Mapping banana plantations in North Australia from object-based classification of SPOT-5 data

The objective of this research was to develop and evaluate an approach for objectoriented mapping of banana plantations from SPOT-5 multi-spectral and panchromatic image data. Non-cultivated areas were first masked out using spectral and elevation data. Banana plantations were then separated from all other cultivated classes using object-oriented segmentation and supervised classification. Within the cultivated areas, separation of banana plantations and other land-cover classes increased when including image co-occurrence texture measures and context relationships. The results showed that a pixel size of ≤ 2.5 m was required to accurately identify the row structure within banana plantations, which enabled separation from other crops based on texture information. The user's and producer's accuracies for object-oriented mapping of banana plantations increased from 73% and 77%, respectively, to 94% and 93% after post-classification visual editing. Classification errors mainly occurred for plantations with damaged, young, or variably aged banana plants. The results derived from the object-oriented approach indicate that this process may be automated in the future. This will require high spatial resolution image data with pixels ≤ 2.5 m captured close to nadir for inclusion of additional textural information