Image retrieval method based on multi-feature fusion

Abstract

The invention discloses an image retrieval method based on image multi-feature fusion, which is mainly used to solve the problem of low accuracy of image retrieval in the prior art. The implementation steps are: (1) extract the three visual features and semantic attribute features of all images in the image set to be retrieved; (2) calculate the adjacency distance matrix between all images in the image set to be retrieved; A feature channel for rough retrieval; (4) According to the rough retrieval results of each feature channel, analyze the semantic attribute characteristics of the images in the coarse retrieval results, and select a reference image; (5) Calculate each A feature fusion template matrix; (6) Obtain a fused distance measure matrix according to the obtained fusion template matrix; (7) Return the retrieval result to the user according to the obtained distance measure matrix. The invention obviously improves the accuracy rate of final image retrieval and can be used for image retrieval.

Description

technical field [0001] The invention belongs to the technical field of information retrieval, and specifically relates to an image retrieval method based on image multi-feature fusion, which can be used in the field of Internet image retrieval. Background technique [0002] Under the current Internet background, most commercial network image search engines such as Google, Image Search, and Bing use text-based retrieval technology, mainly using annotations such as titles and description text around images to make similar queries to query text. Sex matching, using this similarity to retrieve the retrieved images. However, techniques based solely on text retrieval often fail to achieve good results due to the semantic gap between text and image content. In addition, due to the presence of ambiguous noise data in image annotation, the correctness of image labels derived from image metadata analysis is often not guaranteed. [0003] In order to improve the accuracy of existing ...

AI Technical Summary

Problems solved by technology

Although today's image retrieval methods based on multi-feature fusion can improve the accuracy of image retrieval and ranking results to a certain extent, essential problems still exist.

Different retrieval situations correspond to different importance of visual features, and it is often difficult to obtain satisfactory image retrieval results by directly performing multi-feature fusion with fixed coefficients.

Images