"Clone Detection - Image" This study advances the field of forensic image analysis by addressing the limitations of current techniques. By integrating the Structural Similarity Index Measure (SSIM) and Error Level Analysis (ELA), our developed tool offers a more comprehensive, interpretable, and efficient approach for identifying image manipulations across diverse applications. In comparison to existing methods such as FotoForensics and FaceForensics, our approach: Enhances accuracy through the combination of structural and error-based evaluations.
Offers visual interpretability rather than merely providing a classification label.
Functions effectively with various image formats, minimizing dependence on compression artifacts.
Demands fewer computational resources, thereby increasing accessibility for practical applications.
These advancements significantly contribute to the improvement of forensic image analysis techniques, safeguarding the integrity of images in fields such as forensics, journalism, cybersecurity, and digital content verification. Technologies Used The development of the tool leverages a combination of programming frameworks, image processing libraries, and forensic analysis techniques to perform image clone detection. The key technologies used in this study are: Programming Language and Framework Python: The primary programming language used due to its extensive libraries for image processing, machine learning, and web development. Flask: A lightweight web framework used to create the web-based interface for image upload, analysis, and result visualization. Image Processing Libraries OpenCV: Used for image reading, conversion, resizing, difference calculations, and color-based visualization. PIL (Pillow): Used for handling image operations, including saving images at different compression levels and performing image enhancement techniques. NumPy: Utilized for numerical operations, such as matrix manipulations and pixel-wise image analysis.

Error Level Analysis (ELA)
ELA serves as a forensic technique for image analysis that identifies variations in image compression levels, thereby exposing areas that may have been manipulated or edited. Digital editing can result in certain sections of an image exhibiting different compression artifacts, which can signal potential alterations. Mechanism of ELA
ELA operates on the principles of JPEG compression, which eliminates specific details based on the image's content. The procedure consists of the following steps:

1. Convert Image to RGB Format
   The initial image is imported and transformed into RGB format to ensure uniform processing.
2. Re-save the Image with Slightly Reduced Quality
   The image is then saved again in JPEG format at a designated quality level. This action introduces minor compression artifacts that assist in uncovering inconsistencies. The JPEG quality selection is adaptive in our approach, based on the image's variance (sharpness level):

• Low-detail images: 50% quality
• Medium-detail images: 70% quality
• High-detail images: 90% quality

3. Calculate the Difference (Error Level) Between the Original and Resaved Image
   The pixel-by-pixel difference between the original and the resaved image is computed using the formula:
   ELA Image = Original Image - Resaved Image If the image remains unaltered, the differences will be consistent throughout. Conversely, if certain areas have been modified, they will exhibit more pronounced variations.
4. Enhance the Differences Using CLAHE and Sharpening
   CLAHE (Contrast Limited Adaptive Histogram Equalization) is utilized on each color channel to amplify subtle error-level differences. A sharpening kernel is then applied to accentuate the edges of the modified regions.
5. Thresholding and Highlighting Edited Areas
   A binary threshold is implemented to underscore significant changes. Areas with elevated ELA intensity values are marked in red, indicating likely modifications. Applications of ELA
   Detection of spliced images (segments copied from other images). Identification of airbrushed or retouched images. Exposing cloned regions in an image. Structural Similarity Index (SSIM) Objective of SSIM SSIM measures the similarity between two images by comparing their structural, luminance, and contrast differences. Unlike traditional pixel-wise difference methods, SSIM models the human visual perception of images, making it more effective for detecting changes. Analyzing SSIM Scores SSIM = 1.0: The images are identical. SSIM > 0.9: The images are very similar with minimal differences. SSIM ≈ 0.5 - 0.8: The images have high amount changes. SSIM < 0.5: The images are significantly different. SSIM Difference Image The difference image produced by SSIM highlights the regions where the two images differ. The difference map is colorized using a heatmap (Jet colormap) for better visualization.

METHODOLOGY This study presents a dual-phase forensic methodology aimed at detecting image clones and edited sections through the use of the Structural Similarity Index Measure (SSIM) and Error Level Analysis (ELA). The approach encompasses both image analysis and comparative assessment between images, establishing a comprehensive framework for the identification of duplications or modifications in digital photographs. Image-to-Image Analysis using SSIM and ELA This methodology is intended to evaluate two images and ascertain whether one is a duplicate or a modified version of the other. It employs a combination of the Structural Similarity Index Measure (SSIM) and Error Level Analysis (ELA) to emphasize differences based on structure and compression. Step 1: Image Preprocessing Both images are resized to ensure uniform dimensions. They are converted to grayscale to eliminate color variations and concentrate on structural discrepancies. Step 2: Calculation of Structural Similarity Index Measure (SSIM) The SSIM algorithm is utilized to assess the two images concerning luminance, contrast, and structure. A similarity score ranging from 0 to 1 is calculated: signifies identical images, Lower values indicate alterations. Step 3: Creation of SSIM Difference Image The SSIM difference image is transformed into a heatmap using OpenCV’s COLORMAP_JET, enabling users to visually pinpoint cloned or altered sections. Step 4: Validation through ELA ELA is independently applied to both images to identify compression-related differences. The ELA images for each input provide further forensic validation by highlighting areas where compression artifacts may indicate potential modifications. Step 5: Interpretation of Results and Decision-Making The tool displays: SSIM Similarity Score SSIM Difference Image (Heatmap) ELA Images for Both Input Images This integrated approach offers forensic analysts a comprehensive analysis, thereby improving the precision of clone detection. Image Analysis using Error Level Analysis (ELA) This phase concentrates on pinpointing altered regions within a single image through Error Level Analysis (ELA). The procedure consists of the following steps: Step 1: Image Preprocessing The uploaded image is transformed into RGB format using the Python Imaging Library (PIL) to ensure it is suitable for forensic analysis. Subsequently, the image is resaved at a reduced JPEG quality, which is adaptively determined based on the image's variance, to introduce compression artifacts. Step 2: Error Level Calculation The resaved image is analyzed in comparison to the original image by computing the pixel-wise differences. Regions exhibiting significant error levels may indicate modifications or cloned areas. Step 3: Enhancement of ELA Output for Improved Clarity To enhance the interpretability of the results, various enhancement techniques are employed: Contrast Limited Adaptive Histogram Equalization (CLAHE): This technique improves localized contrast, thereby enhancing the visibility of manipulated regions. Sharpening Filter: A convolution kernel is applied to accentuate altered areas. Thresholding and Red Highlighting: A threshold is applied to the ELA output, with manipulated regions highlighted in red for improved visualization. Step 4: Interpretation of Results The final ELA heatmap is produced, with suspicious areas highlighted in bright colors. The ELA findings are presented to the user, enabling forensic investigators to scrutinize the potentially edited sections of the image. Image URL using Error Level Analysis (ELA) Follow the same steps as like image analysis, but user enters the URL of the image instead of uploading the image

Conclusion This study concentrated on identifying image alterations through the use of the Structural Similarity Index (SSIM) and Error Level Analysis (ELA) within an image-to-image analysis framework. The main goal was to create and assess a tool capable of accurately pinpointing edited areas in an image by comparing it to its original version. This research tackled the increasing issue of digital image manipulation and the necessity for dependable forensic methods to detect such changes. The tool was designed to

• Compare an original image with its edited version to detect modifications.
• Calculate the SSIM score to evaluate structural similarity and produce an SSIM difference image that highlights altered sections.
• Conduct ELA analysis to reveal inconsistencies in compression levels and uncover manipulated areas.
• Offer a user-friendly interface for forensic analysis, ensuring precise detection of altered images. Key Findings and Results The developed tool effectively illustrated that the integration of SSIM and ELA significantly improves the accuracy of detecting image forgery. The main findings include:
• SSIM Analysis: The SSIM score served as a quantitative indicator of image similarity, with lower scores indicating substantial modifications. The SSIM-difference image effectively showcased the areas where the image's structural integrity was compromised.
• ELA Analysis: The heatmap generated by ELA successfully identified regions with varying compression artifacts, especially in edited or spliced areas. This technique proved effective for detecting changes in JPEG-compressed images, where modified regions exhibited distinct brightness differences.
• Compression Impact: The effectiveness of ELA was affected by the level of JPEG compression. Higher-quality images retained more details, resulting in improved accuracy, while heavily compressed images diminished ELA's effectiveness due to uniform compression artifacts. Contributions of the Work This study has made several significant contributions to the domain of image forensic analysis:

1. Integration of SSIM and ELA for effective detection: By utilizing both methodologies, the tool offered numerical data (SSIM score) alongside visual representations (SSIM difference image and ELA heatmap) for identifying tampering.
2. Improved visualization of changed areas: The inclusion of an SSIM difference image and an ELA heatmap enabled precise identification of changed areas.
3. Enhanced usability for forensics: The application was designed to be simple and efficient, making it suitable for use in digital forensics, journalistic fact-checking, and educational research.
4. Empirical validation of the impact of compression: The study emphasized the influence of image compression on ELA accuracy, providing important insights into the optimal conditions for identifying manipulations. Overall, this work contributes to the field of digital image forensics by proposing an effective and understandable approach for recognizing edited images, thus increasing the reliability of forensic and verification processes.