Agricultural Fields Classification in Semi-Arid Central Tunisia using SPOT 7 Image

This paper reports on classification methods applied and tested for land use classification in a semi-arid environment. Our study, conducted on two irrigates sites located in the Kairouan region, the largest irrigated region in Tunisia, compared Support Vector Machine (SVM) and Maximum Likelihood classification of SPOT-7 data. To produce a per-field classification a Mean-Shift Segmentation has been performed on the pansharpened SPOT-7 images. A field survey has been conducted to. Accuracy assessment was done to evaluate the performance of the proposed using collect ground truth data on land use and extend of all the agricultural fields within the study areas obtained through filed survey.

Publication year:2016

Keywords:remote sensing, classification, segmentation