RU2592712C1 - Method and system for determining vehicle speed - Google Patents

Abstract

FIELD: data processing; transportation.

SUBSTANCE: group of inventions relates to traffic control, namely to determination of vehicle speed. Method and system for determining vehicle speed consist in fact that image of road surface and vehicles located on it is formed using video camera, and with help of electronic computing device average path of vehicle is formed, and then with help of statistical analysis of data on vehicle position in projection on plane of road surface at moments of fixation of image frames resultant vehicle speed is estimated.

EFFECT: high accuracy of determining speed without restrictions on vehicle view of observation, simple design.

7 cl, 4 dwg

Description

The invention relates to the field of regulation of the movement of road transport, and in particular to methods and systems for determining the speed of a vehicle, and can be used to control the observance of speed limits by drivers of vehicles.

A known method of determining vehicle speed using locators described in patent RU 2237266, in which they emit electromagnetic radiation in the direction of moving vehicles and receive the reflected signal with a modified frequency due to the Doppler effect. By changing the signal, the vehicle speed is calculated.

The disadvantage of this method is that it cannot be used in fully automated complexes for fixing violations of traffic rules, since the measurement of speed in it is not accompanied by automatic identification of the vehicle: locators are used by traffic police who decide on which vehicle refers to the measurement made using this method speed.

A known system for determining the speed of a vehicle, consisting of locators and video cameras, described in patent RU 2382416, in which locators are used to determine the speed and position on the roadway of vehicles, and video cameras are used in combination with an electronic computing device to identify state registration plates of vehicles funds. In this system, the readings of the locators are automatically compared with the images of vehicles formed with the help of video cameras, thereby identifying vehicles that violate the high-speed mode.

The disadvantage of this system using cameras and a locator is that although it is able to work automatically without operator intervention to identify high-speed offenders, it is complex and, therefore, not accurate enough, since pairing the radar zones on the road with the image of vehicles a difficult task, and in practice, in the case of intersecting trajectories of vehicles, the measured speed may not be correctly compared with the identified vehicle.

A common drawback of the analogs described above, using radar to determine the speed of a vehicle, is the availability of available detectors, which pre-signal the drivers of vehicles that a measuring-control complex is functioning along the way. This reduces the degree of control of compliance with traffic rules.

A known method of determining the speed of a vehicle using a single camera and an electronic computing device described in patent RU 2419884, in which the position of the state registration plate of the vehicle in the image from the camera is automatically allocated when performing the procedure for identifying the vehicle registration number. The method is based on comparing the coordinate space of the image formed by the video camera with the coordinate space of the roadway along which the vehicle is moving. Since the position of the registration marks is fixed on successive frames of the video stream, the speed is defined as the offset along the roadway obtained after converting the displacement of the state registration plate of the vehicle from the camera image space to the roadway space divided by the time elapsed between registration of the offset frames.

This method of determining the speed of a vehicle using a single camera is passive from the point of view of radio emission, therefore, the location of the measuring system cannot be detected remotely using household automotive radar detectors. In addition, identification by the number plate of a vehicle makes it possible with almost 100% probability to identify a vehicle of an intruder, even if several vehicles are present in the camera’s field of view. However, this method relies on a number of assumptions set forth in the claims, which are not satisfied for all vehicles, which can lead to a significant increase in speed measurement for such vehicles. This method is based on the assumption that the plate of the state registration plate of the vehicle is not deformed and not dirty, which allows you to accurately determine the rectangular region of the frame of the plate, and based on the selected quadrangle accurately determine the position of the projection of the vehicle onto the plane of the roadway. In practice, there are examples when determining the exact position of the plate frame of the state registration plate of a vehicle is difficult due to deformation of the plate as a result of an accident. Most often, this arises due to deviations from the regulations and GOSTs, which oblige vehicle drivers to monitor the cleanliness and serviceability of the license plate with a state registration mark. In particular, in the Russian Federation, the fastening of plates with state registration marks to vehicles is regulated by GOST R 50577-93. However, violations of the rules and GOSTs for the appearance and mounting of license plate plates are legally in no way associated with violations of the high-speed mode and should not be interpreted by control and measuring systems in such a way as to fix excess speed. Another disadvantage of this method is that it does not apply statistical processing of data on the position of the number plate of the vehicle obtained on the set of frames of the video stream that record the passage of the vehicle in the field of view of the video camera. Use only the first and last frame, which reduces the accuracy of measuring the speed of the vehicle.

Closest to the claimed invention is a method for determining the speed of a vehicle, described in patent GB 2503328, which is similar to the method of the previous analogue described in RU 2419884, but differs from it in that in order to determine the speed of the vehicle in the field of view of the video camera there must be a contact point of the wheels vehicle with roadbed. This method is selected as a prototype of the claimed invention.

The disadvantage of the prototype method is that not in all angles of view of the vehicle, the point of contact of the vehicle with the roadway is visible (for example, see Fig. 3.) Another significant drawback of this method is that the relative error in measuring the speed in it reaches 8%, judging by the graph in the figure, i.e. when the vehicle speed is 100 km / h, the absolute error is 8 km / h, which is unacceptable in the certification of modern measuring means of vehicle speed in the framework of the legislation of the Russian Federation.

In addition, the prototype method, which proposes to use a stereo system of two cameras for measuring speed, allows you to more accurately assess the position of the projection of the license plate on the plane of the roadway compared with the analogue method described in patent RU 2419884, respectively, more accurate are obtained stereo cameras measure speed. However, the use of two chambers instead of one leads to the fact that the measuring complex becomes significantly more difficult from a technical point of view, which leads to its appreciation both in production and in operation. It should also be noted that in the prototype method, as well as in the analogue method described in patent RU 2419884, measurement statistics for a plurality of frames of a video image are not used, which reduces the accuracy of determining the speed.

The objective of the claimed invention is to provide a method and system for determining the speed of a vehicle with increased accuracy in determining the speed; without restrictions on the view of the vehicle; as well as with a simpler and therefore cheaper design due to the use of a single video camera and the absence of active emitters.

The problem is solved by creating a method for determining the speed of a vehicle, in which using a video camera to form an image of the roadway and vehicles located on it, and using an electronic computing device:

- at the preliminary calibration stage, a correspondence is established between the pixel coordinates of the image points formed by the video camera and the metric coordinates of the roadway points, while for several image points and the corresponding roadway points, data on the measured distance from the image point to the corresponding road point is obtained and used canvases;

- at the preliminary calibration stage, the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway are also calculated;

- determine the fixation time for each frame of the image formed by the camera;

using the correspondence between the coordinates of the points of the image formed by the video camera and the coordinates of the points of the roadway, and also using the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway using an electronic computing device:

on each frame of the video sequence containing the vehicle, the position of the elements of the state registration plate of the vehicle and the coordinates of the characteristic elements of the vehicle in the vicinity of the license plate are determined;

determine a vehicle model, which determines the coordinates of the characteristic elements of the vehicle of this model relative to each other and the height of the characteristic elements of the vehicle of this model above the plane of the roadway;

statistically determine the characteristic points of the vehicle with which you can most accurately determine the position of the vehicle on the plane of the roadway;

recalculating the coordinates of the image of the characteristic points of the vehicle in the coordinates of the orthogonal projection of the characteristic points on the plane of the roadway, using information about the height of the characteristic points above the roadway;

statistically distinguish the trajectories of characteristic points, the movement of which most closely matches the model of a solid body, and form from these trajectories an average trajectory of the vehicle, reducing the random positioning error;

using statistical analysis of data on the position of the vehicle in the projection onto the road surface at the time of fixing image frames, the resulting vehicle speed is estimated, compensating for the random error by multiple positioning on successive frames of the video stream.

In a preferred embodiment of the method, in the calibration step, a correspondence is established between the pixel coordinates of the image points formed by the video camera and the metric coordinates of the roadway points, while for several image points and the corresponding roadway points, the distance from the image point to the corresponding one is measured using a laser range finder points of the roadway; at the calibration stage, using a laser range finder, measure the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway.

In a preferred embodiment of the method, a vehicle model is determined, the state number of the vehicle is determined on the number plate and a request is sent to it in the database of vehicles registered in the traffic police connected to the electronic computing device, and in response to the request, it is obtained from the database vehicle model data.

In a preferred embodiment of the method, a vehicle model is determined, and a request is sent to the vehicle database located in the memory of the electronic computing device and containing data about the vehicle models and the corresponding coordinates of the characteristic elements of the vehicles in the vicinity of the license plate, and in response upon request, data about the vehicle model are obtained from the database.

The problem is also solved by creating a system for determining the speed of a vehicle, comprising a interconnected video camera and an electronic computing device, the video camera configured to form an image of the roadway and vehicles on it; and the electronic computing device is configured to:

- at the preliminary stage of calibrating the installation of correspondence between the pixel coordinates of the image points formed by the video camera and the metric coordinates of the roadway points, while for several image points and the corresponding roadway points, receiving and using data about the measured distance from the image point to the corresponding road point canvases;

- at the preliminary stage of calibration, also calculating the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway;

determining a fixation time for each frame of the image formed by the video camera; in addition, the electronic computing device is configured to:

on each frame of the video sequence containing the vehicle, determining the position of the elements of the state registration plate of the vehicle and the coordinates of the characteristic elements of the vehicle in the vicinity of the license plate;

determining a vehicle model, and from it determining the coordinates of the characteristic elements of the vehicle of this model relative to each other and the height of the characteristic elements of the vehicle of this model above the plane of the roadway;

statistically determine the characteristic points of the vehicle with which you can most accurately determine the position of the vehicle on the plane of the roadway;

recalculating the coordinates of the image of the characteristic points of the vehicle in the coordinates of the orthogonal projection of the characteristic points on the plane of the roadway, using information about the height of the characteristic points above the roadway;

statistically identifying trajectories of characteristic points, the movement of which most closely matches the model of a solid body, and forming from these trajectories an average trajectory of the vehicle, reducing the random positioning error;

using a statistical analysis of data on the position of the vehicle in the projection onto the plane of the roadway at the time of fixing the image frames to estimate the resulting vehicle speed, while compensating for random errors by multiple positioning on successive frames of the video stream.

In a preferred embodiment of the system, the electronic computing device is configured to determine the model of the vehicle, while determining on the number plate the state number of the vehicle and sending a request to it in the database of vehicles registered in the traffic police connected to the electronic computing device, and in a response to a request for obtaining data on a vehicle model from a database of vehicles registered in the traffic police.

In a preferred embodiment of the system, the electronic computing device is configured to determine the vehicle model, while sending a request to the vehicle database, which is in the memory of the electronic computing device and containing data about the vehicle models and the corresponding coordinates of the characteristic elements of the vehicles in neighborhood of the license plate, and in response to a request for obtaining vehicle model data from the database.

For a better understanding of the claimed invention the following is a detailed description with the corresponding graphic materials.

FIG. 1. A block diagram of a vehicle speed detection system according to the invention.

FIG. 2. Scheme of the projection of the coordinates (x, y) of the roadway on the pixel coordinates (u, ν) of the image formed by the video camera, made according to the invention.

FIG. 3. An example of the allocation of characteristic points in the image of the vehicle, made according to the invention.

в ортогональную

выполненная согласно изобретению.FIG. 4. Illustration of recalculation of the central projection

into orthogonal

made according to the invention.

Items:

1 - video camera;

2 - electronic computing device.

Consider in more detail the functioning of the claimed system and method for determining the speed of a vehicle (Fig. 1-4).

The claimed system consists of interconnected video cameras 1 and electronic computing device 2.

At the calibration stage of the claimed system, a correspondence is established between the coordinates of the image points (u, ν) in pixels and the metric coordinates (x, y) of the points of the roadway, and the height of the suspension h of the camera 1 above the plane of the roadway is measured and the projection of the position of the camera 1 on the road is determined (x ₀ , y ₀ ). In the approximation of geometric optics within the framework of the model of a thin lens (the model of a thin lens assumes the optical system of a video camera 1 by the optical properties of an equal thin lens), the coordinate transformation (u, ν) → (x, y) is performed using an electronic computing device 2 in accordance with the formulas

where A, B, C, D, E, F, G, H are the coefficients determined by the focal length of the video camera 1, the height of the video camera 1 and the projection of the installation point of the video camera 1 on the roadway, as well as the angle of rotation of the video camera 1 relative to its optical axis, relative to the axis of the suspension and the angle of inclination of the camcorder 1 with respect to the roadway.

In a more general form, the coordinate transformation is determined using functions

x = f (u, ν),

y = g (u, ν),

where the functions f, g can be determined at the calibration stage when putting the declared system into operation, for example, in a tabular way: the image space (u, ν) is covered with a grid, and at each grid point, the corresponding coordinates in the plane of the road surface are measured by geodetic methods. The density of the grid and, therefore, the difficulty in determining the calibration functions depends on the model of the lens of the camera 1 and on the degree of optical distortion of the images obtained with the camera 1. For cameras with lenses matching the model of a thin lens, four points in the grid that are not lying on one are enough line in order to restore the values of the functions f, g at other points.

The initial data in the form (x ₀ , y ₀ ), h, x = f (u, ν), y = g (u, ν) are used to estimate the speed by performing the following operations using electronic computing device 2.

, которые соответствуют характерным элементам транспортного средства, где i - индекс характерной точки транспортного средства. В первую очередь к таким точкам относятся элементы автоматически локализуемой и идентифицируемой номерной пластины: координаты отдельных символов, угловые точки рамки номерной пластины. Также к характерным точкам относятся точки контуров элементов конструкции транспортного средства, например фары, решетки радиаторов, лобовые стекла, контуры капотов и т.д., в которых линии контуров пересекаются под значительным углом, что позволяет точнее определять положение этих точек в любых ракурсах. Сам алгоритм выделения характерных точек строится на принципах машинного обучения: параметры критериев отбора, является ли рассматриваемая характерная точка изображения удачной, формируют при подготовке программного обеспечения на основе статистического анализа и оптимизации.In the first step, on each image j formed by the video camera 1, in which the vehicle number is fixed, a plurality of points $$\left\{\left(u_i^j,{\nu }_i^j\right)\right\}$$

that correspond to the characteristic elements of the vehicle, where i is the index of the characteristic point of the vehicle. First of all, such points include elements of an automatically localized and identifiable number plate: coordinates of individual characters, corner points of the number plate frame. Characteristic points also include the contour points of vehicle structural elements, such as headlights, radiator grills, windshields, hood contours, etc., in which the contour lines intersect at a significant angle, which allows you to more accurately determine the position of these points in any angle. The algorithm for distinguishing characteristic points is based on the principles of machine learning: the parameters of the selection criteria, whether the considered characteristic point of the image is successful, are formed when preparing the software based on statistical analysis and optimization.

транспортного средства приписывают высоту над дорожным полотном, а для некоторых симметричных элементов кузова транспортного средства, соответствующих парам характерных точек

в силу симметрии расположенных на одной высоте, приписывают расстояние между точками $$d_{i_1{i}_2}$$

. Тип информации, приписываемый характерным точкам и парам характерных точек, выбирают по критерию оптимальности решения конечной задачи по оценке скорости с минимальной погрешностью. Это означает, что критерии выбора определяют заранее на этапе машинного обучения перед вводом заявленной системы в эксплуатацию.In the second step, on the basis of the image of the vehicle formed by the video camera 1, a model of the vehicle is determined in the vicinity of the characteristic points of the vehicle. If the claimed system and method have access to a database of vehicles registered in the traffic police, then the type of vehicle is determined by querying this database. If the claimed system and method do not have access to the database of vehicles registered in the traffic police, then the type of vehicle is determined by querying the database of vehicles located in the memory of electronic computing device 2 and containing data on vehicle models and their corresponding coordinates characteristic elements of vehicles in the vicinity of the license plate, and in response to a request, mode data is obtained from the database in the memory of the electronic computing device 2 and the vehicle. Automatically assess the posterior probability of the correct determination of the vehicle model. If it is high enough, then based on the design features of the vehicle and reference data to some of the characteristic points

the vehicle is attributed to the height above the roadway, and for some symmetrical elements of the vehicle body corresponding to pairs of characteristic points

due to symmetry located at the same height, attribute the distance between points $$d_{i_{\mathrm{one}}{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="00000030.png"\; state="\{\{state\}\}">i</figure-callout>}}_2}$$

. The type of information attributed to characteristic points and pairs of characteristic points is selected according to the optimality criterion for solving the final problem of estimating speed with a minimum error. This means that the selection criteria are determined in advance at the stage of machine learning before putting the claimed system into operation.

с помощью формул x=f(u,ν), y=g(u,ν) определяют центральные проекции

этих характерных точек на плоскость дороги относительно точки крепления видеокамеры 1.In the third step for each of the characteristic points

using the formulas x = f (u, ν), y = g (u, ν) determine the central projections

these characteristic points on the road plane relative to the mounting point of the camcorder 1.

с известной высотой h_i ортогональную проекцию на плоскость дорожного полотна определяют по формуламIn the fourth step for characteristic points

with a known height h _{i, the} orthogonal projection onto the road surface is determined by the formulas

which follow from the similarity of the triangles shown in FIG. four.

между которыми известно расстояние $$d_{i_1{i}_2}$$

в мировом пространстве и которые в силу симметрии расположены на одной высоте

над плоскостью дорожного полотна, соответствующие ортогональные проекции на плоскость дороги

определяют на основании системы уравнений, составленной из соотношений, выведенных в на предыдущем шаге:In the fifth step for symmetric pairs of characteristic points

between which the distance is known $$d_{i_{\mathrm{one}}{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="00000030.png"\; state="\{\{state\}\}">i</figure-callout>}}_2}$$

in world space and which, by virtue of symmetry, are located at the same height

above the road surface, corresponding orthogonal projections onto the road plane

determined on the basis of a system of equations composed of the relations derived in the previous step:

в результате решения данной системы получают следующий результат:Assuming that the attachment point of the camcorder 1 is located above the vehicle, i.e. $$h>\tilde{h},$$

as a result of solving this system, the following result is obtained:

и вычисления соответствующих им ортогональных проекций

сопоставляют одинаковые характерные точки на последовательных кадрах видеоизображения, при этом выделяют набор траекторий {T_k} движения характерных точек транспортного средства, где T_k на каждом кадре j соответствует проекция одной из характерных точек на плоскость дорожного полотна. Проводят статистический анализ данных для траекторий {T_k}, при этом удаляют из данного множества траекторий те траектории, которые не соответствуют модели движения твердого тела. В результате выделяют множество траекторий {T_l}, являющихся подмножеством {T_k}, движение которых согласовано в рамках модели движения твердого тела.In the sixth step, after highlighting the characteristic points of the image

and calculating their corresponding orthogonal projections

match the same characteristic points on successive frames of the video image, while a set of trajectories {T _k } of movement of the characteristic points of the vehicle, where T _k on each frame j corresponds to the projection of one of the characteristic points on the plane of the roadway. A statistical analysis of the data for the trajectories {T _k } is carried out, while those trajectories that do not correspond to the model of motion of a rigid body are removed from this set of trajectories. As a result, many trajectories {T _l } are distinguished, which are a subset of {T _k }, the motion of which is consistent within the framework of the model of motion of a rigid body.

раз, где

- количество отобранных на предыдущем шаге траекторий. В среднем для транспортных средств выделяют более четырех характерных точек, которые можно с высокой степенью надежности отследить на множестве кадров видеоизображения, поэтому данный шаг позволяет более чем в 2 раза уменьшить погрешность конечной оценки скорости.At the seventh step, from all the trajectories {T _l } of the movement of the characteristic points of the vehicle, one trajectory T of the vehicle is formed on the road surface, which in each frame j is composed of points corresponding to the geometric center of the points of the trajectories {T _l } on this frame. Averaging can reduce the random positioning error of the vehicle on the roadway in

times where

- the number of trajectories selected in the previous step. On average, more than four characteristic points are distinguished for vehicles that can be tracked with a high degree of reliability on a plurality of frames of a video image; therefore, this step allows reducing the error of the final speed estimate by more than 2 times.

раз по сравнению с разностной формулой на основе лишь первого и последнего отсчетов, где N - количество точек в траектории T. Соответственно, в

раз уменьшится случайная погрешность оценки модуля скорости, вычисляемой по формулеAt the eighth step for geometric centers _{{(-X j, Y j)} }, corresponding to the trajectory T, performed statistical analysis step, the estimate vehicle speed ν based on the data (X _j, Y _j) and time stamps frames t _j. One way to do this analysis is to find the linear regression coefficients (ν _x , ν _y ) for the corresponding sequences of values {{X _j , t _j )} and {(Y _j , t _j )} using the least squares method, which allows you to reduce the random error of the estimate projection of speed (ν _x , ν _y ) onto the road surface in

times compared with the difference formula based on only the first and last samples, where N is the number of points in the trajectory T. Accordingly, in

times, the random error in the estimation of the velocity modulus calculated by the formula

In a typical installation of the claimed system, the vehicle is in the field of view N = 10 frames of the video image even at high vehicle speeds, i.e., the error in determining the speed is additionally reduced by 3 times.

The accuracy of estimating the height of the characteristic points of the vehicle over the roadway, determined on the basis of the recognition of the brand of the vehicle, can be twice as high as the methods for estimating the height used in the analogue and prototype described in patents RU 2419884 and GB 2503328, respectively. additional gain obtained by calculating the center of mass of the identified characteristic points (from 2 times) and using statistical processing, for example, by the least squares method (from 3 times) the final gain in accuracy and compared with the analogue and prototype described in patents RU 2419884 and GB 2503328, respectively, exceeds 10, i.e. the relative error in the estimation of speed using the inventive method is less than 1%, if it is taken for granted that the relative error in the estimation of speed by the prototype method described in patent GB 2503328 reaches 8%.

Although the above-described embodiment of the invention has been set forth to illustrate the present invention, it is clear to those skilled in the art that various modifications, additions and substitutions are possible without departing from the scope and meaning of the present invention disclosed in the attached claims.

Claims (7)

1. The method of determining the speed of a vehicle, in which using a video camera to form an image of the roadway and vehicles located on it, and using an electronic computing device:
- at the preliminary calibration stage, a correspondence is established between the pixel coordinates of the image points formed by the video camera and the metric coordinates of the roadway points, while for several image points and the corresponding roadway points, data on the measured distance from the image point to the corresponding road point is obtained and used canvases;
- at the preliminary calibration stage, the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway are also calculated;
- determine the fixation time for each frame of the image formed by the camera;
using the correspondence between the coordinates of the points of the image formed by the video camera and the coordinates of the points of the roadway, and also using the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway using an electronic computing device:
on each frame of the video sequence containing the vehicle, the position of the elements of the state registration plate of the vehicle and the coordinates of the characteristic elements of the vehicle in the vicinity of the license plate are determined;
determine a vehicle model, which determines the coordinates of the characteristic elements of the vehicle of this model relative to each other and the height of the characteristic elements of the vehicle of this model above the plane of the roadway;
statistically determine the characteristic points of the vehicle with which you can most accurately determine the position of the vehicle on the plane of the roadway;
recalculating the coordinates of the image of the characteristic points of the vehicle in the coordinates of the orthogonal projection of the characteristic points on the plane of the roadway, using information about the height of the characteristic points above the roadway;
statistically distinguish the trajectories of characteristic points, the movement of which most closely matches the model of a solid body, and form from these trajectories an average trajectory of the vehicle, reducing the random positioning error;
using statistical analysis of data on the position of the vehicle in the projection onto the road surface at the time of fixing image frames, the resulting vehicle speed is estimated, compensating for the random error by multiple positioning on successive frames of the video stream. 2. The method according to p. 1, characterized in that at the calibration stage, a correspondence is established between the pixel coordinates of the image points formed by the video camera and the metric coordinates of the roadway points, while for several image points and their corresponding roadway points using a laser range finder the distance from the image point to its corresponding roadway point; at the calibration stage, using a laser range finder, measure the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway. 3. The method according to p. 1, characterized in that the model of the vehicle is determined, the state number of the vehicle is determined on the number plate and a request is sent to it in the database of vehicles registered in the traffic police, connected to the electronic computing device, and in the response to the request is received from the database data about the vehicle model. 4. The method according to p. 1, characterized in that they determine the model of the vehicle, while sending a request to the database of vehicles in the memory of the electronic computing device and containing data about the models of vehicles and the corresponding coordinates of the characteristic elements of vehicles in neighborhood of the license plate, and in response to a request, data on the model of the vehicle are obtained from the database. 5. A system for determining the speed of a vehicle, comprising a interconnected video camera and an electronic computing device, the video camera configured to form an image of the roadway and vehicles on it; and the electronic computing device is configured to:
- at the preliminary stage of calibrating the installation of correspondence between the pixel coordinates of the image points formed by the video camera and the metric coordinates of the roadway points, while for several image points and the corresponding roadway points, receiving and using data about the measured distance from the image point to the corresponding road point canvases;
- at the preliminary stage of calibration, also calculating the height of the camera’s suspension above the plane of the roadway and the coordinates of the projection point of the video camera’s position on the roadway;
determining a fixation time for each frame of the image formed by the video camera;
in addition, the electronic computing device is configured to:
on each frame of the video sequence containing the vehicle, determining the position of the elements of the state registration plate of the vehicle and the coordinates of the characteristic elements of the vehicle in the vicinity of the license plate;
determining a vehicle model, and from it determining the coordinates of the characteristic elements of the vehicle of this model relative to each other and the height of the characteristic elements of the vehicle of this model above the plane of the roadway;
statistically determine the characteristic points of the vehicle with which you can most accurately determine the position of the vehicle on the plane of the roadway;
recalculating the coordinates of the image of the characteristic points of the vehicle in the coordinates of the orthogonal projection of the characteristic points on the plane of the roadway, using information about the height of the characteristic points above the roadway;
statistically identifying trajectories of characteristic points, the movement of which most closely matches the model of a solid body, and forming from these trajectories an average trajectory of the vehicle, reducing the random positioning error;
using a statistical analysis of data on the position of the vehicle in the projection onto the plane of the roadway at the time of fixing the image frames to estimate the resulting vehicle speed, while compensating for random errors by multiple positioning on successive frames of the video stream. 6. The system according to p. 5, characterized in that the electronic computing device is configured to determine the model of the vehicle, while determining on the number plate the state number of the vehicle and sending a request to it in the database of vehicles registered in the traffic police, connected to an electronic computing device, and in response to a request for obtaining data about a vehicle model from a database of vehicles registered in the traffic police. 7. The system according to p. 5, characterized in that the electronic computing device is configured to determine the model of the vehicle, while sending a request to the database of vehicles located in the memory of the electronic computing device and containing data about the models of vehicles and the corresponding the coordinates of the characteristic elements of vehicles in the vicinity of the license plate, and in response to a request for obtaining vehicle model data from the database.

Images