Homologous local copy detection method based on superpixel multi-feature matching

Abstract

The present invention discloses a homologous local copy detection method based on superpixel multi-feature matching. The method comprises the steps of: performing Gaussian smoothing filtering preprocessing of an image to be detected, employing a BEMD algorithm to obtain an image texture degree proportion, and performing adaptive initialization of the number of superpixel blocks; employing a superpixel segmentation algorithm based on SLIC to obtain image superpixel blocks, and employing a color look-up table color quantification technology and a texture moment analysis method to obtain featureinformation of the superpixel blocks; employing Rg2NN algorithm and a BBF algorithm to perform feature matching of the superpixel blocks; and finally, employing an SIFT feature point extraction algorithm, a RANSAC method, a ZNCC algorithm and a morphological method to perform postprocessing.

Description

technical field [0001] The invention belongs to the technical field of digital image forensics, and relates to an image tampering detection method for copying and pasting, in particular to a homologous partial copying tampering detection method based on superpixel multi-feature matching. Background technique [0002] In recent years, images have become an indispensable part of people's lives, especially in news, medicine, military, intelligence and other fields, where images play an indispensable and important role as a carrier of information and digital evidence. With the rapid development of computer network technology and portable digital devices and the continuous expansion of application scope, it brings efficiency to people's life and work, and also makes it easier to tamper with images. The rapid growth of image processing software has led to a large number of tampered images without obvious traces. The frequency of tampered images is getting higher and higher, and th...

AI Technical Summary

Problems solved by technology

Among them, the block-based detection method has the advantages of high positioning accuracy, and has a good detection effect on images processed by noise and compression, but the time complexity of this type of detection method is extremely high; while the detection method based on feature points extracts the Detecting the feature points of the image, and using the corresponding local feature descriptors to match the feature points, the detection time complexity of this method has been greatly reduced, but this type of method is not effective for smooth tampered area detection. ideal

Therefore, most of the existing tamper detection methods still have the problem that the detection accuracy is improved, and the time complexity is also increased, or the time complexity is reduced, but the detection accuracy is also reduced.

Images