Highly Efficient Mobile Visual Search Algorithm

Abstract

In this paper, we propose a highly efficient mobile visual search algorithm. For descriptor extraction process, we propose a low complexity feature detection which utilizes the detected local key points of the coarse octaves to guide the scale space construction and feature detection in the fine octave. The Gaussian and Laplacian operations are skipped for the unimportant area, and thus the computing time is saved. Besides, feature selection is placed before orientation computing to further reduce the complexity of feature detection by pre-discarding some unimportant local points. For the image retrieval process, we design a highperformance reranking method, which merges both the global descriptor matching score and the local descriptor similarity score (LDSS). In the calculating of LDSS, the tf-idf weighted histogram matching is performed to integrate the statistical information of the database. The results show that the proposed highly efficient approach achieves comparable performance with the state-of-the-art for mobile visual search, while the descriptor extraction complexity is largely reduced. key words: mobile visual search, descriptor extraction, feature selection, reranking

Analysis of the Visual Design and Expression Effect of Virtual Reality and Three-Dimensional Space Technology in Art Space

“…Using instancing can greatly reduce the number of polygons in the scene, saving a lot of memory. Using this technique in a distributed simulation will greatly reduce the amount of data transfer [22]. e instantiation algorithm has attracted much attention in current research and is famous for its various advantages.…”

“…e instantiation algorithm has attracted much attention in current research and is famous for its various advantages. It uses matrix transformation and sacrifices time in exchange for memory space [21,22].…”

Feature Fusion for Image Retrieval With Adaptive Bitrate Allocation and Hard Negative Mining

“…Finally, the global descriptor g, the local descriptor which contains a set of compressed local features, and the other coding information are merged to generate the CDVS bitstream. In the retrieval stage, the global and local descriptors are separated out from the received query bitstream on the server end [20]. Then, the query global descriptor g is compared with each global descriptor in the global database, and based on the similarity score, the topranked N candidate images are returned.…”

“…For each query image, the corresponding global and local descriptors are separated out from the bitstream. Then the CDVS global descriptor and the corresponding CNN embedding are combined to conduct retrieval, and based on the matching scores calculated by (20), a shortlist S with the top-ranked S list images, such as 500, is returned.…”

“…the query image and i th image of S. In each matching, first an exhaustive pairwise comparison is performed to find all matched hand-crafted feature pairs between query and the image in S. Then, the ratio test and GCC are conducted to remove the wrong matched pairs, and finally the remaining P good match pairs are used to generate the matching score m q 1×S list (i) between query and i th image according to (21) [20].…”