Image forgery detection device and method

Abstract

The invention provides an image forgery detection device and method, and the method comprises the steps: receiving two parameters T and R which represent the threshold value of a Low's algorithm and the proportion of the number of pairs of matched feature points in all feature numbers respectively; preprocessing the image, namely converting the original image into a grey-scale map; calculating scale invariant feature transformation features of the image, wherein the scale invariant feature transformation features comprise feature point coordinate information and feature vectors; carrying out feature matching on the extracted features; filtering the matched feature point pairs according to a certain filtering strategy, and counting the number of the matched point pairs conforming to the filtering strategy; finally, judging whether the proportion of the number of the matching point pairs to the total number of the feature points is larger than or equal to a parameter R or not: if yes, determining that the image is forged, otherwise, determining that the image is non-forged, and for the image considered to be forged, connecting lines of the matching point pairs are drawn on the original image. According to the device and the method, image tampering behaviors such as multiplexing, rotating, overturning, zooming and the like in academic periodicals can be actively screened in batches.

Description

technical field [0001] The present invention relates to the technical field of image forgery detection, in particular to an image forgery detection device and method. Background technique [0002] With the progress of society and the development of technology, the use of cameras, mobile phones and other equipment has become more and more common, and digital images have become a multimedia resource that can be seen everywhere in real life. At the same time, the prevalence of image editing software also makes it easier for people to tamper with images, and with the development of technology, such tampering is becoming less and less detectable, and the security problems of digital images are becoming more and more serious. Image forgery detection technology (Image Forgery Detection) can realize automatic detection of forged images through computers and using certain methods. [0003] Image forgery detection technology is widely used in natural scene graphs. It can detect real ...

AI Technical Summary

Problems solved by technology

For the practical needs of academic image tampering detection, some researches have been successful, and some solutions are feasible, but for different types of academic images, these methods still have certain limitations

First, there is a single detectable counterfeiting method

Usually, academic image falsification methods include multiple methods such as multiplexing, rotation, flipping, scaling, etc. It is necessary to detect different falsification methods as much as possible, but some current solutions cannot well support this requirement

Second, it is impossible to actively and batch-type detection

For a large number of paper journal data, it is very important to be able to support active and batch detection, which is not ideally solved in the existing methods

The third is that the strip images commonly used in the biomedical field often have "afterimages", which is called "smearing" in professional terms. This image feature will seriously affect the detection effect of existing methods

Fourth, the existing feature extraction and matching technology will generate many useless matches in academic image detection, which seriously affects the final result judgment

Images