Image retrieval method based on incremental learning of large vocabulary tree

Abstract

The invention discloses an image retrieval method based on incremental learning of a large vocabulary tree. The generated large training picture vocabulary tree which contains millions of images is used as a dictionary, leaf node number information associated with feature descriptors of newly added pictures is extracted from the vocabulary tree so as to form a vector for description, the incremental learning of the newly added pictures is completed, leaf node number information associated with feature descriptors of to-be-queried pictures is extracted in a same mode as the incremental learning so as to form a vector, and the two leaf node information vectors are compared so as to find out an image represented by the vector with a highest contact ratio with the leaf node number information of the to-be-queried pictures as a retrieval result to be returned. The image retrieval method has the advantages of high accuracy of the retrieval result and good robustness, and a real-time image retrieval task of a database increment can be satisfied.

Description

technical field [0001] The invention belongs to the technical field of image retrieval, in particular to an image retrieval method based on incremental learning of large-scale vocabulary trees. Background technique [0002] In recent years, with the rapid development of the mobile Internet and the popularization of mobile terminals, people can take and share their own pictures as they like. Through the retrieval and identification of pictures, user behavior can be analyzed to provide data support for various industries. Such as: understanding customers, meeting customer service needs, optimizing machine and equipment performance, improving safety and law enforcement, improving urban transportation, financial transactions, etc. Therefore, image retrieval and recognition technology has extremely important application value. Therefore, quickly and effectively managing and retrieving valuable information has quickly become an urgent need for people, and then Content-Based Imag...

AI Technical Summary

Problems solved by technology

At present, due to the limitation of computer development level, the CBIR system cannot really support semantic-based image retrieval. In order to improve the efficiency of matching, it takes a lot of time to train the feature vectors in the process of indexing to achieve fast retrieval. Effect

The image retrieval technology based on the vocabulary tree is an effective way, which can quickly find the retrieval results in millions of pictures, but its defect is that the training time is too long, and the training time for millions of pictures is as long as dozens of Hours, if you want to add new image data to the original database model and maintain a high retrieval accuracy rate, the cost of training the newly added images to the database with the images in the original database is too high. Therefore, a solution is proposed, using the existing trained large-scale tree index structure containing millions of pictures as a dictionary, quickly extracting data from the tree structure to represent newly added pictures, and completing the purpose of incremental learning

Images