Video analysis is becoming an ever more important part of many investigations. There are a variety of different analysis methods and techniques that can be used; which method is appropriate for each case is dictated by the footage itself, and multiple other factors may also need to be considered when determining the best way to analyse footage.
The most common reason to analyse footage is to calculate the speed of a collision-involved vehicle. We might also be interested in pedestrian or cyclist movements, analysing the lines of sight which are available to each party, evaluating average traffic speeds, or even tracking the movements of another object entirely in non-collision related investigations, such as a sportsperson or boat.
For comparison in this article, we will show some of the ways to calculate the distance travelled by a vehicle and therefore its speed. Important considerations in choosing the correct analysis method include:
- Movement of the vehicle relative to the camera
- Distance of the vehicle from the camera
- Other objects that are in view
Where a vehicle is moving side to side across a camera view, with its wheels visible and identifiable, a simple desktop analysis using the vehicle’s dimensions can be conducted. For this method to be possible it is very important to know what the vehicle specification is; once we have the length or wheelbase, we can calculate the distance it has travelled in a particular time.
In this example, a reference line has been placed in the footage, aligned on the centre of the front wheel. In the next frame, 0.12 seconds later, the centre of the rear wheel is at the same point, meaning the vehicle has travelled its wheelbase in this time. Once we have the distance and the time, a calculation of the vehicle’s speed can be carried out.
Another method of determining distance travelled relates the position of the vehicle to fixed physical features at the scene, such as street furniture or road markings. There are several possible ways to take measurements of these physical features, including Google aerial imagery or using a police scale plan or survey data. If these measurements cannot be obtained reliably, it will be necessary to visit the scene and take measurements ourselves.
In the first frame here, the front of the vehicle is aligned with a change in road surface (obscured by the imposed red line). In the second frame, the rear wheel is aligned with the start of the cross hatched markings. Measuring between the two points on the road, plus the extra distance from the front of the vehicle to the front of the rear wheel, gives us the distance travelled.
It is important in all methods to consider the viewing angle of the camera and account for any uncertainty in the positioning by applying a range to the calculations. The more precise the measurements can be, the smaller the range needs to be to account for the uncertainty.
The fixed physical features used to position a vehicle might also need to be considered in three dimensions. This is particularly true with vertical objects such as lamp posts; measuring the distance between two points may not always be a case of simply measuring a straight line between the two – it must also take account of the viewing angle of the camera relative to the physical feature and the moving vehicle. The most reliable method is to use 3D laser scan data which provides measurements accurate to within a few mm. It might be possible to obtain such data from a police investigation or, if this is not available, our experts can laser scan scenes themselves.
Combining laser scan data with CCTV allows the relative positions of the camera, the fixed features and the moving vehicle to all be modelled. This can result in highly accurate tracking of a vehicle’s movements and therefore speeds.
In summary, many features of a video recording are considered when deciding which analysis method should be used. In some instances, a desktop analysis can provide an accurate result, but in others it is not possible to calculate a speed without visiting a scene and perhaps performing more advanced enhancement techniques.
Video analysis can involve many varied and complex techniques and should only be performed by an expert with the knowledge and experience to understand how much reliance can be placed upon a recording. The content of the footage is far more important in determining the time an analysis will take, than the length of the recording. Our experts can review a recording or a series of cameras and advise on the best way to effectively carry out any analysis, to provide the most scientifically robust results.