A Fast Interest Points Feature Descriptors Algorithm for Mobile Image Retrieval Applications

High smart phones capability increases visual data uploaded and downloaded between users and service providers. Query by image content means applying computer vision techniques to the image retrieval problem or searching for digital images in large databases. Such applications are becoming vital: medical, forensics, military, and novel Geographical Information Systems (GIS). Future may enable a new class of applications which uses mobiles to search about objects. Google goggles, kooaba, and Snaptotell are all considered pioneers that use image retrieval algorithms. This paper introduces a novel mobile server-based visual search application. The proposed method introduces Interest Points (IPs) feature descriptors from query that could overcome image capturing problems: noise, rotation, and cropping effects. The paper applies the proposed IPs descriptors on an adequate image database. Comparison with related work, error and time measures are also considered.