Respond Projects

Research Areas

Techniques for multi-date data registration and mosaics

Image-registration can be defined as ‘the process of overlaying images (two or more) of the same scene taken at different times, from different viewpoints, and/or by different sensors’. It plays an essential role in remote sensing as it is a prerequisite for the analysis of multidate image sequences. Image co-registration issues might seriously influence the final quality of the remote sensing analysis and products because small misalignments in the input data could give large errors in the final outputs. e.g. for change detection applications. Image-to-image methods have two important advantages. The first one concerns the analyzed data: many applications require the comparison of multi-temporal and multisource data that should be precisely aligned. The relative co-registration of different images is therefore more important than their absolute geo-referencing. Secondly, imageto-image co-registration methods are based on the analysis of dataset of the same type. This leads to a general simplification of the co-registration process especially in the case automatic co-registration techniques are used. The co-registration essentially required for Bhuvan portal, where the multiple images has to match each other previous layer. It includes free download of AWIFS and LISS-3 orthorectified products and High Resolution satellite of multiple years, which are being used for feature extraction. The co-registered product enables automatic updation of newly identified features in latest images.