Visual Positioning in a Smartphone

Abstract

Numerous techniques for obtaining motion and location related information are needed for obtaining seamless positioning capability with smartphones. High-quality cameras are nowadays widely available in portable devices and can provide necessary redundant data about the user’s surroundings in addition to the other sensors usable for positioning purposes. In this chapter, after introducing methods of image processing related to feature extraction, applicable methods for visual positioning are discussed with state-of-the-art examples. Regardless whether the visual-based positioning is based on reference images in a database or using information obtained from consecutive images, the first steps of pre-processing are similar to obtain noiseless images for accurate calculations and to retrieve the required camera parameters. Smartphones have limited computational resources and that restricts the methods available for image processing. To carry out the visual positioning function, features in images are either matched to corresponding features in consecutive images, or to a database. Obtaining the location can be performed with matching the query image to a database of reference images equipped with location information. Alternatively, the attitude and position change can be resolved from consecutive images to provide localization augmentation that may be fused with other sensor information. When smartphones are concerned, the restricted resources however bring about challenges that are the focus in this chapter.