Field-based Classification of Agricultural Crops Using Multi-scale Images

This paper presents field-based classifications performed using the multi-resolution images of SPOT4 XS, SPOT5 XS, IKONOS XS, QuickBird XS, and QuickBird Pansharpaned (PS) covering an agricultural area located in Karacabey, Turkey. The objective was to assess the classification accuracies of different spatial resolution images in an agricultural land using the field-based classification techniques. To do that pre-field classification was performed using the common bands of the images. For each field, the statistical measures of the mean, median, and mode values were calculated. Then, a Maximum Likelihood Classification (MLC) was carried out using the derived bands. After computing the accuracies of the classifications, it was observed that similar results were obtained, for each image, when the mean values were used. Of the images used, the 0.61m resolution QuickBird PS image provided the highest overall accuracy of 85.2% using the median bands classification. On the other hand, the lowest overall accuracy was found to be 42.9%, when the SPOT4 XS image was classified using the median bands.