Ting Pan

Ms. Ting Pan

orcid

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
6 December 2019 in Remote Sensing
Verified
Journal Article
Open access
PDF file
JWJinwang Wang
Jinwang Wang
JDJian Ding
Jian Ding
HGHaowen Guo
Haowen Guo
WCWensheng Cheng
Wensheng Cheng
TPTing Pan
Ting Pan
Wen YangWen Yang
Wen Yang
Object detection in aerial images is a fundamental yet challenging task in remote sensing field. As most objects in aerial images are in arbitrary orientations, oriented bounding boxes (OBBs) have a great superiority compared with traditional horizontal bounding boxes (HBBs). However, the regression-based OBB detection methods always suffer from ambiguity in the definition of learning targets, which will decrease the detection accuracy. In this paper, we provide a comprehensive analysis of OBB representations and cast the OBB regression as a pixel-level classification problem, which can largely eliminate the ambiguity. The predicted masks are subsequently used to generate OBBs. To handle huge scale changes of objects in aerial images, an Inception Lateral Connection Network (ILCN) is utilized to enhance the Feature Pyramid Network (FPN). Furthermore, a Semantic Attention Network (SAN) is adopted to provide the semantic feature, which can help distinguish the object of interest from the cluttered background effectively. Empirical studies show that the entire method is simple yet efficient. Experimental results on two widely used datasets, i.e., DOTA and HRSC2016, demonstrate that the proposed method outperforms state-of-the-art methods.

14 October 2019 in Remote Sensing
Verified
Journal Article
Open access
PDF file
TPTing Pan
Ting Pan
DPDong Peng
Dong Peng
Wen YangWen Yang
Wen Yang
HLHeng-Chao Li
Heng-Chao Li
Despeckling is a longstanding topic in synthetic aperture radar (SAR) images. Recently, many convolutional neural network (CNN) based methods have been proposed and shown state-of-the-art performance for SAR despeckling problem. However, these CNN based methods always need many training data or can only deal with specific noise level. To solve these problems, we directly embed an efficient CNN pre-trained model for additive white Gaussian noise (AWGN) with Multi-channel Logarithm with Gaussian denoising (MuLoG) algorithm to deal with the multiplicative noise in SAR images. This flexible pre-trained CNN model takes the noise level as input, thus only a single pre-trained model is needed to deal with different noise levels. We also use a detector to find the homogeneous region automatically to estimate the noise level of image as input. Embedded with MuLoG, our proposed filter can despeckle not only single channel but also multi-channel SAR images. Finally, both simulated and real (Pol)SAR images were tested in experiments, and the results show that the proposed method has better and more robust performance than others.

17 September 2019 in Remote Sensing
Verified
Journal Article
Open access
PDF file
DPDong Peng
Dong Peng
TPTing Pan
Ting Pan
Wen YangWen Yang
Wen Yang
HLHeng-Chao Li
Heng-Chao Li
In this paper, we present a novel method for change-pattern mining in Synthetic Aperture Radar (SAR) image time series based on a distance matrix clustering algorithm, called K-Matrix. As it is different from the state-of-the-art methods, which analyze the SAR image time series based on the change detection matrix (CDM), here, we directly use the distance matrix to determine changed pixels and extract change patterns. The proposed scheme involves two steps: change detection in SAR image time series and change-pattern discovery. First, these distance matrices are constructed for each spatial position over the time series by a dissimilarity measurement. The changed pixels are detected by using a thresholding algorithm on the energy feature map of all distance matrices. Then, according to the change detection results in SAR image time series, the changed areas for pattern mining are determined. Finally, the proposed K-Matrix algorithm which clusters distance matrices by the matrix cross-correlation similarity is used to group all changed pixels into different change patterns. Experimental results on two datasets of TerraSAR-X image time series illustrate the effectiveness of the proposed method.

20 September 2019 in The Journal of Engineering
Verified
Journal Article
Open access
PDF file
TPTing Pan
Ting Pan
DPDong Peng
Dong Peng
XYXiangli Yang
Xiangli Yang
PHPingping Huang
Pingping Huang
WYWen Yang
Wen Yang
(Dis)similarity measures play an important role in the interpretation of polarimetric synthetic aperture radar (PolSAR) images. Here, the authors introduce a kind of similarity measures for PolSAR images based on the concepts of Hölder pseudo‐divergence and Hölder divergence. Authors’ similarity measures are more generalised as the derived formulas indicate that they contain several widely used measures, such as Bartlett or Bhattahcharyya distances and Chernoff distance. Also, their similarity measures are derived from the complex Wishart distribution, so these measures are good at quantifying the similarity between two covariance matrices and perform well while dealing with classification problems for PolSAR images. Experimental results on unsupervised and supervised classification of PolSAR images also verify the effectiveness of their measures.

1 January 2017
Verified
Conference Paper
Open access
PDF file
BCBao-Lei Cheng
Bao-Lei Cheng
TPTing Pan
Ting Pan
JFJian-Xi Fan
Jian-Xi Fan
JZJing-Ya Zhou
Jing-Ya Zhou
Zhao LiuZhao Liu
Zhao Liu