Abstract:

Pixel based classification are often used to extract land use information. However, fragmentation results make the accuracy is limitted. Image classification can be combined with expert knowledge and spatral values to improve the accuracy of classification results. In this study, expert knowledge was used in object - oriented classification to extract land use information in Thai Thuy district, Thai Binh province. Object-oriented classification is an effective method by dividing the image into hierarchies. The parameters in this hierarchical include: spectral value, NDVI index, distance of objects, geomorphological factors and soil factors. Thê ovêrall accuracy of classification rêsults was ρ = 0.713.