https://scholar.google.com/citations?hl=en&user=7QwnQC0AAAAJ&view_op=list_works&authuser=4&gmla=AH70aAXSgsGfbihg4XfTuewCeQeYGy1HTwvT72Ir9iHrnZEDh1XFE7EzcqgkFv5kr1vS-lIMrz6MeOglUi59DhKE

Document Type : Original Research Paper

Authors

1 Department of Surveying Engineering, Faculty of Engineering, University of Zabol, Zabol, Iran

2 Department of Civil Engineering, Faculty of Engineering, University of Birjand, Birjand, Iran

Abstract

target detection. The most common method for target detection in satellite images is pixel-based detection, in which each pixel is assigned to a specific class with only its spectral information and without considering neighboring pixels. With recent advances and the creation of images with high spatial resolution, it is necessary to use both spectral and spatial information to detect hyperspectral images. This research deals with the detection of roofs with special coverage as a target, in an urban environment, through a series of hyperspectral images. Since an urban environment has complex characteristics in terms of physics, geometry and elements used in buildings, hyperspectral data effectively helps to identify, extract and produce a map of the elements that make up an urban environment. Identifying the type of roof of buildings in urban environments is very important in various applications, such as mobile phone communications, virtual reality, architecture and urban modeling, planning, and city management.
Methods: In this research, the spatial information strategy is investigated along with the spectral information to improve target detection in the analysis of hyperspectral images. For this purpose, the spectral-spatial algorithm of marker-based minimum spanning forest, which is used in the image classification process, is used to detect the roofs of buildings with special coverage. The markers were selected from the support vector machine classification map in the proposed method. For this purpose, the analysis of the labeling of the connected components was done based on 8 neighboring pixels. The minimum spanning forest is obtained after creating the minimum spanning tree and removing the ridges related to the added vertex in the last step. In the minimum spanning forest algorithm, each tree grows on one of the vertices of the image, and by assigning the class of each marker to all the pixels grown from it, a spectral-spatial detection map is obtained.
Findings: The above techniques were applied on a series of CASI sensor image data taken from the urban area of ​​Toulouse located in the south of France. The results of quantitative and qualitative evaluations show that the proposed method has improved the value of the Kappa coefficient by 38% in comparison with the spectral angle measurement detection algorithm. This shows the importance of using spatial information in the detection process, while the spectral angle measurement algorithm only needs the spectral information of the desired target for detection.
Conclusion: Simultaneously with the growth of urbanization and the development of urban areas, the need of managers and planners for very accurate maps of urban areas has increased significantly. The use of spatial information, especially in the case of images taken from urban areas where several adjacent pixels belong to the same class or complex, can improve detection accuracy. It is intended to reduce the amount of error in the spectral-spatial detection of the target in the future research. The conditions of creating mixed pixels, such as the overlap of terrestrial phenomena and the heterogeneity of most phenomena, and as a result, the increase of the internal variance of the target, increase the detection error in hyperspectral images. Therefore, it is tried to reduce the above errors by using different methods.

Keywords

Main Subjects

COPYRIGHTS 
© 2024 The Author(s).  This is an open-access article distributed under the terms and conditions of the Creative Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) (https://creativecommons.org/licenses/by-nc/4.0/)  

[9] Carvalho, O.A., & Meneses, P.R. (2002). Spectral Correlation Mapper (SCM): An Improvement on the Spectral Angle Mapper (SAM), Asa Norte, 70910-900, Brasília, DF, Brasil, 1-9.
http://dx.doi.org/10.1007/978-1-4419-9170-6
https://doi.org/10.1109/21.256541
[14] Frolov, D., & Smith, R.B. (1999). Locally Adaptive Constrained Energy minimization for AVIRIS image, Eighth JPL Airborne Earth Science (AVIRS), http://www.microimages.com/papers/.
https://doi.org/10.1672/08-194.1
[24] Fauvel, M. (2007). Spectral and Spatial Methods for the Classification of Urban Remote Sensing Data, PhD thesis, Grenoble Institute of Technology, 1-180.
[27] Jain, R., Kasturi, R., & Schunck, B.G. (1995). Machine Vision. McGraw-Hill series in Computer Science., McGraw-Hill, Inc, 1-549.
[28] Soille, P. (2006). Morphological Image Analysis, IEEE Transactions on Pattern Analysis and Machine Intelligence, 28(5): 673-683. https://doi.org/10.1109/TPAMI.2006.99
[31] Homayouni, S., & Roux, M. (2005). Hyperspectral image analysis for material mapping using spectral matching, ISPRS04-Istanbul, GET, Telecom Paris, UMR 5141 LTCI, Department TSI, 46 rue Barrault, 75013 Paris, France.
[36] Hou, Y., Zhang, Y., Yao, L., Liu, X., & Wang, F. (2016). Mineral target detection based on MSCPE_BSE in hyperspectral image, the IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Beijing, China, 1614–1617. https://doi.org/10.1109/IGARSS.2016.7729412
[38] Akbari, D., Saadatseresht, M., & Homayouni, S. (2008a). Hyperspectral Detection Improvement by using Spectral-Spatial feature, the Map Asia 2008 Conference, Kuala Lumpur, Malaysia.
[40] Nasrabadi, N.M. (2008). Regularized Spectral Matched Filter for Target Recognition in Hyperspectral Imagery, SPLetters, 1: 317-320.
https://doi.org/10.1007/978-1-4757-3264-1
[43] Homayouni, S., & Roux, M. (2003). Material Mapping from Hyperspectral Images using Spectral Matching in Urban Area, IEEE Workshop on Advances in Techniques for analysis of Remotely Sensed Data, NASA Goddard center, Washington DC, USA.