CASE EXAMPLES

CONVERTING A CCTV VIDEO AND DEBLURRING A LICENSE PLATE

Situation

A hit and run is caught on CCTV footage outside a local supermarket, the vehicle is seen leaving the scene. The video evidence team has been tasked with identifying the license plate of the vehicle.

Workflow

The CCTV footage is in a proprietary format and is unable to play on standard media players, so the video evidence team use Amped FIVE’s powerful conversion engine to view the video.

After converting and viewing the footage they can see that the license plate has a blur effect obscuring the details of the license plate. After successfully applying the motion deblur filter they realize that the license plate’s resolution and pixel quality in a single frame is not enough. They decide the best course of action would be to stabilize the video and carry out a frame average. The frame average will remove any noise and enhance the details of the license plate further. This proves successful and after cropping and resizing the license plate they can successfully trace the vehicle back to the owner.

Conclusion

The police are now able to bring the suspect in for an interview.

ANALYZING A MOBILE PHONE VIDEO TO IDENTIFY AN OBJECT

Situation

The police are handed a video recording taken on a mobile phone of an assault occasioning grievous bodily harm.

Workflow

The footage taken on the mobile phone suffers from severe camera shake. The video evidence team use Amped FIVE to apply local stabilization on the area of the fight making the video and the actions of each individual much easier to follow. After reviewing the footage it appears that one of the suspects has an object in one of his hands. Using the filter Spotlighting they highlight this area and track it throughout the fight to see if the object was used. It appears the suspect uses it within the fight and could be what caused the serious injuries in the assault. This new video is exported as evidence and shown to the investigators.

The team is now tasked to try to identify the object. They apply a range selector to the video to select only the relevant frames and begin to analyse them further. By adjusting the contrast in the video, cropping the object and resizing using the smart resize filter they can clearly make out what appears to be an empty bottle.

Conclusion

The police are now confident that the suspect in question used a bottle as a weapon during the assault which caused the injuries to the victim.

UNDISTORTING CCTV VIDEO AND MEASURING THE HEIGHT OF A SUSPECT

Situation

The police are handed CCTV footage of an unknown suspect involved in a high value theft.

Workflow

The police requested the Video Analysts to determine the height of the suspect from the CCTV footage. On review of the footage the experts were able to determine that it suffered from distortion created by the camera lens. Using Amped FIVE they are able to apply the undistort filter to correct this before attempting to analyse the suspects height.

The footage is of high enough resolution and there are enough references on the scene to apply the Measure 3D tool. After the investigators supplied a known measurement from the crime scene the analyst applied the Measure 3D filter to determine the suspects height with a given error.

Conclusion

The police are now confident that the suspect falls within a certain height range and can narrow their search for the suspect.

DEBLURRING A MOBILE PHONE CHAT CONVERSATION TO MAKE IT READABLE

Situation

The police are handed video footage captured on a mobile device. The footage is a recording of another mobile device displaying a chat conversation between a minor and adult.

Workflow

After reviewing the footage, they realize there’s too much shake and motion blur for the chat to be readable. They submit the video to the video evidence team who are able to successfully stabilize the footage in Amped FIVE. They then apply a Motion Deblur filter and are able to successfully create a transcript from the once unreadable footage.

Conclusion

The police now have sufficient evidence to charge the suspect.

REDACTING SENSITIVE DATA AND AUDIO IN A VIDEO

Situation

The Military Police are investigating an individual for an alleged drugs job when they come across body worn camera footage from Afghanistan. Reviewing the footage they witness a murder of a wounded and unarmed terrorist.

Workflow

The Multimedia Evidence team have been tasked with producing two pieces of evidence from the footage. They need to help identify the suspect and also produce a redacted version of the footage to be released to the media.

In order to identify the suspect they select the relevant frames in which the suspect is stationary, adjust the levels, crop the head of the suspect, and then apply a frame average to enhance the detail and remove noise. Finally by resizing the image they have managed to get a clear visual of the suspect.

To create a redacted version of the video they select the relevant frames. Due to strong language and sensitive data they also apply the Redact Audio filter. This enables them to remove any audio required before releasing the video to the media. Finally they sensitize the video by applying Hide Selection to the faces of other people present in the video.

Conclusion

Using Amped FIVE they have successfully managed to release the required footage to the media and create sufficient evidence to identify the suspect.

REDACTING SENSITIVE IMAGES

Situation

An investigator is working on a case involving CSE material. A first-generation image of a victim is seen amongst the material, not known on any database and the investigator needs to send a redacted image to another department responsible for identifying and safeguarding vulnerable victims. Unfortunately, the image is indecent in nature and needs to be censored in order to leave only unique, identifying features whilst removing any obscene content.

Workflow

Using the seamless integration of Griffeye Analyze and Amped Software, the investigator selects the image and loads it into Amped FIVE using the integration plugin feature.

Once the image is opened in FIVE, the investigator can use the Redaction filter within FIVE in order to isolate only the face and distinguishing marks in order for the child to be identified and ultimately protected from further harm. The process of opening the image directly in FIVE from Griffeye means that no further copies of the image are produced, the process is streamlined and a redacted copy can be saved and securely forwarded to the right department instead.

As a result of this quick and easy method of redaction and integration, the identification department can request the relevant information from other agencies whilst avoiding producing any obscene content and further exposure to otherwise traumatic material.

Conclusion

The child is successfully located and safeguarded from further harm and exploitation by the relevant authorities.

DEBLUR SCENE PHOTOGRAPH

Situation

A case involving the downloading of CSA material has reached court. The defendant is disputing the installation of peer-to-peer file sharing software, but photographs taken at the scene of the warrant show the suspect’s computer switched on at the time with the software live and running. Unfortunately, the photograph suffers from a blur disturbance and the time and date shown on the active computer is unclear.

Workflow

Amped FIVE was used to deblur the photograph and allowed for the time and date to be clarified. A full forensic report and an Amped FIVE project was created to show the court the actions taken and to prove no unnecessary editing or planting had taken place.

Conclusion

The suspect pleaded guilty to the offence following this new evidence submitted by the prosecution.

DVR SCREEN CAPTURE TO PRESERVE ONLINE DATA

Situation

An investigator must perform some open-source intelligence, accessing social media in order to obtain information about a suspect in a case. This access is required to be recorded to show steps taken and to preserve data that could disappear from the online world.

Workflow

Using Amped FIVE’s built in DVR Screen Capture, the access and pertinent information are both captured using the desired frame rate and codec, and any lost frames are noted.

Conclusion

The capture was processed straight away within FIVE for any presentation or redaction needs.

COMPARING IMAGES TAKEN AT DIFFERENT TIMES

Situation

Following a deadly collapse of a building, investigators need to assess the cause of the incident. Two aerial pictures acquired years apart have to be compared to assess whether the building had been illegally reshaped in time. The older picture has no colors and suffers from a noise pattern that complicates comparison. Moreover, pictures were taken with different equipment, and the perspective is not consistent between the two.

Workflow

Using Amped FIVE's Fourier filter, the older picture is clarified. Thanks to the Perspective Aligner tool, the perspective of the two images is registered by simply clicking on a few pairs of reference elements that are present in both pictures (road intersections, historical buildings, etc.). The more recent picture is converted to grayscale allowing for direct pixel comparison. Images can then be overlaid, and a difference image is also produced which confirms the good match between elements that remained unaltered in history, while modifications in the questioned building clearly stand out.

Conclusion

By examination of the produced images, changes in the building shape become evident and indisputable. Building owners are prosecuted as responsible for the fatal incident.

A PROJECT THAT MIGHT HAVE TAKEN 2-3 HOURS WITH STANDARD VIDEO EDITING SOFTWARE CAN BE DONE IN MINUTES

We have two licenses for Amped FIVE and found that it has dramatically sped up our workflow. A project that might have taken 2-3 hours with standard video editing software can be done in minutes. We use it for everything from quickly redacting video for public disclosure requests to getting tricky video to play (and creating a playable derivative for detectives) to extracting still images for bulletins. Of course, it does so much more than that. Money can be tight in government. We’ve found that Amped FIVE is the best bang for the buck for video forensic work.
Detective Steve Paxton, USA