A SAR Image Segmentation Method based on Otsu and Level Set

Yu, Jinao; Wu, Min; Li, Changqing; Zhu, Shibing

doi:10.1109/itaic49862.2020.9339053

Cited by 4 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on the high amplitude of the layover area, a threshold is set to segment the SAR image. The efficiency and outcome of segmentation show slight variation depending on the threshold and the image segmentation algorithm applied, such as the unsupervised image segmentation algorithm suggested by He [16], the CFAR algorithm adopted by Qin [17], and the OSTU method applied by Yu [18], etc. Besides, Prati et al [19] demonstrated that slopes could cause spectral shift of SAR interferogram, which constitutes the theoretical basis for the layover detection based on local frequency.…”

Section: Introductionmentioning

confidence: 99%

Multibaseline InSAR Layover Detection Based on Local Frequency and Eigenvalue

Wang

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

Layover detection has long been the focus of attention for Interferometric Synthetic Aperture Radar (InSAR) data processing. As for the existing layover detection methods, most of them are applied to single-baseline scenarios. Since multi-baseline InSAR provides more image samples, the performance of layover detection can be further improved. In this paper, a joint detection method of layover is proposed on the basis of local frequency and eigenvalue of multi-baseline InSAR. To obtain a precise local frequency, maximum likelihood estimation is performed for flattened interferometric phase fusion and removing the reference terrain phase is chosen to ensure the proportionality of flattened phase to the baseline length. Adaptive image selection based on coherence is proposed to obtain a distinct eigenvalue. Then, the joint detection method is elaborated on. Based on the results of local frequency estimation and eigenvalue decomposition, precise thresholds are set for joint detection, and joint judgment is made by combining the layover detected by local frequency, the layover detected by eigenvalue and the thresholds. Finally, the proposed method is validated through comparative experiments on simulated data and real data. Both theoretical analysis and experimental results show the feasibility and superiority of the proposed method.

show abstract