Digital Image Source Verification and Fake Screenshot Detection System

Abstract

The invention relates to a system that enables verification of whether a digital image is a true screen image by examining technical characteristics specific to the image creation process, such as subpixel structure, font rendering traces, rendering traces generated by the graphics processing unit, image compression residues, and anti-aliasing patterns.

Description

Digital Image Source Verification and Fake Screenshot Detection System Technical Area The invention determines whether a digital image is a true screen image by means of sub-pixels. structure, font rendering traces, generated by the graphics processing unit such as image traces, image compression remnants, and anti-aliasing patterns by examining the technical characteristics specific to the image creation process It is related to a system that enables verification. State of the Art Nowadays, visual inspection and file verification are commonly used in the verification of digital images. Metadata (EXIF) analysis and various digital forensics methods are used. These methods can provide limited information about the source and integrity of the image. However, current techniques have significant shortcomings. EXIF ​​and other metadata. They can be easily deleted or modified with image editing software. The manipulations that are carried out cannot always be reliably detected and the truth... Very close fake screenshots can be created. Therefore, a digital image is not actually an original screenshot taken from a device screen. is it an image or a subsequently edited or artificially created image? a solution that provides reliable, technical and objective verification of whether something is true It is needed. Application number US20230005122A1 includes pixel characteristics and metadata information from the image. It is related to the system that determines whether the image is fake or real by extracting it. However The application includes subpixel sequences, anti-aliasing properties, rasterization traces, and font design. The image generation method is determined by examining the sequence and GPU-based rendering patterns. It is not determined. In conclusion, due to the negative aspects described above and the current solutions being the subject of discussion... Due to its shortcomings, an improvement is needed in the relevant technical field. It has been made. 2 Purpose of the Invention The invention was created by drawing inspiration from existing situations and overcoming the aforementioned drawbacks. It aims to solve the problem. The main purpose of the invention is to create a digital image that is truly authentic, taken directly from a device screen. whether it is a screenshot or a subsequently edited, reproduced image The goal is to determine with high accuracy whether or not the image is a forged image. Instead of examining the visual content or modifiable metadata of an image, the invention... device and imaging process-specific events that occur during image formation It analyzes technical traces. This includes subpixel arrangement and font style. (font) render sequence, GPU rasterization traces, compression artifacts, and anti-aliasing Characteristic features of the image production process, such as those related to models, are evaluated. Within the scope of the invention, a pixel adjacency matrix is ​​created from the image, and this data... A render track model is obtained using this method, and the resulting model is displayed on the relevant device screen. The image is compared with the reference screen model. As a result of the comparison, the image... Is it a genuine screenshot or an edited or fake one? It is determined whether it is an image or not. The invention relates to image forensics and machine learning-based classification. by utilizing existing methods and techniques such as frequency domain analysis (FFT) situations where verification methods are insufficient in terms of reliability and accuracy It aims to provide solutions in the fields of banking, digital forensics, and e-commerce. platforms, courts, insurance companies and others that require digital document verification reliable and technically verifiable screenshot analysis in systems This is ensured. Detailed Description of the Invention This detailed explanation describes the digital image source verification and forgery process that is the subject of the invention. Screenshot detection systems and preferred configurations are not only better understood in this context. It is explained in a way that facilitates understanding. The invention is a system for digital image source verification and counterfeit screen image detection. The digital image uploaded by the user that needs to be verified is entered into the file input interface. an image that is accepted via the system and converted into a data format that can be processed by the system. 3 The acquisition module processes the pixels of the image transferred to the system by the image acquisition module. examining the structure, pixel neighborhood relationships, and subpixel arrangements within an image, Pixel analysis that analyzes anti-aliasing properties and compression artifacts. the engine, on the analysis outputs obtained by the pixel analysis engine To make an assessment, the rasterization traces and font drawings present in the image... The render trail detection unit analyzes the sequence and GPU-generated patterns of render trails. The analysis results obtained by the unit were used with a CNN-based machine learning model. classifier that classifies with the help of, classification produced by the classifier a reporting module that converts the result into a verification output to be presented to the user. It includes. The image acquisition module transfers the digital image to be verified into the system. It is the input unit that provides this. This module receives the image uploaded by the user as a file input. It accepts the data through the interface and converts it into a data format that can be processed by the system. Thus, the image data is converted into a suitable format that can be used in subsequent analysis stages. and is prepared for the analysis process. The pixel analysis engine analyzes the image transmitted to the system by the image acquisition module. It is an analysis unit that examines the pixel structure. This unit analyzes the pixel neighborhoods in an image. relationships, subpixel arrangements, anti-aliasing properties and analyzes compression artifacts. As a result of these analyses... The characteristic features of the image relating to physical screen production are determined, and the image's Technical data regarding the production process is obtained. Render trace detection unit, analysis outputs obtained by the pixel analysis engine. It is the unit that evaluates the image. This unit evaluates the image on a screen hardware. was it produced by or by a graphic editing software? rasterization traces and font drawings in the image are used to determine how it was created. It analyzes the sequence and GPU-generated patterns. Based on the evaluation results... Technical inferences are obtained regarding the method of image creation. The classifier uses the analysis results obtained by the render trace detection unit. It is the unit that enables image classification. This unit uses a pre-trained CNN. With the help of a machine learning model based on the image, it can be rendered as a "real screenshot" or It classifies the image as "manipulated / fake image". Thus, the technical analysis of the image... A verification decision is made based on the results. 4 The reporting module displays the classification result generated by the classifier to the user. It is the unit that converts the image into a verifiable output. This module takes the image source and... Digital data including the technical verification result, confidence score, and review details. It generates a verification report. The generated report shows the verification result of the image. It is presented as output in a way that allows the user to evaluate it. The working principle of the invention The digital image to be verified within the scope of the invention is primarily the image acquisition module. It is uploaded to the system via this method and converted into a processable data format. Then the image The data is examined by the pixel analysis engine. The pixel analysis engine analyzes the pixels in the image. neighbor relationships, subpixel sequences, anti-aliasing properties and by analyzing compression artifacts, the physical display characteristics of the image can be determined. It reveals. The resulting analysis outputs are transferred to the render trace detection unit. The render trace detection unit, whether the image was generated by a display hardware device or a graphics editing process rasterization traces, font to determine if it was generated by software It evaluates the drawing sequence and GPU-generated patterns. The classifier then uses a pre-trained CNN model to analyze the image. It classifies them as either "genuine screenshot" or "edited / fake image". Based on the classification result, the reporting module generates a verification output. This Output, technical verification result regarding the image source, confidence score and review. It is presented to the user as a digital verification report containing the details. Thanks to this structure, the system focuses not on the image content itself, but on the method by which the image was created. It enables the technical detection of fake screenshots by analyzing them.

Claims

1. It is a digital image source verification and fake screenshot detection system. feature;  The digital image uploaded by the user that needs to be verified is a file Data that can be processed by the system by accepting it through the login interface. image acquisition module that converts to format,  The pixel structure of the image transferred to the system by the image acquisition module Examining the image, analyzes the pixel neighborhood relationships and subpixel arrangements. pixels that analyze anti-aliasing properties and compression artifacts analysis engine,  on the analysis outputs obtained by the pixel analysis engine To make an assessment, we will examine the rasterization traces and fonts present in the image. The render trace detection unit analyzes the drawing sequence and GPU-generated patterns.  CNN-based analysis results obtained by the render trace detection unit A classifier that classifies using a machine learning model.  Presenting the classification result produced by the classifier to the user It includes a reporting module that converts this into a verification output.

2. The system, according to Claim 1, is characterized by its technical verification of the image source. digital verification report including result, confidence score and review details It includes a reporting module that generates it.