A New Index to Perform Shadow Detection in GeoEye-1 Images

With the introduction of new satellites for earth monitoring characterized by very high resolution (VHR) sensors, new algorithms to recognize shadow in the supplied images are necessary. Automatic shadow detection can enhance the interpretability of the images in several applications such as classification and change detection. Several approaches are present in literature for shadow detection and their adaptation and particularization for VHR satellite images are still in evolution. The goal of this paper is to propose a new index for shadow detection based on multispectral files processing. GeoEye-1 satellite data are used for this study: IHS pan-sharpening method is applied to transfer pixel dimensions of the panchromatic image (spatial resolution: 0.5 m x 0.5 m) into the multispectral images (2 m x 2 m); an index named ERGAS is used to test the quality of the resulting raster files. Dealing with the problem of the shadow detection, a new index is defined to identify the affected pixels both in the original as well as pan-sharpened images. The results are compared with them by another index named ratio that is generally applied for shadow detection in multispectral images: issues and advantages, derived by using the proposed technique, are discussed.