System for detecting drones / missiles in the video stream. Part one

Part one. Technical description.

Introduction

To understand the basics of the system for video detection of drones and missiles in a video stream, it is necessary to start with a description of existing technical solutions, their advantages and disadvantages. (Note: All information is from public sources and does not contain confidential data or detailed technical information, please contact software developers or hardware manufacturers for more details).
Also, this article does not discuss the need to introduce such a system, since this need is clear to anyone who is a little interested in world news. We will not describe here the threat itself, which has long been known to professionals and there is no need to list all the risks once again.

Description

Any anti-drone (unmanned aerial vehicle) system consists of three components: drone detection, drone identification and tracking, and drone countermeasures.
Drone detection is the first and most necessary step, without which further technical measures are impossible. At the same time, the earliest possible detection of a drone is already partly the solution to the third stage of counteraction, since regardless of further steps and measures, the greatest response time is the key to the effectiveness of the entire system.

Existing drone detection systems include radar systems, radio signal detection systems, sound detection systems, and video stream detection systems. Let’s take a closer look at the advantages and disadvantages of each system.

A historically well-known way to detect drones and missiles is a radar installation. There are many radar technologies available, full details are available on the Wikipedia website. The advantages of using radars are the greatest range and reliability of threat detection, the possibility of their use in any weather conditions (!) and time of day. Regardless of the frequencies or type of radar used (including the world’s most advanced variable angle phased array radars), all radars are based on the same laws of physics, and there has been no breakthrough in this area in the last century.
Therefore, for each type of radar, the main limitation is the area (size) of the flying object, which determines the range and accuracy of object detection. Various methods are currently being used to reduce drone reflectivity for radar, but this technology is very expensive, cannot be used in mass production, and the size of the object is still the most significant factor.

Also, the size of the drone directly affects its carrying capacity and, as a result, the degree of its danger. Therefore, the direct relationship between the size of a flying object and the possibility of its long-range detection makes the radar the most reliable and stable means of detecting a threat. But, like any technology, radar has a number of disadvantages.
Not a technical, but organizational disadvantage is its cost and, as a result, the impossibility of widespread installation. The technical limitation is the terrain, which inevitably creates dead zones for the radar. A clear example is the use of low-flying drones, missiles and even aviation in a number of armed conflicts over the past decades, examples of this use are widely available on video in instant messengers and streaming platforms.

The next way to detect drones is radio detection of drone control signals or drone video transmission signals to the operator’s console. The method, which is less expensive than radar, even sometimes (mainly for mass-produced drones) allows you to determine the UAV model from the imprints of its control signals. The main disadvantage of this method is its absolute uselessness when flying a drone (or rocket) that is not controlled by an operator, but uses GPS coordinates or an even more technologically advanced inertial system.

The presence of a satellite orientation system in space today is provided even in the simplest and most inexpensive drones. Inertial systems that cost tens of thousands of dollars a decade ago have also become widely available with the development of commercial drones. As for missiles using various navigation methods from thermal imagers to active methods or passive radars, the radio detection method is not applicable in principle.

Systems for detecting drones by their sound prints, in other words, by the sounds of propellers or engines, are currently only being developed. The devices we tested did not work at all (subjective remark, but based on personal testing experience), but this technology is developing and may become more reliable in the near future. Of the physically insurmountable limitations of this technology are its fundamental inoperability.

A feature within the boundaries of urban development, which is associated with an abundance of sounds. And also the unreliability of drone detection outside the city due to the presence of various noises, such as wind, sound attenuation with a distance that is practically not available at a height of a couple of hundred meters (although this height still does not allow radar to detect a drone) and noise reduction of modern drones.

So, let’s take a look at the last (of the physically available at the current level of technology development) detection methods — detecting a drone in a video stream. The disadvantage of this method is the small detection range of the drone due to a number of factors. The first limitation is another law of physics of decreasing the size of an object with its removal from the camera lens. The second limitation is a non-public shortcoming of existing video analytics systems associated with the inability to distinguish objects close to the camera from objects far from the camera.
As a result, for the matrix of a digital video camera, a drone flying a few hundred meters and a fly flying a few meters from the lens have the same size. It is also worth noting that to date, there are no available systems for video detection of a drone by a video camera at all. None. A simple search query will cast doubt on this claim, as the internet and the websites of many well-known manufacturers of drone detection equipment offer such systems.

However, upon closer inspection, it turns out that all of these systems offer tracking, not drone detection. That is, they do not detect a drone and do not work as a separate system, but use coordinates from a radar or a radio detection system (both sources, except for a phased array radar costing tens of millions of dollars, give very approximate coordinates) to detect a drone and only then aim at him a video camera at known coordinates.
That is, this technology does not allow the autonomous drone detection system to work in the video stream, but requires the presence of a radar or radio detection of a drone (which in some cases does not work at all) to detect a drone.
As a result, such systems can be called drone tracking systems based on the coordinates received from other technical methods of detection, but not a system for detecting drones in a video stream, none of the systems offered by global manufacturers today works autonomously.

Now let’s look at the advantages of a drone (missile, aircraft) detection system in a video stream. The main advantage of this method of determination is, oddly enough, its cost. The use of conventional video cameras (both existing and newly installed) as equipment for drone detection allows you to deploy such systems within a company, city or country.
Ensuring the life and health of people is always the highest priority for any company or state. But the impossibility of implementing such a decision due to the limited budget always makes its own objective adjustments to our capabilities.

The drone detection system in the video stream can also detect drones in the city, which, in principle, is not available for a radar that has a cost thousands of times higher than the cost of a video detection system.
The radio detection system is also often blind in urban areas, since, firstly, it has a very high radio noise level (you can imagine the number of WiFi routers in one city area), and secondly, it determines the signatures of only drone control systems known to it, leaving only self-assembly systems without attention radio control (this statement is not true for more professional radio detection systems, but their cost of tens of thousands of dollars per system does not allow it to be used in a mass installation and levels out its advantages, not to mention its uselessness when detecting drones flying along GPS coordinates) .

As we noted above, the video detection system also has limitations associated with the impossibility of distinguishing distant objects from closely flying insects. At the moment, many developers of video analytics systems claim a solution to this problem, but their technical solution is to increase the size of a moving object, which eliminates false alarms associated with flying insects, but at the same time reduces the possibility of analyzing distant objects.

Simply put, they increase the size of the area that can trigger a video analytics system alarm and, as a result, a moving object (car, person, aircraft or drone) that is large but distant from the video camera is not detected by the system.
This technological lag of video analytics systems from currently inexpensive high-resolution video cameras (up to 8K) in mass production is designed to be solved by the Pterygota Marker, which allows you to distinguish close-flying objects from distant ones and makes it possible to mass-produce video detection technology for drones (aircraft or missiles) in the video stream.

Thus the use of the Pterygota Marker can help realize the capabilities of already existing low-cost high-resolution video cameras both in drone detection systems and in other areas of video analytics systems, for example, in systems developed by car manufacturers all over the world, systems for analyzing traffic conditions or various accidents at work or eliminating rare, but financially costly false positives of various machine vision systems.
Such systems are now widely used in various industries, on the basis of such a system Tesla autopilot is built, sorting and quality control systems at industrial enterprises, and so on.

Summing up the choice of UAV detection systems, it is worth noting the following. The only method available for mass adoption is the video detection method. This method provides an optimal balance between the stability and reliability of drone detection and its cost, which is always an important factor in the creation and implementation of mass spaced systems.
Such a system is practically the only method for detecting drones in the city and can be used both independently and in conjunction with other systems and types of equipment.

Long-range detection of drones is undoubtedly provided by radars, and detection of drones at high altitude is possible only with the help of radars. A drone video detection system can replace or supplement radars for detecting flying objects at low altitudes and is indispensable in urban areas.

Next article: System for detecting drones / missiles in the video stream. Part two.

Next article:

System for detecting drones / missiles in the video stream. Legal support and certification.