Approach of Texture Signature Determination - Application to Forest Cover Classification of High Resolution Satellite Image

This paper presents an approach of textural signature identification for the classification of high resolution satellite image of forest. We are looking for the most appropriate combination of features from texture measures. This combination which forms our signature should allow the discrimination between different types of textures present in the image that will be classified. We improve our signature by a step of weighting features. The weight of each feature reflects its degree of confidence. We finish with an experimental step which is an application of our combined weighted signature for the purposes of classification of high resolution satellite image of forest.