Expert audio and video forensics

Case Study

It’s hard to imagine our modern-day world without video surveillance cameras, smartphones, tablet cameras, dashboard cameras, action cameras, and body cams. Thousands of cameras capture thousands of different events. Like unbiased witnesses, they are ready to share what they saw. Thus, when it comes to crime investigations, it’s not surprising that video recordings become an imperative form of evidence. However, it’s not uncommon for a recording’s setting, camera’s technical features, and other factors to impact the quality. Sometimes, there’s no certainty that the video reflects real events.

As a result, the involvement of video forensics is often necessary for a video recording to contribute to uncovering the truth. These experts leverage their expertise and specialized equipment to ensure that video recording can serve as uncontested and informative evidence during legal proceedings. Investigators need to overcome many challenges, and even simple videos can present significant problems. There are dozens of modern digital recording formats and analog recordings. In many cases, law enforcement agencies are limited in their ability to reconstruct this data.

Converting video to a necessary format cannot be reliably achieved by simply using general-purpose consumer software. Innumerable surveillance system models have their own video formats without user-friendly and easily obtainable converter formats. Even if these conversion tools exist, there’s no guarantee that conversion will not destroy the image’s microscopic details that will be critical in the future. An expert with video coding training and reliable video forensic equipment gives you the best chance at an accurate outcome.

Expert involvement is also critical when video is damaged. For instance, a dash cam captured the entire incident while the camera was functional, and the data remains on the memory card. The last recorded video that captured an impact cannot be reproduced using regular tools. In this case, experts can often compile small fragments of video data in the right sequence. This technique can enable the reconstruction of video captured seconds before the impact, which can then be presented during court proceedings.

Retrieval of necessary information often requires more than just viewing the video. For instance, multiple cameras can record the same event. If recordings are reproduced one after the other, it’s hard to match details. It’s much easier to accomplish this by analyzing all recordings simultaneously. But, first, recordings need to be synchronized. Synchronized recordings allow investigators to view events from different angles and capture critical details.

Expert contribution is often required when a video is missing the necessary details. It can happen for various reasons:

• Recording on handheld video devices results in shaky video.
• Recording from a distance yields low-resolution images.
• Rapidly moving recorded objects and slow-working camera autofocus systems result in blurred images.
• Insufficient light and light reflecting in the camera’s direction (backlighting) produces dark images.

The camera’s characteristics, such as resolution and frame dimensions, also impact quality.

Taking all these factors into consideration, it’s not guaranteed that a video can clearly capture a car’s license plate or criminal’s face.

When it comes to forensic video enhancement, it’s naïve to expect a miracle from an expert, similar to stories shown in fictitious crime shows. However, modern video enhancement software can sometimes produce miraculous results.

Forensic Video Enhancement

This complex task is accomplished through two phases. During the first phase, forensic experts evaluate the video to determine whether it is an original or a copy. The term original video is only used to refer to the video captured on the device during recording. Copying can simply mean that the video file was copied on a different device, the device that was not used during video recording. Copying can also refer to editing the file format, such as resampling and recompression. An expert trained in video coding from different devices and video editing can help conclusively indicate the video’s authenticity.

The best way to distinguish between an original and a copy is to obtain a video copy recorded on the same device as the video in question. Discovering recompressed video raises a red flag for a video forensic expert. Video recompression can indicate video editing.

However, recompressed videos are not always fabricated. For instance, all YouTube videos are recompressed, yet very few YouTube videos were edited from the original.

As a result, the most complicated part of authenticity determination is finding the remaining evidence of counterfeit videos. Sometimes, easily observed, simple artifacts can serve as this proof. For instance, a suddenly appearing object in the video frame or hands of a clock that rapidly changed to a new time can prove that the video is a fake.

Professional forgery is not easily detectable without special tools such as modern forensic software. In this case, forensic experts can conclude the video’s authenticity by closely inspecting various qualities of the video, such as video format, coding parameters, file structure, metadata, and, more importantly, verification of the image in every frame. It’s a very tedious and lengthy process. Without performing this analysis, experts can’t be confident that the video depicts real events and not what someone wanted us to see.

From time to time, experts are presented with a new challenge – source camera identification. This is a specialized area of forensic analysis requiring advanced tools and expertise. Experts need to record dozens of frames in a specialized setting using proven devices. After that, specialized software will isolate the camera’s imprint. Camera imprints are an image fragment that cannot be seen with human eyes, and that remains the same in all images. Collected imprints are then compared against imprints on proven authentic videos and photos. This explanation might lead the reader to believe that camera identification is simple, but it’s not true. The captured image’s quality determines the simplicity with which forensic software will discover camera imprints. Poor image quality can even produce invalid data imprints. In order to arrive at the correct conclusion, experts need to have a deep understanding of “imprint” search algorithms, video comparison techniques, and software data interpretation skills.

Forensic expert’s contribution doesn’t stop even after proving the video’s authenticity, enhancing its quality, and ensuring that all stakeholders can view it. Why? Because it’s not enough to see a vehicle on a video. It’s vital to be sure that a specific vehicle is shown, or, at the very least, to prove that the vehicle’s make is within the scope of the investigation.

Objects Identification

Thus, object identification comes into play. In order to solve this issue, experts need to find identifying characteristics using an image. These characteristics need to distinguish an image from thousands of similar images. Next, it’s vital to compare a verifiable object’s characteristics against duplicate object’s characteristics. This job is similar to a child’s “Spot the Difference” game. A forensic expert’s ability to skillfully play this game can impact the destiny of stakeholders involved in video investigation.

Forensic Photogrammetry

During investigation, questions can arise about size and distance. Forensic photogrammetry, a branch of video forensics, can provide the answers. An individual with the knowledge of simple rules of advanced spatial geometry can solve this riddle using established geometric measurement principles and forensic modeling tools. When a computer performs geometric measurements, results will be obtained more rapidly, and data will be more accurate.

However, advanced spatial geometry is not always enough. Luckily, modern computers have the capability to produce 3D models of geographic areas and objects. Forensic engineers can determine several points with known coordinates on video or images, easily calculate coordinates of other points, and determine the distance between these points.

An image can demonstrate the distance between two objects. A video can also determine the time between the events. When distance and time are known, speed can easily be calculated. Cases that involve the determination of guilt after a vehicular accident often require calculation of speed. Accumulated forensic expertise and video forensic methodologies allow experts to determine the speed under different conditions. Many factors are at play, such as the time of day, distance from the subject of the recording, and movement of the camera. However, vehicle speed can often be estimated using forensic measurement methods.

Video forensics can’t solve all riddles. However, experts know that even the tiniest details can make a significant impact during an investigation. Video forensic experts are not magicians, but they still try to work toward reputable and actionable findings.