wang

2017

EURASIP J. Adv. Signal Process.

Self Cite

View full text Add to dashboard Cite

show abstract

Certain investigations on image segmentation algorithms on synthetic aperture radar images and classification using convolution neural network

Srinitya¹,

Sharmila

Logeswari³

et al. 2021

Concurrency and Computation

View full text Add to dashboard Cite

Synthetic aperture radar (SAR) image segmentation is the most important step in detecting the land and sea areas in many applications like military in particular. This article experiments four prominent image segmentation algorithms on SAR images and we propose a hybrid method to segment SAR images, which shows better efficiency.First among them is OTSU thresholding method that segmented the image with less interest in the texture component. Further, adaptive thresholding, artificial bee colony segmentation, and wavelet transforms were experimented; wavelet transforms give better segmentation results compared to thresholding methods. In this article, wavelet along with masking threshold including segmentation method is proposed, which gives a better result based on texture and executes in almost the same time as other methods and the segmented images give better classification results using convolution neural network when compared to other methods.

show abstract