Using the mixture-tuned matched filtering method for lithological mapping with Landsat TM5 images

作者：Mehr Saeed Goodarzi; Ahadnejad Vahid^*; Abbaspour Rahim Ali; Hamzeh Mohammad

来源：International Journal of Remote Sensing, 2013, 34(24): 8803-8816.

DOI：10.1080/01431161.2013.853144

Hyperspectral images are widely employed for geological mapping because of their high spectral resolution. In this article, we develop the mixture-tuned matched filtering (MTMF) method, which can provide highly accurate mapping using the minimum ground-based data. This method is applied in the Malayer region of western Iran, which is composed of various lithological units. MTMF and minimum noise fraction (MNF) methods were applied to a Thematic Mapper 5 (TM5) image, and minimum number of training data based on field observation were used to produce a suitable false-colour image in which locations of lithological units were given. Finally, classification of six desired lithological units was done by the maximum likelihood classification (MLC) method. Results of lithological mapping show that although minimum ground-based data were used, the accuracy of classification is 82.3%. In addition, evaluation of the above-mentioned false-colour image reveals that the algorithm presented enhances the separability of units in the image by 7.3%, which can partially compensate for the low spectral resolution of the image used.

• 出版日期2013-12-20