Skip to main content

Image Integrity Analysis with Amped Authenticate: Search the Web for Reference Images

Reading time: 5 min

Dear Amped blog fellows, welcome to a brand new Tip Tuesday! When you need to perform image integrity analysis with Amped Authenticate, nothing is better than having a reference picture to compare with. Unfortunately, it is common not to have it! That’s why Amped Authenticate lets you search for images from the same device model on the web. Keep reading to find out how!

Why Reference Images Matter in Image Integrity Analysis

If you’re a loyal follower of this blog, then you surely know that digital images carry a lot of information besides pixels. There are metadata, and there is information relative to how the image is stored to file: format, file structure, JPEG Quantization Tables, etc. This kind of information turns extremely useful when your task is to check image integrity. Indeed, if someone claims that an image is an unaltered, camera original file, then we expect its features to be totally compatible with other pictures produced by the declared device model.

Let’s work with an example. As part of an investigation following a car incident, we are asked to assess whether the image below is a camera original file through image integrity analysis. The image is produced by a smartphone, model Sony D5503, commonly known as Xperia Z1 Compact (click here to download the full-resolution file).

Close-up photo of a dark-colored car parked outdoors reflecting an adjacent orange vehicle in its highly polished side panels and windows. The reflection includes surrounding trees, a building, and blue sky with white clouds, highlighting the car's glossy finish and exterior details like door handles and trim.

Using Amped Authenticate to Find Reference Images Online

We fire up Amped Authenticate and begin with filters in the Overview category. The Visual Inspection filter does not raise any concern. We then turn to the File Format filter, which produces this output:

Screenshot of a digital image metadata analysis report showing detailed EXIF data for the file "D12_T_nat_0072.jpg." The data includes file path, MD5 hash, image dimensions (5248 x 3936), camera make and model (Sony D5503), software version, JPEG quality (97), and date/time information. The report also lists JPEG compression signatures, EXIF tag counts, thumbnail sizes, and SHA hashes. A warning in the “JPEG QT Hash” row indicates: "No compression signature available for this camera in the DB."

We are happy to see there are no hard warnings (that would be written in red in the right column). However, we have a soft warning (written in yellow): it says Authenticate’s JPEG Quantization Table database does not contain any compression signature for this camera. Alas, we miss that one! Does that mean that the image has an issue? No! It just adds a layer of complexity to the image integrity analysis. Authenticate cannot confirm that its QTs are the expected ones, simply because this device model is not in Authenticate’s internal DB.

Are we left alone? Of course, not. Amped Authenticate has a tool that allows you to search for images of a given make and model on the web. Just click on Tools -> Search Images From Same Camera Model… and you’ll get this dialog:

Popup window titled "Search Images From Same Camera Model" showing search parameters for locating images based on camera metadata. Fields include "Camera Make" set to Sony, "Camera Model" set to D5503, and date range set to search images taken up to 3000 days ago. Other filters specify a minimum of 10 pictures, resolution of 5248 x 3936, and options to exclude pictures with known software tags or different make/model. Buttons at the bottom read "Search" and "Cancel."

Authenticate will look for images that declare in their metadata to come from a Sony D5503. Not only: it will exclude by default all images whose Exif “Software” tag contains any known software name (because we want good reference files, not modified ones!). In our case, forcing the same resolution of our evidence file would return only two files. So we opt for allowing images of any size, thus obtaining this table:

Search results window displaying a list of images captured with the Sony D5503 camera model, showing metadata columns: Web Photo ID, Owner ID, Image Size, Make, Model, Software version, Date Uploaded, and direct image URLs from staticflickr.com. All images are linked to their respective upload details and vary in resolution, upload date, and software version, offering traceability for forensic or verification purposes.

The table lists images that are freely available on Flickr and tells you their URL, the date they were uploaded, the ID of the user that uploaded them, and more info.

We can select them all, right-click and Download Images, then choose a destination folder. Once they’re on our computer, we can run the Batch File Format Comparison tool, from the Tools menu. This is what we get (we’ve hidden some column of the table for better visibility):

Screenshot of the "Batch File Format Comparison" window in Amped Authenticate, showing a tabular comparison of 15 JPEG images. Columns display data such as filename, format, resolution, thumbnail size, metadata counts, EXIF make and model (Sony D5503), EXIF software version, JPEG quality, JPEG QT hash, chroma subsampling, and whether JPEG Huffman Tables are standard. Highlighted entries and consistent patterns in EXIF data and software hashes are used to assess image authenticity and origin.

Our evidence image details are written in the first row of the table, then other images are listed. For each reference image, values that are different from those in the evidence are highlighted in red. Notice that, while most images have different JPEG QT, there are two at the bottom of the list that match exactly. Interestingly, the penultimate is also the only one that has the same firmware as our evidence image (according to the Exif Software field).

It is worth the effort to load this image as a reference for a deeper comparison. Just right-click on that row and select Load as Reference. We can now compare the two images’ characteristics in detail: let’s begin with the File Format filter:

Screenshot of Amped Authenticate software displaying a side-by-side metadata comparison of two JPEG images taken with a Sony D5503 camera. The evidence image and reference image share similar EXIF data, including JPEG format, quality (97), and software version, but differ in resolution, aspect ratio, and metadata timestamps. Fields like image size, EXIF field count, and SHA values show discrepancies, highlighted with “Different” labels in red, aiding forensic analysis and authenticity verification.

As expected, the reference image has a lower resolution. Is it a supported resolution for this device? A quick search reveals that yes, it is. So that does not invalidate our comparison. We then see we have a different amount of Exif metadata. Let’s check what the Exif filter has to say:

Screenshot from Amped Authenticate showing EXIF metadata comparison between an evidence image and a reference image, highlighting discrepancies. Numerous GPS-related tags such as GPSLatitude, GPSLongitude, GPSAltitude, and GPSTimeStamp appear only in the evidence image. Differences are also noted in thumbnail size, software version, image size, and resolution fields. Fields like SceneCaptureType and CustomRendered are labeled as “Different,” indicating metadata inconsistencies that may suggest image manipulation or different camera settings.

Good! The different amount of metadata is simply due to the fact that the GPS localization feature was probably turned off in the device that captured our reference image. This is a user-configurable option, so we must allow for differences between the evidence and reference file. We can then compare the JPEG Structure filter output, to see whether files are the same even from the structural point of view. This time, we have total compatibility (we don’t expect the offset values, that are the 8-digit numbers written on the left, to be the same; we only expect lines to be in the same number and order).

Hexadecimal comparison view from Amped Authenticate displaying JPEG file structures of an evidence image and a reference image. Both images show sequences of JPEG markers such as SOI (Start of Image), APP1 EXIF, DQT (Define Quantization Table), SOF0 (Start of Frame - Baseline DCT), DHT (Define Huffman Table), SOS (Start of Scan), and EOI (End of Image). The sequence is repeated in both images, indicating standard JPEG formatting, although file offsets differ, suggesting variations in image size or editing history.

We managed to find an image on the web that is strongly compatible with our evidence image. Of course, that does not provide rock-solid proof about our evidence image integrity. We are certainly more inclined towards considering it a camera’s original file than before.

Before concluding, let us just mention that Amped Authenticate features another, even more powerful, tool to search the web for reference material. It is called Search Images from Same Camera Model (CameraForensics)… , under the Tools menu. We’ll dedicate a future Tip to this tool and to show the difference from the one we’ve used today.

Final Note

This week’s takeaway is: comparison with reference material is the best way to go for image integrity analysis! If you don’t have reference images from the same device, you can use Authenticate to search the web for them.


 Marco Fontani

Marco Fontani is the Forensics Director at Amped Software, a software company developing image and video forensic solutions for law enforcement agencies worldwide. He earned his MSc in Computer Engineering in 2010 and his Ph.D. in Information Engineering in 2014. His research focused on image watermarking and multimedia forensics. He participated in several research projects funded by the EU and EOARD, and authored/co-authored over 30 journal and conference proceedings papers. He has experience in delivering training to law enforcement and provided expert witness testimony on several forensic cases involving digital images and videos. He is a former member of the IEEE Information Forensics and Security Technical Committee, and he actively contributed to the development of ENFSI’s Best Practice Manual for Image Authentication.

Subscribe to our Blog

Receive an email notification when a new blog post is published and don’t miss out on our latest updates, how-tos, case studies and much more content!