Collision investigation: CCTV playback and validation using a light board
It’s one of those frustrating myths of TV crime drama that a traffic collision often makes for an easy solve. The crash occurs within full view of several active CCTV cameras, leaving the investigators to simply and swiftly retrieve the footage. They cross-reference the multi-angle, high-resolution images with admissible dash-cam or seized CCTV to calculate speed. Faced with the incontrovertible evidence, the suspect confesses, and the courts do the rest.
But as anyone working in law-enforcement will know, a collision investigation that smooth is a rare thing indeed. If a prosecution using CCTV evidence is to ensue, the reality is that investigators face a maze of potential obstacles, starting with the very first step. Retrieving the CCTV footage and being able to replay it is a complex job and often taken for granted, but verifying the data it produces can be an even knottier procedure.
CCTV cameras record at varying rates of frames per second, meaning that it is often difficult to accurately determine the speed of a vehicle from seized footage. The Frame Interval Timer, more colloquially known as the ‘lightboard’, was developed by the Home Office to verify and calibrate the frame interval of the original footage. Once this has been calculated, investigators are significantly better equipped to extrapolate very accurate driver speeds, thus leading to stronger evidence and, eventually, more reliable convictions. The problem is that using a lightboard and extracting the data it produces is a painstaking procedure. In response to this issue, Sira have developed a tool to automate a large part of the lightboard process, leading to decreased investigation time and cost.
In this article, we will discuss the issues surrounding the use of CCTV and lightboard technology to assist collision enquiries, and take a look at Sira’s Lightboard Reader as a way of saving time and resources during those investigations.
CCTV and COLLISIONS: The current position
For decades, CCTV footage has been used to aid the reconstruction of collisions. Initially, if a police force needed analysis of CCTV in the context of potential prosecution, they’d need an experienced scientist. But poor quality of recordings, impenetrably complicated VCRs and the need for specialist hard- and software hindered investigation teams in their attempts to gather useful information from whatever footage was available.
As technology outgrew the VCR however, digital CCTV proliferated at a rate of knots. From the late 2000s, CCTV systems - including commercial, domestic, vehicle-mounted and wearable units - became viable, cheap and increasingly ubiquitous. And although more footage invariably means a greater chance of hard evidence when it comes to solving crime, for the collision investigator this explosion in CCTV use can be a double-edged sword.
Before expert analysis or reconstruction of a collision is even on the cards, CCTV evidence can only be of value if the skills and equipment to extract and playback the material are present. Because of the sheer numbers of units and CCTV manufacturers, all of whom use different software to recover and replay the footage they produce, there are often significant delays. Listening to the police and staff at the Home Office it became apparent to us at Sira that was a dire and urgent demand for a universal CCTV viewer, which is why we developed SiraView. Our universal CCTV viewer enables law-enforcement agencies to play many formats with a single user interface. It’s easy to use with minimal training, is capable of dealing with thousands of CCTV formats, preserves evidential quality, and includes forensic export with data dump.
The forensic export with data dump exporter converts the video to a series of still images (BMP or JPEG), along with a CSV data file containing a timestamp for every frame extracted from the embedded metadata. SiraView’s forensic export gives easy access to large quantities of timestamp data so collision investigators can perform large quantities of statistical analyses relatively quickly. But if the CCTV doesn’t offer adequate metadata for a meaningful calculation of impact speed, investigators have a problem.
This is where the lightboard comes in. The lightboard operates effectively as an optical clock that displays time at a precision of 0.001s through the sequence of LEDs illuminated at any given point in time. It gives law-enforcement agencies a way to determine the actual frames-per-second rate (FPS) from any given camera, and therefore extrapolate vehicle speed to a very high level of accuracy. Regardless of which lightboard is used, the methodology remains the same.
The lightboard is a timing device usually comprised of two banks of Light Emitting Diodes (LEDs) positioned vertically and horizontally. The first bank consists of a grid of 10 rows of 4 LEDs where each row is numbered from 0 to 9.
The LEDs of the first column illuminate one at a time from 0 to 9 at intervals of 1 second.
During each 1 second period the 10 LEDs in the second column follow the same sequence, at intervals of 0.1 second.
During each 0.1 second period the 10 LEDs in the third column follow the same sequence, at intervals of 0.01 second.
During each 0.01 second period the 10 LEDs in the fourth column follow the same sequence, at intervals of 0.001 second.
The lightboard is designed to be filmed by a CCTV camera from which evidence needs to be verified. After filming a sample, the user will note the time data using frame-by frame playback. From there, frame intervals can then be calculated.
The second bank of LEDs that completes the lightboard is identical to the first though oriented horizontally. When in operation an infra-red communication link between the two banks ensures that they operate in synchrony. The purpose of the two banks is to identify any effect caused by a rolling shutter and where possible, quantify any error that it may introduce.
Using a lightboard enables investigators to establish frame intervals and to test the whole system from the camera lens right through to the recording equipment, playback software, and ultimately the analysis process, without expert understanding of the CCTV system in question.
Process for reading the CCTV lightboard
When an investigator uses a lightboard, a sample of footage of the board in operation is recorded by the CCTV camera in question. The Home Office recommends a recording sample of 10 minutes. The footage of this test must then be played back, frame by frame. The time from each sequential frame is recorded on a spreadsheet, allowing a calculation of the frame rate of the camera.
Clearly, if done manually, this stage will take some time. A 10-minute clip playing at 25 frames a second will mean it’s someone’s job to take 15,000 individual readings. If it takes about 10 seconds to record each reading, it will take an investigator close to a week to tabulate all the data.
Automatically reading a CCTV light board
Sira started work developing a piece of software to read the data in a fraction of the time. We discussed the requirements for such a tool with the Institute of Traffic Accident Investigators, various police collision investigators, and Forensic Collision Investigation and Reconstruction Ltd who have been closely involved in the development of a Home Office-supported lightboard.
What evolved was the outline specification for the SiraView Lightboard Reader. We undertook the development of a reader that would:
- be easy to set up and configure; able to identify erroneous readings as they are taken
- work on any reputable lightboard
- export data alongside existing metadata
- export data in a form that can be easily analysed by collision investigators
- carry out the 15,000 readings in minutes rather than days.
Our reader is configured to look for pixel changes showing when a timing LED comes on, thereby converting the lightboard information into digital data that can in turn be read and analysed without the need for time-consuming input of each individual reading into a spreadsheet. The user starts the process by manually calibrating the reader by telling it where on the screen to look for the individual LEDs of the lightboard. Once the software has locked these co-ordinates, the footage is run, and the software makes its readings by determining whether each LED is on or off. This speeds up the whole process with no loss of integrity as all the data can be manually spot-checked.
It should be stressed that this does not encroach on the expertise of a collision investigator analysing the data. SiraView Lightboard Reader is a tool to extract data from a light board in several ways and display it alongside data extracted from the CCTV metadata. It will remain the responsibility of the collision investigator to check the veracity of the data, and the implications of the data presented to the court.
For example, the dataset should always be compared to the incident footage. Any anomalies in the dataset should be explored, and the collision investigation will still be primarily responsible for the evidence presented to the court. It is not envisaged that the software presents a ‘click and forget’ approach, merely a tool to assist the collision investigator.
Trials and beyond
The beta version of the software was sent out to about a dozen collision investigators to trial. So far, the results are very encouraging. Sira have had a number of suggestions as to how the user interface can be changed to make it more intuitive to. Comparisons with results taken manually have shown excellent correlation.
Our next step will be to implement the suggested changes and continue to make the software simpler to use and more robust. Following that, we will initiate another round of trials to ensure our newest product brings the expected benefits and is of sufficient quality to make it a 'must have' tool in the collision investigator's bag.
Sira plan to release the Lightboard Reader as an add-on to the full version of SiraView in 2018. We anticipate the add-on to start in the region of £325 + VAT, with discounts for multiple copies.