Slot-cutting evidence obtaining method based on LBP and extended Markov features

Abstract

The invention belongs to the technical field of image processing and digital image evidence obtaining, proposes an algorithm for more effective detection of slot-cutting tampering, and solves an insensitive problem when the conventional Markov features are used for detecting the slot-cutting tampering proportion. A slot-cutting evidence obtaining method based on LBP and extended Markov features comprises the steps: respectively calculating Markov transition probability matrixes with the differential two-dimensional JPEG matrix steps q=1 and q=4 in four directions in the LBP domain: horizontaldirection, vertical direction, leading diagonal direction and secondary diagonal direction; carrying out the fusion of the features obtained under the condition q is equal to one and the features obtained under the condition q is equal to 4; extracting 648(324*2)-dimensional features, and carrying out the mathematic modeling of the 648-dimensional features in the obtained transition probability matrixes; taking the model as a feature vector, and carrying out the training and detection of the feature vector through an SVM (support vector machine), so as to detect whether the slot-cutting tampering of the image is carried out or not. The method is mainly used in an image processing occasion.

Description

technical field [0001] The invention belongs to the technical fields of image processing and digital image forensics, and specifically relates to a thin-slit forensics method based on LBP and extended Markov features. Background technique [0002] With the rapid development of digital technology and the rapid spread of the mobile Internet, digital images have been integrated into people's work, study and life, such as images displayed on social networks, images that appear when browsing news, and images on the display screens on buses and subways. .....Digital images are becoming more and more closely related to humans. However, all kinds of simple and easy-to-get digital image editing software that came into being brought convenience and entertainment to people, and at the same time damaged the authenticity and integrity of digital images. The traditional concept of "seeing is believing" has gradually been broken. , Malicious digital image tampering hinders judicial identi...

AI Technical Summary

Problems solved by technology

But it can also be used for malicious tampering of images, such as removing a certain target in the image, such as figure 2 As shown, the image on the left shows a pair of walking men and women. After delineating the area to be removed, the selection of the slit will give priority to the pixels in the delineated area to be deleted when the image is shrunk, and the final zoomed out effect As shown in the figure on the right, it can be seen that the semantic content of the original image has been tampered with by thin seam cropping, but it is impossible to judge whether the image has been tampered with visually, so it is necessary to detect such tampering

Images