AUTHORS: Maria C. Proença, Jorge M. Marques

Download as PDF

ABSTRACT: Automatic ship detection in synthetic aperture radar (SAR) images is a promising subject for maritime surveillance that exploits the characteristics inherent to SAR technology. Since the radars are active devices that do not depend on sunlight and use wavelengths with high atmospheric penetration, the acquisition of images day and night and in virtually all weather conditions becomes possible. Sentinel-1A is the first of a constellation of two European Synthetic Aperture Radar satellites, and it is operational since 3 October 2014. The data is provided free of charges by the European Space Agency (ESA) and the European Commission within the Copernicus Program. The data set used to implement and test this algorithm consists in 32 images of the same area, near the port of Singapore, acquired between October 2014 and January 2017. This is a harbor known by its intense commercial traffic, where there is also high activity of small fishing boats, and a constant presence of large floating platforms, probably involved in drainage and maintenance of the port bathymetry. The algorithm is based on a wavelet filter variation that can provide a very smooth image, reducing speckle effects and allowing a precise detection of the targets. It can be used in small computers and take seconds to process large areas. The accuracy of the results makes possible the use of morphological criteria to filter the candidates detected, in addition to the usual radiometric criteria. The quality of the initial detection of targets is compared with the detection achieved with two of the most popular algorithms used in ship detection with similar processing times, SUMO and LM. The concept of false positives does not apply, as all the candidates are (most probably) ships, but of dimensions not relevant in this case study. Each image has its own land/sea mask, defining the area of interest (AOI) to work, because there are small variations in the field of view of the SAR, as well as in the image itself: a few images include strips of saturated pixels where discrimination is not possible, and this areas are eliminated from the AOI to save processing resources. This work intends to make available a quick and efficient algorithm implemented in Matlab for SENTINEL-1 images of level 1 with a pixel nearly square of 10 m, directly downloaded from ESA site (https://scihub.copernicus.eu/dhus/#/home), which can be used at any facility with usual informatics support. Visual interpretation was used to validate the results; validation based on ground truth data acquired by VMS/AIS installed in the considered ships and harbor was not in reach for this amount of data and time spanning.

KEYWORDS: Automatic maritime surveillance, ship detection, maritime traffic, maritime security

REFERENCES:

[1] D. J. Crisp, The state-of-the-art in ship detection in synthetic aperture radar imagery, DSTO Information Sciences Laboratory, DSTO–RR–0272, 2004.

[2] W. Juan, S. Lijie and Z. Xuelan, Study evolution of ship target detection and recognition in SAR imagery, Proceedings of the 2009 International Symposium on Information Processing (ISIP’09), P. R. China, 2009, pp. 147-150.

[3] A. Marino, M. J. Sanjuan-Ferrer, I. Hajnsek, K. Ouchi, Ship detection with spectral analysis of synthetic aperture radar: a comparison of new and well-known algorithms, Remote Sensing, Vol. 7, No.5, 2015, pp. 5416-5439.

[4] H. Greidanus, P. Clayton, M. Indregard, G. Staples, N. Suzuki, P. Vachoir, C. Wackerman, T. Tennvassas, J. Mallorqui, N. Kourti, R. Ringrose, H. Melief, Benchmarking operational SAR ship detection, Proceedings of International Geoscience and Remote Sensing Symposium ( IGARSS 2004), Vol. 6, 2004, pp. 4215-4218, USA.

[5] T. N. Arnesen and R. B. Olsen, Literature review on vessel detection, FFI/RAPPORT2004/02619, Norwegian Defence Research Establishment, Kjeller, Norway, 2004.

[6] H. Greidanus and N. Kourti, Findings of the DECLIMS project – detection and classification of marine traffic from space, Proceedings of SEASAR 2006, Italy, 2006.

[7] M. Tello, C. López-Martínez, J. Mallorqui, A novel algorithm for ship detection in SAR imagery based on the wavelet transform, IEEE Geoscience and Remote Sensing Letters, Vol. 2, No. 2, 2005.

[8] M. Tello, C. López-Martínez, J. Mallorqui, Automatic vessel monitoring with single and multidimensional SAR images in the wavelet domain, ISPRS Journal of Photogrammetry and Remote Sensing, Vol. 61(3–4), 2006, pp. 260-278.

[9] F. Meyer and S. Hinz, Automatic ship detection in space-borne SAR imagery, ISPRS Proceedings, Hannover, 2009.

[10] https://sentinels.copernicus.eu/documents/2479 04/685163/Sentinel-1 User_Handbook (accessed 05-04-2017).

[11] https://sentinel.esa.int/documents/247904/3494 49/S1_SP-1322_1.pdf (accessed 05-04-2017).

[12] S. Mallat, A theory of multiresolution signal decomposition: the wavelet representation, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 11, No. 7, 1989, pp. 674 – 693.

[13] I. Daubechies, The wavelet transform, timefrequency localization and signal analysis, IEEE Transactions on Information Theory, Vol. 36, No. 5, 1990, pp. 961-1005.

[14] P. Soille, Morphological image analysis: principles and applications. Springer-Verlag, pp. 170-171, 1999.

[15] W. K. Pratt, Digital image processing. Third edition, chap. 14, Ed. J. Wiley & Sons, 2001.

[16] https://www.google.pt/maps/place/Sudong+Isla nd/@1.2224565,103.6231072,1978m/data=!3m 1!1e3!4m5!3m4!1s0x31da02d055d51961:0x90 c04c4cc7de2776!8m2!3d1.207916!4d103.7202 285?hl=pt-PT (accessed 12-04-2017).

[17] L. Gagnon, H. Oppenheim and P. Valin, R&D Activities in Airborne SAR Image Processing/Analysis at Lockheed Martin Canada, Internacional Conf. on Applications of Photonic Technology III, 1998, pp. 998-1003.