Home Articles Integration of Optical and Radar Imageriesfor Precise Land-Cover Mapping

Integration of Optical and Radar Imageriesfor Precise Land-Cover Mapping

Hosein Torabzadeh
Bu Ali Sina University,
Email: [email protected]

Ali Akbar Abkar
Jahad ministry
Email: [email protected]

Abbas Alimohamadi
K.N. Toosi University

Data integration is lately being used as a new and efficient technique to improve the capability of different image types. Due to numerous advantages this technique is even employed by some commercial companies involved in satellite data business. Present study aims to provide a more precise land cover map using the integration of
two different images. Due to high spatial and spectral resolution as well as its similarity
with human’s vision optical data is the most commonly used data among the users. Ever developing technology of optical sensors is a valid evidence for this fact. However due to passive mode of optical sensors they mostly depend on sun as energy source so they are lacking some capabilities. Optical sensors are not able to receive data under all weather condition and of all time of the day and night. Furthermore structural and textural characteristics of land surface features are not precisely detected by means of optical sensors. Microwave systems mostly eliminate such deficiencies but they don’t offer enough details of the composition. Therefore proper integration technique avails to use both advantages for pattern recognition purpose. Landsat-TM and RADARSAT-SAR have been employed as the two major data, which have been integrated during this research study. The western portion of Rasht a city along northern part of Iran is selected as study area. Different image processing techniques such as HIS, PCA, Brovey as well as another feature-based technique (FI) are employed for image integration purpose. The results of all employed techniques are evaluated and compared. Final results of present study revealed 67, 58, 67, and 92 percent of accuracy respectively .The accuracy of different land cover classes has also been investigated during the research. Consequently all findings approve FI as the most efficient technique to integrate TM with SAR data. Key Words : Remote Sensing, Image Integration, SAR, Classification.