Exploiting IKONOS and Hyperion data fusion for automated road extraction

Abstract

We present here the examples that show how fusing data from hyperspectral sensors with data from high spatial resolution sensors can enhance overall road detection accuracy. The fusion of hyperspectral and high spatial resolution data combines their superior respective spectral and spatial information. IKONOS (MSS) and Hyperion images were fused using the principal component analysis (PCA) method. The approach for road extraction integrates multiresolution segmentation and object oriented classification. Road extraction is done from an IKONOS (MSS) image and a Hyperion and IKONOS (MSS) merged image and comparisons are made depending on accuracy and quality measures such as completeness and correctness. This article also emphasises the types of roads which are giving better accuracy of extraction after fusion with hyperspectral image. This can vary because of types of material and condition of roads. The methodology was applied on roads of Dehradun, India.Keywords: hyperspectral image; image fusion: automatic road extraction IntroductionAttaining geospatial information is most important in spatial decision making and many applications in remote sensing and geographic information systems. When considering geospatial information for decision making for problems related to an urban environment, roads form an integral part of the data requirement. Rangwala (2000) described roads as a man-made linear feature, generally rectangular, which divides the town in more or less rectangular sectors. Detailed and accurate information about the road network is the foundation for comprehensive management and planning of transportation infrastructure and assets. Quality standards for the required data have evolved considerably over the last decades. Remote sensing has the potential to provide detailed road mapping and may offer more up-to-date and economical methods to improve road network updating. The geometry and structure of road surfaces result in a spectrally indistinct response with a high amount of spatial variability. These factors become even more complex and complicated within urban areas, where the road network is generally dense. The abundant presence of structures such as roofs, transportation surfaces, vegetation,

Mapping Lithological and Mineralogical Units Using Hyperspectral Imagery

“…EO-1 Hyperion hyperspectral image (Earth Observation-1) designed by NASA's in (Nov. 2000) (Tiwari et al, 2010;White et al, 2007). According to (Jafari & Lewis, 2012), the EO-1 satellite follows the same Landsat orbits of about 1 min., and the spatial resolution is about (30 m) with a standard scene of 42 km long and 7.7 km wide.…”

Decision Tree Algorithm Based on Regional Growth for the Automatic Oil Field Road Extraction

“…The image feature extraction is the extraction of useful information from the image [1,2,3]. The road characteristics are different in road extraction [4,5,6].…”

A Methodology to Assess the Accuracy with which Remote Data Characterize a Specific Surface, as a Function of Full Width at Half Maximum (FWHM): Application to Three Italian Coastal Waters

“…However, hyperspectral data have been used in a variety of fields [ 3 , 4 ]. For example, Hyperion satellite data have been used to study buried archaeological structures [ 5 ], minerals and soil [ 6 , 7 ], oil slick thickness [ 8 ], seagrass and the algal blooms [ 9 – 11 ], shallow water bathymetry [ 12 , 13 ], snow and ice [ 14 , 15 ], the urban surface [ 16 – 18 ], vegetation [ 19 , 20 ], volcanic plumes [ 21 ] and water column constituents [ 22 – 24 ].…”