• Each image is divided into three horizontal strips. The upper one stands for sky, the middle one for trees/foliage and the bottom one for grass/ground

• Different features are extracted from each region and then training and test are carried out on them respectively

• Classification results from different regions are compared