Trademark image retrieval method and system, medium and equipment

Abstract

The invention relates to a trademark image retrieval method and system, a medium and equipment. The trademark image retrieval method comprises the following steps: according to all similar group numbers of each trademark classification number, carrying out library division on a feature library containing feature vectors of trademark images; clustering the feature vectors of the trademark images ineach sub-library according to the determined number of the clustering centers, and dividing a sub-library for the feature vectors belonging to the same clustering cluster in each sub-library; determining a sub-library to which the to-be-queried trademark belongs according to the classification number of the to-be-queried trademark and the trademark image; calculating the distance between the feature vector of the trademark image of the trademark to be inquired and the feature vector of the trademark image in the subordinate sub-library; and outputting the trademarks corresponding to the feature vectors of the first N trademark images with the smallest distance as retrieval results. According to the method, the search range during online retrieval can be greatly reduced, and especially when the vector dimension is relatively high, the retrieval efficiency can be remarkably improved.

Description

technical field [0001] The present invention relates to the field of data retrieval, in particular to a trademark image retrieval method, system, medium and equipment. Background technique [0002] Trademark search is an important task in procedures such as registration application, trademark examination, trademark management, and trademark rights protection. The main purpose of the trademark search is to find the same or similar trademarks in the sense of the trademark law, so as to confirm the information on the scope of ownership of a certain trademark. The existing trademark retrieval method is to pre-extract the feature vector of the trademark image in the existing trademark database, establish a trademark feature database, and use the same method to extract the feature vector of the trademark image of the trademark to be queried, and then compare them one by one The feature vectors in the trademark feature database calculate the similarity, and then output the retriev...

AI Technical Summary

Problems solved by technology

The existing trademark retrieval method is to pre-extract the feature vector of the trademark image in the existing trademark database, and establish a trademark feature database. The feature vectors in the trademark feature database calculate the similarity, and then output the retrieval results sorted according to the similarity. Due to the huge amount of data in the trademark database, this retrieval method has a large search range, and due to the similarity between vectors Calculation is a very time-consuming operation, especially when the dimension of the vector is high, the amount of calculation is also very large, and the retrieval efficiency is extremely low

Images