Abstract for presentation at The 13th Australasian Remote Sensing and Photogrammetry Conference

Semi automatic planimetric feature extraction from landsat TM imagery using fuzzy reasoning rules

  • PhD Candidate Ali Mohammadzadeh, K.N.Toosi University of Technology - Remote Sensing Eng. Dept., Iran
  • Assistant professor Mohammad Javad Valadan Zoej, K.N.Toosi University of Technology - Remote Sensing Eng. Dept., Iran
  • Associated professor Ahad Tavakoli, Amirkabir University of Technology - Electrical Eng. Dept., Iran
  • BSc Student Naeim Mohammadzadeh, Imam Hossein University, Iran

    • Fuzzy logic (FL) has become of great interest, to remote sensing and Geographic Information Systems expertise, since it can provide more reliable information. This paper presents a method for classification and extraction of crops from Landsat TM (Thematic Mapper) images using fuzzy reasoning method. Each pixel is transformed into a matrix of membership degrees representing the fuzzy inputs. Then a minimum-reasoning rule is applied to infer the fuzzy outputs, and finally, a defuzzification step is applied to extract features.
    The accuracy and performance of the explicit fuzzy method is compared with Maximum Likelihood Classification (MLC) method for both simulated image and Landsat TM image. The results do not show any considerable differences. The strength of the fuzzy method is its simplicity and has high flexibility for easily inserting new bands or removing bands without disturbing the remaining parts of the classifier. Another advantage of this fuzzy classification method is its ease in introducing on artificial neural networks.

    Conference Organiser - ICMS Pty Ltd