Mobile Visual Location Recognition By Using Panoramic Images Applications

Mobile Visual Location Recognition(MVLR) has attracted a lot of researchers' attention in the past few years. Existing MVLR applications commonly use Query-by-Example(QBE) based image retrieval principle to fulfill the location recognition task. However, the QBE framework is not reliable enough due to the view of query image may be too narrow, containing relatively small information. To solve the above problem, we make following contributions to the design of a panorama based MVLR system.We provide users two kinds of query modes,using panorama mode of the phone's camera, or shooting several pictures then stitching them. To address the limitations of projective transformation, we combine projective transformation and rigid transformation, global transformation and local transformation, improving the deformation and blurring problem. The database images are complete 360-degree panoramas, while the submitted query may not be. To filter out the disturbed areas, we design a heading(from digital compass) aware BOF(Bag-of-features) model to generate the descriptors of panoramic images. To reduce the memory and computational time requirements, we adopt a compressed sensing based encoding method to encode the descriptors. While the measurement matrix is extreme large, we propose an effective bilinear compressed sensing based encoding method. While being fast and accurate enough for on-device implementation, our algorithm can also reduce the memory usage of projection matrix significantly.Experimental results prove the effectiveness of the proposed methods for MVLR applications.