Looking at the CCTV camera market today, we can see that it is booming in three areas: Hardware, software and, most importantly, demand. While the hardware and software side of the industry has made significant advances in the last decade, as have the requirements of the customer, the planning and installation companies do not appear to be progressing at the same rate. In fact, many integrators still approach CCTV system design and installation the same way they did when CCTV systems first appeared in the market.
Planning and installation procedures remain unchanged, with a mix of site-gathered information and personal experience used to plan the CCTV surveillance project.
The project’s requirements are often based on little more than an estimation of the number and type of cameras needed, and the choice of mounting locations; but such guess work often leads to uncertainty and extra work during the post-installation phase.
The post-installation man-hours required can be significant because of the need for testing, using testing models and/or using mini monitor-testing devices. Many readjustments may be required to correct height, angle of view, focus and lens selection – the camera itself might even need changing.
Many camera companies provide calculation tools for use in the mounting of their products but a lot do not; other than a simple lens calculator that provides little help in choosing the camera’s optimum height or its angle of view.
To understand what a downward-pointed camera can see and cover, I will illustrate its properties with the help of some sketches.
Fig. 1 is a side view of a tilted camera. The inner and the outer red lines represent the upper and lower limits of the camera’s vertical view angle. From this illustration, we can also see the camera’s maximum and minimum horizontal limits at ground level, as determined by the camera’s parameters (lens, sensor) and its mounting parameters (height, angle of view).
So the camera can see only the feet of a person standing on the edge of the camera’s maximum viewing limit and the whole body, but not the feet, of a person standing on the edge of the camera’s minimum viewing limit.
Fig. 2 (2-D top view) sketch shows the exact shape of what the camera can see, or the true field of view, (FOV) covered by the camera at ground level.
Changing the detection level to one other than ground level, will change all of the camera view distances according to the detection level height.
Fig. 3 is a side view of a tilted camera. The inner and the outer red lines represent the upper and lower limits of the camera’s vertical view angle. From this illustration, we can see the maximum and minimum horizontal limits at object level (the object level in this article’s sketches is two metres) as determined by the camera’s parameters (lens, sensor) and mounting parameters (height, pointing angle).
So the camera can see the whole body of a person standing on the edge of the camera’s maximum viewing limit at object level, but only the top of the head when that person is standing on the edge of the camera’s minimum viewing limit.
Fig. 4 (2-D top view) sketch shows the exact shape of what the camera can see or the true FOV covered by the camera at the object level.
By comparing Fig. 2 and Fig. 4 we can see how the FOV size and horizontal distances have been changed by changing the height of detection level from ground level to an object level of two metres high.
Looking at distances referred to in these examples, we can see that the camera can detect a person entering the monitored area at 17.1 metres from the camera but we can only see his face at 8.6 metres.
The other factor that affects the quality of monitoring is the depth of field (DOF). As shown in Fig. 5, it is not enough to detect a person, but this person should also be in focus at the detection distance.
Two important factors that should also be taken into account are recognition and identification. These are related to the camera resolution level which, together with other camera parameters such as lens, height and pointing angle, can result in unacceptable image quality.By looking at the following graphs, we can see that even slight changes in camera parameters can result in very large differences in distances.
We can see in Graph 1 how the camera’s height influences the maximum (max0) and minimum (min0) ground level detection distances.
We can see in Graph 2 how the camera’s height influences the maximum (max2) and minimum (min2) object-level detection distance (object height in this case is 2m).
We can see in Graph 3 how the camera’s pointing angle influences the maximum ground-level detection distance at different heights.
We can see in Graph 4 how the camera’s pointing angle influences the maximum object-level detection distance at different heights (object height in this case is 2m).
We can see from these graphs that small changes in camera height or pointing angle can result in big changes as to what the camera can see.
An article published in a UK newspaper states that about half of the UK’s CCTV cameras are not properly installed, while here in Sweden, a company says that the police are happy with their cameras because they can provide very good identifiable images.
However, when looking at what type of camera they are using, we can conclude without doubt that it is capable of delivering high quality results because it is an 8-megapixel camera. But when talking price, this type of camera may not be the right choice, because if we know what is needed and what the camera can deliver, then we might only need one that is half the price of an 8–megapixel camera.
To conclude, it is obvious that today’s estimation methods cannot visualise what the camera can see during the installation phase, and reaching a level of accuracy depends very much on the installer’s experience and his ability to visualise camera views.
In order to solve today’s camera installation problems, here are two suggestions:
- Many tests and re-adjustments should be carried out during the post-installation phase to reach the required level of monitoring.
- Choose high-end, expensive hardware that can help to reduce inaccuracies.
- This means that today’s planning costs are much higher due to the need for more man-hours to complete the task, as well as the need for more expensive hardware.
- This problem can be solved by using mathematics, not estimation.
- The following example shows how calculations can help to accurately determine how a monitored area can be covered by a PTZ camera. I have chosen to give an example with only two lens settings (min 20mm and max 180mm) and with only two tilt-angle settings (10 degrees and 4 degrees) to make the example easier to understand.
- The sketch shows the top view of the PTZ pan movement arcs. Each arc is related
- to certain parameters but I will not explain them all.
- The outermost arc at 217.3m is where the camera can detect an object entering the monitored area.
- The second important arc at 184.7m is where the camera can detect the face of a person entering the monitored area. For both arcs, the lens setting is at 20mm and the tilt is 10 degrees.
- The third important arc, at 180.1m, is where the camera can detect the face of a person entering the monitored area at a lens setting of 180mm, with a tilt of 4 degrees.
- The fourth important arc at 182.3m is where a person’s features can be detected.
- The fifth important arc at 148.3m is the minimum limit of DOF.
So, to identify a person, he should be within the 180.1m arc so that the face can be seen and, if the number of pixels is enough, a true identification can be made within that arc of 182.3m, and before the 148.3m minimum DOF limit has been crossed.