Nowadays, urban and vehicular surveillance systems are col-lecting large amounts of image data for feeding recognition systems e.g.towards proposing localization or navigation services. In many cases,these image data cannot directly be processed in situ by the acquisitionsystems in reason of their low computational capabilities. The acquiredimages are transferred to remote computing servers through various com-puter networks, then analyzed in details towards object recognition. Theobjective of this paper is twofold i) presenting image-based cypheringmethods that can eciently be applied for securing the image transferagainst consequences of image interceptions (e.g.; man-in-the-middle at-tacks) ii) presenting generic image-based analysis techniques that can beexploited for object recognition. Experimental results show end-to-endimage-based solutions for fostering developments of surveillance systemsand services in urban and vehicular environments.