Robust image Hashing method based on SIFT (Scale-Invariant Feature Transform) and LBP (Local Binary Pattern) mixing

Abstract

The present invention discloses a robust image Hashing method based on SIFT (Scale-Invariant Feature Transform) and LBP (Local Binary Pattern) mixing. The method has the beneficial effects that two major features of SIFT and LBP are combined and after feature descriptors of SIFT extraction are transformed into a binary system through a quantitative rule, the storage space can be greatly compressedat the same time, so that the conditions of the hash coding are met; the rotation operation and the scaling operation on a received image can be restored according to the pairing of the SIFT descriptors, the LBP features of the received image are extracted and the received LBP codes are compared with the extracted LBP codes to ensure that the tampered areas of the images can be detected; and in the meantime, the robustness of hash codes to the content retention operation and the sensitivity of the hash codes to the content tampering attack are both considered.

Description

technical field [0001] The invention relates to a robust image hashing method based on the mixture of SIFT (scale-invariant feature transform) and LBP (local binary pattern), belonging to the field of digital image processing. Background technique [0002] Image hashing algorithm has become more and more important and widely used in image content authentication, image retrieval in image database, and image forgery recognition in recent years. Image hashing technology is based on the content of the image, extracting a hash code unique to different images as the identity mark of the image. The image hash code should have the following properties: (1) One-way: the hash code can be easily extracted from the image content, but it is impossible to deduce the image content from the hash code. (2) Perceptual robustness: The same as the visual characteristics of the human eye, the hash codes obtained by the images with the same or similar content perceived by the human eye should be...

AI Technical Summary

Problems solved by technology

[0003] After the hash code has the above characteristics, it is not easy to design an algorithm that is robust to the content-preserving operation of the image and has high sensitivity to the content-offensive operation, and the hash code should have a comparative Due to the small length of the original image, the hash code with good functions requires more space resources, so the function of the hash code and the consumption of space become a pair of contradictions

Although there are already many image hashing algorithms, the existing image hashing methods only target specific image attacks, and are not robust to image content preservation and are vulnerable to joint content tampering attacks.

Images