Near-repetitive image retrieval method based on consistent region deep learning features

Abstract

The invention discloses a near-repetitive image retrieval method based on consistent region deep learning features. The near-repetitive image retrieval method specifically comprises the following steps: extracting SIFT features of all images in an image library, quantizing the SIFT features into visual words, and establishing an inverted index file for all SIFT features; reserving K target regionsof each image, and calculating CNN features C(Rc) of the target region; extracting SIFT features of the query image, and quantizing the SIFT features into visual words; finding out candidate images by utilizing the inverted index file; finding out a nearly repeated region approximately repeated with each target region of each candidate image in the query image; extracting a CNN feature C(RQ) of the near-repetitive region; calculating cosine similarity between any C(Rc) and C(RQ) corresponding to the CNN feature, and taking the cosine similarity as a similarity score of the group; and in eachcandidate image, selecting a group of scores with the highest cosine similarity as similarity scores between the candidate image and the query image. According to the near-repetitive image retrieval method, the retrieval efficiency is improved, and meanwhile, the image retrieval accuracy is greatly improved.

Description

technical field [0001] The invention belongs to the field of information security, and in particular relates to a near-duplicate image retrieval method based on consistent region deep learning features. Background technique [0002] Due to the widespread use of powerful image processing tools and the rapid development of Internet technology, digital image data is increasingly easy to be illegally copied, tampered with and transmitted on the network. In fact, these illegal images are near-duplicate images that share small duplicated areas and undergo various image modifications such as rescaling, occlusion, noise addition, and brightness and color changes. To prevent image content from unauthorized use and privacy violation, detecting illegal partially reproduced versions of copyrighted images has become an urgent problem. Therefore, as a branch of content-based image retrieval, near-duplicate image retrieval plays a very important role in the fields of copyright and privacy...

AI Technical Summary

Problems solved by technology

[0009] Purpose of the invention: In order to solve the problems that the existing retrieval technology is not suitable for near-duplicate images sharing a small part of the duplicated area, and the retrieval efficiency is low; the present invention provides a near-duplicate image retrieval method based on deep learning features of consistent regions

Images