Adaptive Hybrid Conditional Random Field Model for SAR Image Segmentation

Wang, Fan; Wu, Yan; Li, Ming; Zhang, Peng; Zhang, Qingjun

doi:10.1109/tgrs.2016.2611060

Cited by 40 publications

(12 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among the supervised methods, training and prediction through graph model and classifier are the most important manners [2], [3], [12], [13], [30]- [35]. These methods use superpixel as basic processing unit, which significantly reduces the computational complexity of graph model and classifier during the training.…”

Section: ) Supervised Methodsmentioning

confidence: 99%

“…Image segmentation using superpixel as basic processing unit has been an important image segmentation manner recently. In this section, we review them from unsupervised methods [1], [4], [11], [26]- [29] and supervised methods [2], [3], [12], [13], [30]- [37].…”

Section: B Superpixel-based Image Segmentationmentioning

confidence: 99%

“…In addition, supervised methods need to construct a classifier for training and learning. For example, Liu et al [3] exploit large-margin framework and deep features generated by trained CNNs [12] for Conditional Random Fields (CRFs) learning respectively; Wang et al [13] utilize a multinomial logistic regression classifier for parameter training. However, supervised methods have a strong dependence of pre-trained models, which require a large number of pixel-level labels for training.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Automatic Image Segmentation With Superpixels and Image-Level Labels

Xie

et al. 2019

IEEE Access

View full text Add to dashboard Cite

Automatically and ideally segmenting the semantic region of each object in an image will greatly improve the precision and efficiency of subsequent image processing. We propose an automatic image segmentation algorithm based on superpixels and image-level labels. The proposed algorithm consists of three stages. At the stage of superpixel segmentation, we adaptively generate the initial number of superpixels using the minimum spatial distance and the total number of pixels in the image. At the stage of superpixel merging, we define small superpixels and directly merge the most similar superpixel pairs without considering the adjacency, until the number of superpixels equals the number of groupings contained in image-level labels. Furthermore, we add a stage of reclassification of disconnected regions after superpixel merging to enhance the connectivity of segmented regions. On the widely used Microsoft Research Cambridge data set and Berkeley segmentation data set, we demonstrate that our algorithm can produce high-precision image segmentation results compared with the state-of-the-art algorithms.

show abstract

Section: ) Supervised Methodsmentioning

confidence: 99%

Section: B Superpixel-based Image Segmentationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automatic Image Segmentation With Superpixels and Image-Level Labels

Xie

et al. 2019

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The definitions of them can be described as follows. If there are m solutions in the current population, the values of the two indices of t-th solution can be defined by equation 3and (4).…”

Section: Definitions Of Multi-objective Optimizationmentioning

confidence: 99%

“…SAR image segmentation has been the subject of widely research since the invention of SAR imaging. Up to now, a multitude of methods have been proposed, which can be classified into clustering-based methods [1], graph-partitioning methods [2], morphologic methods [3], and model-based methods [4]. The paper mainly concerns the methods of clustering-based SAR image segmentation.…”

Section: Introductionmentioning

confidence: 99%

An Automatic SAR Image Segmentation Framework by Multi-objective Clustering and Artificial Immune Learning

Yang¹,

Zhang²,

Lv³

et al. 2018

Proceedings of the 2018 International Conference on Mathematics, Modelling, Simulation and Algorithms (MMSA 2018)

View full text Add to dashboard Cite

Though several algorithms inspired by theoretical immunology have been applied to the domain of pattern classification, little focus has been placed on the issues that simultaneously optimize more than one objective-functions. Here, an efficient multi-objective automatic segmentation framework (MASF) is formulated and applied to SAR image unsupervised classification. In the framework, four important issues are presented: 1) two reasonable image preprocessing techniques are discussed at the initial stage; 2)then, an efficient immune multiobjective optimization algorithm is proposed; 3) besides, a locusbased adjacency representation in individual encoding is introduced; 4) two very simple, but very efficient conflicting clustering validity indices are incorporated into the framework and simultaneously optimized. Both simulated data and real images are used to quantitatively validate its effectiveness. In addition, four other state-of-the-art image segmentation methods are employed for comparison. Experimental results show that the proposed framework is efficient and effective for SAR image segmentation.

show abstract