Texture based change detection of built-up on SPOT panchromatic imagery using PCA

This paper addresses built up change detection using texture features derived from SPOT Panchromatic imagery and principal component analysis (PCA). SPOT imagery combines the adequate ground sampling distance (GSD) and large area coverage that extends beyond cities to include the peri-urban areas where urban sprawl (built-up change) typically occurs. This study uses SPOT 1 and SPOT 5 panchromatic data over Casablanca. The texture based “Built up area presence index” is used to provide a likelihood of built up area. The changes in texture in t
₀
and t
₁
are then quantified using PCA. The procedure is simple to implement and requires minimum manual intervention. The analysis shows that the method is reliable for rapid assessments built up changes but that these detected changes need to be visually inspected if operational products have to be derived. The technique performs adequately when changes occur over large areas. The Casablanca study results are encouraging despite the coarse GSD of SPOT 1 and the different illumination and viewing angle. The procedure should be tested over cities in different geographical regions. Finer resolution input data and data with similar acquisition and illumination angle are likely to yield better the results.