LiDAR sensörünün hiperspektral verilerden gölgelik alan çıkarımı başarımına etkisi

Öz

Hiperspektral görüntülerin analiziyle, tek bir pikselden o pikseldeki materyalin ne olduğu anlaşılabilmektedir. Bu özelliğiyle hiperspektral görüntüleme, yeryüzünün uzaktan algılanmasını gerektiren jeoloji, zirai ve askeri alanlarda, özellikle sınıflandırma ve hedef tespiti uygulamalarında tercih edilen bir yöntem olmaktadır. Ancak, hiperspektral görüntülemede gölgelik alanlarda kalan hedeflerden çok az miktarda foton yansımakta, bu yüzden de toplanan spektral verilerin genlikleri çok düşük düzeylerde kalmaktadır. Bu durum, gölgede kalan hedefin bulunamamasına neden olabilmektedir. Özellikle yüksekliğin sık değiştiği yerleşim yerlerinden alınan verilerde, gölgelik alanların etkisinin sınıflandırma başarımına etkisi katlanarak artmaktadır. Bu çalışmada, gölge tespiti yapan iki algoritma geliştirilmiş ve karşılaştırılmıştır. İlk yöntemde, hiperspektral verilere ek olarak LiDAR sensöründen alınan veriler de kullanılmıştır. LiDAR verilerinden gölge tespiti amacıyla, hiperspektral verilerin toplanma anındaki güneşin açılarını ve ortamdaki yükseltileri dikkate alan bir görüş hattı algoritması geliştirilmiştir. İkinci yöntemde ise, gölgeler sadece hiperspektral veriler kullanılarak tespit edilmiştir. Öncelikle bir referans imza oluşturulmuş, sonra diğer pikseller bu referans imzaya olan uzaklıklarına göre sınıflandırılmıştır. Sonuçta, hiperspektral görüntüdeki gölge alanlar tespit edilmiş ve iki yöntemin eşleşme sonuçları ve güvenilirlikleri karşılaştırılmıştır.

Anahtar Kelimeler

• LiDAR,Hiperspektral,Gölgelik alanlar

Effect of LiDAR sensor on the success of shadow detection from hyperspectral data

Abstract

With the analysis of hyperspectral images, it is possible to understand the underlying material from a single pixel. Due to this characteristics, hyperspectral imaging (HSI) is becoming a preferred method in geology, agriculture and defense fields which require the remote sensing of the environment for the purposes of classification and target detection. However, in HSI images, there are only a few photons that get reflected from areas that are under shadow. Hence, the amplitudes of the spectral signals received from shadow areas are very small, which leads to tremendous difficulties in target detection in shadowy areas. These difficulties become much more pronounceable in areas with varying elevations. In this study, we developed two methods to find the shadow regions in hyperspectral images and compared their results. The first method, line-of-sight, uses an external sensor, the Light Detection and Ranging (LiDAR), which provides elevation information. We use the LiDAR data and detect the shadows at the time of the hyperspectral data collection. Then we match the shadows to the hyperspectral image using UTM coordinates. The second method uses only the hyperspectral data and compares each pixel to a pre-determined shadow signature to arrive at a shadow/non-shadow decision. Comparison of both methods gives insight into the reliability of both methods and allows to better deal with the shadows in hyperspectral data.

Keywords

• LiDAR,Hyperspectral,Shadowy areas,Line of sight

References

1. Rencz AN. Remote Sensing for the Earth Sciences: Manual of Remote Sensing. 3rd ed. New York, USA, John Wiley and Sons, 1999.
2. Clark RN. Spectroscopy of rocks and minerals, and principles of spectroscopy. Editors: Rencz AN. Manual of Remote Sensing, 3-58, New York, USA, John Wiley and Sons, 1999.
3. Bati E, Akın Ç, Koz A, Alatan A. "Hyperspectral anomaly detection method based on auto-encoder". Proc. SPIE 9643, Image and Signal Processing for Remote Sensing XXI, 96430N, Toulouse, France, 15 October 2015.
4. Ertürk A, Çeşmeci D, Gerçek D, Güllü MK, Ertürk S. "Integrating anomaly detection to spatial preprocessing for endmember extraction of hyperspectral images". IEEE International Geoscience and Remote Sensing Symposium-IGARSS, Melbourne, Australia, 21-26 July 2013
5. Bi̇lgi̇ AS, Durmuş E, Kalkan H, Ortaç G, Taşdemir K. "An automated system for detecting the infected figs by hyperspectral image analysis". IEEE 23rd Signal Processing and Communications Applications Conference, Malatya, Turkey, 16-19 May 2015.
6. Alam MS, Elbakary MI, Aslan MS. "Object detection in hyperspectral imagery by using K-means clustering algorithm with pre-processing". Proc. SPIE 6574, Optical Pattern Recognition XVIII, Orlando, Florida, United States, 9 April 2007.
7. Omruuzun F, Baskurt DO, Daglayan H, Cetin YY. "Shadow removal from VNIR hyperspectral remote sensing imagery with endmember signature analysis". Proc. SPIE 9482, Next-Generation Spectroscopic Technologies VIII, 94821F, Baltimore, Maryland, United States, 3 June 2015.
8. Zhang Q, Pauca P, Plemmons RJ, Nikic DD. “Detecting objects under shadows by fusion of hyperspectral and LiDAR data: A physical model approach”. 5th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Gainesville, FL, USA, 26-28 June 2013.

9. Sakarya U, Demirkesen C, Teke M. "Unsharp masking filter based shadow-invariant feature extraction for hyperspectral signatures". IEEE 22nd Signal Processing and Communications Applications Conference (SIU), Trabzon, Turkey, 23-25 April 2014.
10. Li Y, Gong P, Sasagawa T. “Integrated shadow removal based on photogrammetry and image analysis”. International Journal of Remote Sensing, 26(18), 3911-3929, 2005.
11. Dare PM. “Shadow analysis in high-resolution satellite imagery of urban areas”. Photogrammetric Engineering & Remote Sensing, 71(2), 169-177, 2005.
12. Dalponte M, Bruzzone L, Gianelle D. “Fusion of hyperspectral and LIDAR remote sensing data for classification of complex forest areas”. IEEE Geoscience and Remote Sensing, 46(5), 1416-1427, 2008.
13. Demirkesen C, Teke M, Sakarya U. "Hyperspectral images and lidar based DEM fusion: A multi-modal landuse classification strategy". IEEE Geoscience and Remote Sensing Symposium, Quebec City, Canada, 13-18 July 2014.
14. Shimoni M, Tolt G, Perneel C, Ahlberg J. “Detection of vehicles in shadow areas”. IEEE 3rd Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Lisbon, Portugal, 6-9 June 2011.
15. Salvador E, Cavallaro A, Ebrahimi T. “Shadow identification and classification using invariant color models”. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP’01), Salt Lake City, UT, USA, 7-11 May 2001.
16. Sarabandi P, Yamazaki F, Matsuoka M, Kiremidjian A. “Shadow detection and radiometric restoration in satellite high resolution images”. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Anchorage, AK, USA, 20-24 September 2004.
17. Tolt G, Shimoni M, Ahlberg J. “A shadow detection method for remote sensing images using VHR hyperspectral and LIDAR data”. IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Vancouver, BC, Canada, 24-29 July 2011.
18. Jiang C, Ward O. “Shadow segmentation and classification in a constrained environment”. CVGIP: Image Understanding, 59(2), 213-225, 1994.
19. Ientilucci EJ. “Leveraging LiDAR data to aid in hyperspectral image target detection in the radiance domain”. Proc. SPIE 8390, Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XVIII, Baltimore, Maryland, United States, 9 May 2012.
20. Rochester Institute of Technology. “SHARE2012 SpecTIR Hyperspectral Airborne Experiment 2012”. http://www.rit.edu/cos/share2012/, (01.03. 2015).
21. Rochester Institute of Technology. “SHARE2012 SpecTIR”. http://www.rit.edu/cos/share2012/spectir.php, (01.03. 2015).
22. SpecTIR-LLC. “SRS Project Report”. Rochester Institute of Technology, Rochester, New York, USA, Scientific Report, 1538, 2012.
23. Rochester Institute of Technology. “SHARE2012 LIDAR”. http://www.rit.edu/cos/share2012/lidar.php (01.03. 2015).
24. Boyacı M, Yuksel SE. “Locating the shadow regions in LIDAR data: Results on the SHARE 2012 dataset”. SPIE Defense and Security: Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XXI, Baltimore, ABD, 21 May 2015.