Improving land-cover classification using recognition threshold neural networks

The use of neural networks to classify land-cover from remote sensing imagery relies on the ability to determine a winner from the candidate land-cover types based on the imagery information available. In the case of a “winnertakes-all” scenario, this does not allow us a measure of how much the prediction of each pixel’s land-cover can be trusted. We present a three-stage method where only winning candidates which are given a clear lead over the other land-cover types are accepted, with a neighborhood relationship and the application of mixed pixels being used to provide full classification. This method allows us to place more faith in the resulting map than simply taking the winner, and results in a higher accuracy of classification. The method is applied to Landsat imagery of an area of the Philippines where natural, urban, and cultivated land-cover types exist.

[1]  B. Steele,et al.  Land Cover Mapping Using Combination and Ensemble Classifiers , 2002 .

[2]  W. Salas,et al.  Mapping secondary tropical forest and forest age from SPOT HRV data , 1999 .

[3]  Eric Mjolsness,et al.  Neural networks, pattern recognition, and fingerprint hallucination , 1986 .

[4]  Xinhua Zhuang,et al.  Optimal Bayesian classifier for land cover classification using Landsat TM data , 2000, IGARSS 2000. IEEE 2000 International Geoscience and Remote Sensing Symposium. Taking the Pulse of the Planet: The Role of Remote Sensing in Managing the Environment. Proceedings (Cat. No.00CH37120).

[5]  Grégoire Mercier,et al.  Estimation and monitoring of bare soil/vegetation ratio with SPOT VEGETATION and HRVIR , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[6]  P M Atkinson,et al.  Linking remote sensing, land cover and disease. , 2000, Advances in parasitology.

[7]  Abbas Alimohammadi,et al.  A NEW APPROACH FOR MODELING UNCERTAINTY IN REMOTE SENSING CHANGE DETECTION PROCESS , 2004 .

[8]  Aitkenhead.  Mapping land use in NE Scotland with neural networks from remote sensing imagery , 2005 .

[9]  D. Civco,et al.  A COMPARISON OF LAND USE AND LAND COVER CHANGE DETECTION METHODS , 2002 .

[10]  Li An,et al.  Using artificial neural networks to map the spatial distribution of understorey bamboo from remote sensing data , 2004 .

[11]  Yannis Avrithis,et al.  Fuzzy image classification using multiresolution neural networks with applications to remote sensing , 1997, Proceedings of 13th International Conference on Digital Signal Processing.

[12]  Sucharita Gopal,et al.  Uncertainty and Confidence in Land Cover Classification Using a Hybrid Classifier Approach , 2004 .

[13]  Kyle Brown.  Per-Pixel Uncertainty for Change Detection Using Airborne Sensor Data , 2004 .

[14]  Grégoire Mercier,et al.  Estimation and monitoring of bare soil/vegetation ratio with SPOT VEGETATION and HRVIR , 2003, IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477).

[15]  S. K. Yen,et al.  Neural classification of SPOT imagery through integration of intensity and fractal information , 1997 .

[16]  Pramod K. Varshney,et al.  Sub-pixel land cover mapping based on Markov random field models , 2003, IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477).

[17]  A. N. C. S. Rosa,et al.  LAND USE/LAND COVER MAPPING IN BRAZILIAN AMAZON USING NEURAL NETWORK WITH ASTER/TERRA DATA , 2004 .

[18]  Peter M. Atkinson,et al.  The integration of spectral and textural information using neural networks for land cover mapping in the Mediterranean , 2000 .