KR20200104551A - Apparatus and method for detecting moving objects - Google Patents

Abstract

The present invention relates to a moving object detection device. The moving object detection device comprises: a generation unit generating silhouette information of a moving object by receiving a reflected signal changed by the moving object as the moving object passes through an observation point on the ground from a laser sensor; and a detection unit detecting information on the moving object through analysis using the silhouette information. The information on the moving object may include type information and moving speed information of the moving object.

Description

Moving object detection device and method {APPARATUS AND METHOD FOR DETECTING MOVING OBJECTS}

The present application relates to a moving object detection apparatus and method. In addition, the present application relates to a moving object detection system.

A vehicle detection system that recognizes vehicles is used to obtain vehicle information from a toll gate on a highway that differentially charges tolls according to the type of vehicle or to determine the amount of traffic running on the road (road traffic conditions). . Such a vehicle detection system typically scans a laser through a passing vehicle and analyzes a reflected wave to obtain information on the passing vehicle.

Meanwhile, as a method of estimating the moving speed of a moving object in the related art, for example, an observer who wants to estimate the moving speed of the moving object is located in a direction parallel to the moving direction of the moving object (ie, a parallel horizontal direction). , There is a method of measuring the speed of a moving object using a change in a reflected signal by projecting a laser in a parallel direction with respect to a moving object. That is, in this conventional method, an observer is located on the axis of the movement direction of the object, and a kind of energy is projected in parallel (frequency or wavelength of a wave) to the moving object. As a method of measuring the speed of a moving object using the Doppler shift or The Doppler effect, for example, the case of measuring the speed of a car approaching and moving away from an observer.

As another example, as a method of estimating the moving speed of a moving object in the related art, an observer is positioned in a direction perpendicular to the side with respect to the moving direction axis of the object, and the change in energy transmitted or reflected back (e.g. For example, there is a method of estimation using opto-Interrupter). This conventional method is a method of estimating by using a change in energy that is transmitted through the energy source in a direction orthogonal to the direction of movement and the side, and reflected and returned. It is possible to observe a passing car through a small window next to the road. For example.

However, most of the vehicle detection techniques or techniques for estimating the speed of a moving object known in the related art have a problem in that accuracy is poor in classifying (identifying) the type of a moving object (eg, a vehicle) or estimating the speed.

The technology behind the present application is disclosed in Korean Patent Publication No. 10-0403822.

The present application is to solve the problems of the prior art described above, and an object thereof is to provide a moving object detection apparatus and method capable of accurately detecting the type, speed, etc. of a moving object (eg, vehicle).

However, the technical problem to be achieved by the embodiments of the present application is not limited to the technical problems as described above, and other technical problems may exist.

As a technical means for achieving the above technical problem, the moving object detection device according to the first aspect of the present application receives the changed reflected signal by the moving object as the moving object passes through the observation point on the ground and moves it from the laser sensor. A generator that generates silhouette information of an object; And a detector configured to detect information on the moving object through analysis using the silhouette information, and the information on the moving object may include type information and movement speed information of the moving object.

In addition, the laser sensor may irradiate an energy source in a direction orthogonal to a moving direction of a moving object passing through the observation point and receive a reflected signal corresponding to the irradiated energy source.

In addition, the changed reflected signal may be a signal received by reflecting an energy source irradiated in a vertical direction toward an upper surface of the moving object passing through the observation point by the laser sensor.

In addition, the silhouette information is information expressed by a relationship between the reflected signal measured over time, and the reflected signal having a preset threshold value among the reflected signals transmitted from the laser sensor is first measured from the first point in time. The reflected signal having a set threshold value may be information generated by using a reflected signal belonging to the second measured second time point as the changed reflected signal.

In addition, the detection unit, based on the pre-learning item based on a deep learning model in which a plurality of learning silhouette information is input as an input value and type information of a moving object matching each of the plurality of learning silhouette information is an output value, the deep learning model By applying the silhouette information as input, information on the type of a moving object corresponding to the silhouette information can be detected.

In addition, the moving object detection apparatus according to the first aspect of the present application further includes a database for matching and storing length information for each type of the moving object, and the detection unit includes the type information of the moving object detected using the deep learning model. To identify the length information of the moving object from the database, and detect the moving speed information of the moving object corresponding to the silhouette information by using the length information of the identified moving object and time information corresponding to the silhouette information. I can.

In addition, the detection unit may detect whether or not a relative moving speed is faster than a reference moving speed of the moving object as information about the moving object, using the detected type information of the moving object and the silhouette information.

On the other hand, the moving object detection system according to the second aspect of the present application irradiates an energy source in a direction orthogonal to a moving direction of a moving object passing through an observation point on the ground and receives a reflected signal corresponding to the irradiated energy source. Laser sensor; And generating silhouette information of the moving object by receiving the reflected signal changed by the moving object as the moving object passes through the observation point from the laser sensor, and analyzing information on the moving object through analysis using the silhouette information. A moving object detection device that detects, and the information on the moving object may include type information and moving speed information of the moving object.

In addition, a moving object detection method according to a third aspect of the present application includes: generating silhouette information of the moving object by receiving a reflected signal changed by the moving object as the moving object passes through an observation point on the ground from a laser sensor; And detecting information on the moving object through analysis using the silhouette information, and the information on the moving object may include type information and moving speed information of the moving object.

In addition, the reflected signal changed in the generating step may be a signal received by reflecting an energy source irradiated in a vertical direction toward an upper surface of the moving object passing through the observation point by the laser sensor.

In addition, the detecting may include, based on a pre-learning item based on a deep learning model in which a plurality of learning silhouette information is input as an input value and type information of a moving object matched with each of the plurality of learning silhouette information is used as an output value. By applying the silhouette information to the running model as input, it is possible to detect the type information of the moving object corresponding to the silhouette information.

As a technical means for achieving the above technical problem, the computer program according to the fourth aspect of the present application may be stored in a recording medium in order to execute the method for detecting a moving object according to the third aspect of the present application.

The above-described problem solving means are merely exemplary and should not be construed as limiting the present application. In addition to the above-described exemplary embodiments, additional embodiments may exist in the drawings and detailed description of the invention.

According to the above-described problem solving means of the present application, it is possible to detect the type and the moving speed of the moving object by using the silhouette information generated using the changed reflected signal received from a single laser sensor.

We can detect information on moving objects while using existing structures (facilities) such as traffic lights or information boards without damaging the road at all and without the need to install additional structures (facilities) for measurement. .

The present application can detect not only the type of the moving object, but also the moving speed of the moving object or traffic conditions on a specific road in real time.

However, the effect obtainable in the present application is not limited to the effects as described above, and other effects may exist.

1 is a diagram showing a schematic configuration of a moving object detection system according to an embodiment of the present application.
2 is a schematic block diagram of an apparatus for detecting a moving object according to an embodiment of the present disclosure.
3 and 4 are views for explaining a moving object detection apparatus according to an embodiment of the present application.
5 is a diagram illustrating an example of a structure in which a laser sensor included in a moving object detection system according to an exemplary embodiment of the present disclosure can be provided.
6 is a diagram illustrating a process of detecting type information of a moving object by a detection unit of a moving object detecting apparatus according to an exemplary embodiment of the present disclosure.
7 is a view for explaining an example of detecting whether a relative moving speed is fast by a moving object detecting apparatus according to an exemplary embodiment of the present disclosure.
8 is a flowchart illustrating a method for detecting a moving object according to an exemplary embodiment of the present disclosure.

Hereinafter, exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present application. However, the present application may be implemented in various different forms and is not limited to the embodiments described herein. In addition, in the drawings, parts not related to the description are omitted in order to clearly describe the present application, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the present specification, when a part is said to be "connected" with another part, it is not only "directly connected", but also "electrically connected" or "indirectly connected" with another element interposed therebetween. "Including the case.

Throughout this specification, when a member is positioned "on", "upper", "upper", "under", "lower", and "lower" of another member, this means that a member is located on another member. It includes not only the case where they are in contact but also the case where another member exists between the two members.

Throughout the specification of the present application, when a certain part "includes" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

1 is a view showing a schematic configuration of a moving object detection system 100 according to an embodiment of the present application. 2 is a schematic block diagram of a moving object detection apparatus 10 according to an embodiment of the present disclosure. 3 and 4 are views for explaining a moving object detection apparatus 10 according to an embodiment of the present application.

Hereinafter, the moving object detection system 100 according to an embodiment of the present application will be referred to as the present system 100 for convenience of description, and the moving object detection device 10 according to an exemplary embodiment of the present application will be referred to as the present device 10. To

1 to 4, the system 100 may include a laser sensor 2 and a device 10 and a moving object detection device 10.

The laser sensor 2 irradiates an energy source in a direction (orthogonal, vertical) that is orthogonal to the moving direction of the moving object 1 passing through the observation point p on the ground, and reflects the energy source corresponding to the irradiated energy source. Can receive signals.

The ground is a ground (terrain) on which the moving object 1 is movable, and may mean, for example, a road. In addition, the moving object 1 is a detection target detected by the system 100 to the device 10, and may mean a vehicle (car) as an example.

Here, the energy source may be an energy source capable of measuring a distance (ie, an energy source for measuring a distance). For example, the energy source may be light (light source), ultrasonic wave, electromagnetic wave, laser, radar, etc., but is not limited thereto, and various energy sources enabling distance measurement may be applied.

Accordingly, the laser sensor 2 may be expressed differently as a distance measurement sensor, a distance sensor, or the like. In the present application, the reflected signal (reflected signal corresponding to the energy source irradiated by the laser sensor) received by the laser sensor 2 may mean a distance measurement value.

The laser sensor 2 is provided in a structure located above the moving object 1 so that the energy source can be irradiated in a vertical direction toward the upper side of the moving object 1 passing through the observation point p. , Can be placed).

For example, the laser sensor 2 may be provided in a detachable form on the surface of the structure. As another example, the laser sensor 2 may be provided in a form in which at least a portion is embedded in the structure.

5 is a view showing an example of a structure that can be provided with the laser sensor 2 included in the moving object detection system 100 according to an embodiment of the present application.

Referring to FIG. 5, in the system 100, the laser sensor 2 is an exemplary structure. As shown in FIGS. 5A and 5B, a speeding camera or a traffic guide sign (speed Restriction signs, road guide signs, etc.) are installed in facilities such as support (support members), tunnels (especially, the upper surface of the tunnel) as shown in Fig. 5(c), and toll gates of highways as shown in Fig. 5(d). Can be.

This is only an example to aid the understanding of the present application, and is not limited thereto, and as a structure to which the laser sensor 2 can be applied (equipped), the energy source in the vertical direction toward the upper side of the moving object 1 Various types of structures (facilities) that are located higher than the moving object 1 to enable irradiation can be applied.

The laser sensor 2 provided in such a structure may irradiate an energy source in a direction orthogonal (vertical) to the upper side (or ground) of the moving object 1 and receive a reflected signal corresponding thereto.

The reflected signal received by the laser sensor 2 may be transmitted (transmitted, provided) to the device 10 through a network.

The network is, for example, 3GPP (3rd Generation Partnership Project) network, LTE (Long Term Evolution) network, WIMAX (World Interoperability for Microwave Access) network, Internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network). ), WAN (Wide Area Network), PAN (Personal Area Network), Bluetooth (Bluetooth) network, NFC (Near Field Communication) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc. , It is not limited thereto, and all types of wired/wireless networks may be included.

The laser sensor 2 applied to the system 100 may be a single sensor, that is, one sensor. In other words, one laser sensor 2 may be applied to the system 100.

The device 10 receives the reflected signal changed by the moving object 1 as the moving object 1 passes through the observation point p from the laser sensor 2, and receives the silhouette information d of the moving object 1 ), and the information on the moving object 1 may be detected through analysis using the generated silhouette information d.

The device 10 may refer to a user terminal or a server. Here, the user terminal refers to PCS (Personal Communication System), GSM (Global System for Mobile Communication), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication) -2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (WCode Division Multiple Access), Wibro (Wireless Broadband Internet) terminal, Smartphone, SmartPad, tablet PC, notebook, wearable device , Desktop PC, etc. may include all types of wired and wireless communication devices.

A more detailed description of the device 10 is as follows.

The device 10 may include a generation unit 11, a detection unit 12, and a database 13.

The generating unit 11 receives the changed reflection signal by the moving object 1 as the moving object 1 passes through the observation point p on the ground from the laser sensor 2 and receives the silhouette information of the moving object 1 (d) can be created. The silhouette information d may be expressed differently as outline information, shape information, or the like.

Here, the changed reflected signal may be a signal received by reflecting an energy source irradiated in a vertical direction toward the upper surface of the moving object 1 passing through the observation point p by the laser sensor 2.

In other words, when the laser sensor 2 irradiates the energy source vertically toward the upper surface of the moving object 1 passing through the observation point p, the changed reflected signal corresponds to the irradiated energy source. It may be a reflected signal (reflected signal reflected from the upper surface of the moving object) that is reflected from the upper surface of) and received by the laser sensor 2.

When the moving object 1 does not pass through the observation point p, the laser sensor 2 measures the distance from the position where the laser sensor 2 is provided to the ground where the observation point p is located as a reflected signal. Can receive (acquire). Here, the distance measurement value received by the laser sensor 2 when the moving object 1 does not pass through the observation point p may be expressed as a first distance measurement value for convenience of description below.

On the other hand, when the moving object passes through the observation point (p) (when passing), the laser sensor (2) is compared to the case where the moving object (1) does not pass through the observation point (p). It is possible to receive the reflected signal changed by 1) (ie, due to the height of the moving object). The laser sensor 2 can transmit the received changed reflected signal to the generation unit 11 of the device 10 through a network, and the generation unit 11 uses the changed reflected signal received from the laser sensor 2 Thus, it is possible to generate the silhouette information (d) of the moving object.

That is, when the moving object passes through the observation point (p), when the moving object (1) does not pass through the observation point (p), the energy source irradiated from the laser sensor (2) is compared to the reflection from the ground. Thus, the energy source irradiated from the laser sensor 2 can be reflected from the upper surface of the moving object 1.

In other words, let's say that the laser sensor 2 is positioned perpendicular to the moving direction (the traveling direction) of the moving object 1 and the distance from the position of the laser sensor 2 to the observation point p on the ground is measured. In this case, if there is no moving object passing through the observation point p, the laser sensor 2 may always measure the first distance measurement value as a constant distance measurement value. On the other hand, when the moving object 1 passes through the observation point p with respect to the lower side where the laser sensor 2 is located, as shown in FIG. 3, from the position of the laser sensor 2 The distance to the ground may be changed by the shape (silhouette, shape) of the moving object 1.

Therefore, when the moving object 1 passes through the observation point p, the laser sensor 2 is a changed reflected signal that changes along the shape line of the upper side of the moving object 1 as shown in FIG. 3. Can receive. At this time, the laser sensor 2 is a reflected signal that is changed when the moving object 1 passes through the point of interest p, and has a second distance measurement value smaller than the first distance measurement value (changed reflected signal) Can receive.

In other words, when the moving object 1 does not pass through the observation point p, the laser sensor 2 may receive a reflected signal having the first distance measurement value. On the other hand, when the moving object 1 passes through the observation point p, the laser sensor 2 obtains a second distance measurement value smaller than the first distance measurement value due to reflection of the energy source on the moving object 1 It is possible to receive the changed reflected signal.

A description of the silhouette information d can be more easily understood with reference to FIG. 4.

Referring to FIG. 4, silhouette information d generated by the generator 11 may be information expressed by a relationship between a reflected signal measured over time. That is, it may be information indicating a reflected signal measured over time. In other words, the silhouette top (d) may be information expressed as a relationship between a time when the reflected signal is measured and a value of the received reflected signal (ie, a magnitude value of the reflected signal, meaning a distance measurement value). This silhouette information d may appear in a waveform form.

The silhouette information (d) generated by the generation unit 11 is information on the shape (silhouette) of the moving object 1, and different shapes according to the speed of the moving object 1 and the type of the moving object 1 Can appear as

In addition, the silhouette information (d) is a preset threshold value from the first time point t ₀ at which the reflected signal having a preset threshold value (Th) among the reflected signals transmitted from the laser sensor 2 is first measured. The reflected signal having Th) may be information generated by using the reflected signal belonging to the second measured second time point t ₁ as the changed reflected signal.

The preset threshold Th is used to identify the reflected signal reflected from the moving object 1 among all of the reflected signals transmitted from the laser sensor 2 (that is, the reflected signal changed by the moving object). It can mean a standard value.

In addition, the first point of view (t _o ) is a point in time at which the head of the moving object 1 (the front end) passes through the observation point p, in other words, the head of the moving object 1 is with the observation point p. It may mean a point in time that is on the same line. The second point of view (t ₁ ) is a point in time when the trailing end of the moving object 1 (the rear end) passes through the observation point p, in other words, the trailing edge of the moving object 1 is on the same line as the observation point p. It can mean the time point to achieve.

The generation unit 11 identifies a section consisting of reflected signals having a threshold value (Th) or higher among all of the reflected signals transmitted from the laser sensor 2, and utilizes the reflected signal belonging to the identified section as a changed reflected signal to form a silhouette. Information (d) can be generated. This silhouette information d may be referred to as information indicating the shape of the moving object 1 that can be used to classify (classify) the type (type) of the moving object.

In addition, time information corresponding to a section consisting of reflected signals having a threshold value (Th) or more may mean time information corresponding to the silhouette information (d). The time information corresponding to this silhouette information d is time information obtained by subtracting the time at the first time point t ₀ from the time at the second time point t ₁ , and the amount of time change information (t ₁ -t ₀ , Δt) Can mean

The detection unit 12 may detect information on the moving object 1 through analysis using silhouette information generated by the generation unit 11. Here, the information on the moving object 1 may include information on the type (type) of the moving object 1 and information on the moving speed of the moving object 1. The type information of the moving object 1 may mean vehicle type information when the moving object 1 is a vehicle. A description of the detection unit 12 may be more easily understood with reference to FIG. 6.

6 is a view for explaining a process of detecting type information of a moving object by the detection unit 12 of the moving object detecting apparatus 10 according to an exemplary embodiment of the present disclosure.

Referring to FIG. 6, the detection unit 12 receives a plurality of learning silhouette information as an input value and a deep learning model (3)-based pre-learning item in which type information of a moving object matching each of the plurality of training silhouette information is used as an output value. On the basis of, by applying the silhouette information (d) generated by the generator 11 to the deep learning model (3) as an input, the type information of the moving object (1) corresponding to the silhouette information (d) is transferred to the moving object (1). ), which can be detected (estimated).

The detection unit 12 may classify the type of the moving object 1 corresponding to the silhouette information d from the silhouette information d using the deep learning model 3.

Specifically, the deep learning model 3 considered in the present apparatus 10 may be a neural network model trained to output type information of a moving object corresponding to the given silhouette information. To this end, the deep learning model 3 takes a plurality of silhouette information for a plurality of types of moving objects or a plurality of silhouette information for a single moving object as learning silhouette information (multiple learning silhouette information) as input values, and It may be learned to output type information of a moving object matching each of the plurality of learning silhouette information as an output value.

The detection unit 12 provides the silhouette information d generated by the generation unit 11 as an input of the deep learning model 3 based on the previously learned items of the deep learning model 3, and as a result, It is possible to acquire (output, detect) type information of the moving object 1 corresponding to the silhouette information d from the running model 3.

As an example, the detection unit 12 extracts a plurality of feature points from the silhouette information d generated by the generation unit 11 and inputs the extracted feature points into the deep learning model 3 to provide a deep learning model. The type information of the moving object 1 corresponding to the silhouette information d can be obtained from (3).

The deep learning model 3 may refer to an artificial intelligence (AI) algorithm model, a machine learning (machine learning) model, a neural network model (artificial neural network model), a neuro fuzzy model, and the like. In addition, the deep learning model (3) is, for example, a convolutional neural network (CNN), a recurrent neural network (RNN), a deep neural network, etc. Neural network model can be applied.

Exemplarily, the detection unit 12 is information on the type of the moving object 1, and according to the displacement and size, a light car, a small-size car, a mid-size car, a full-sized car, etc. Can be detected. In addition, the detection unit 12, as type information of the moving object 1, detects a passenger car, a van (ambulance, etc.), a freight car, a special vehicle (camping car, etc.), a truck, a bus, etc. according to the type of the moving object 1. I can.

Specifically, the detection unit 12 may detect Kia Morning, Ray, Chevrolet Spark, or the like as a compact vehicle. The detection unit 12 is a compact car and can detect Kia Stonic, Chevrolet Trax, Hyundai Kona, and the like. The detection unit 12 is a medium-sized vehicle and can detect the Kia K5, Renault Samsung SM5, Chevrolet Malibu, Hyundai Sonata, and the like. The detection unit 12 may detect a Kia Mojave, a Kia K9, a Kia Carnival as a large vehicle.

As such, the detection unit 12 may detect vehicle type information, vehicle model name information, and the like as type information of the moving object 1 using the deep learning model 3.

After the type information of the moving object 1 corresponding to the silhouette information d is detected using the deep learning model 3, the detection unit 12 provides the type information and the generation unit 11 of the detected moving object 1 Using the silhouette information d generated in ), it is possible to detect whether or not the relative moving speed of the moving object 1 is faster than the reference moving speed of the moving object 1 as information about the moving object. This can be more easily understood with reference to FIG. 7.

7 is a view for explaining an example of detecting whether or not the relative moving speed is fast by the moving object detecting apparatus 10 according to an exemplary embodiment of the present disclosure.

Referring to FIG. 7, as an example, assume that the moving object 1 is moving at a reference moving speed (first moving speed, v ₁ ) as shown in FIG. 7A. That is, suppose that the speed v of the moving object 1 to be moved (moving) is equal to the reference moving speed v ₁ .

When the moving object 1 passes through the observation point p at the reference moving speed v ₁ , the silhouette information generated based on the changed reflected signal received from the laser sensor 2 is shown in FIG. 7(b). It may be of the same shape (shape). In other words, silhouette information of the moving object ₁ moving at the reference moving speed v ₁ may be generated as shown in (b) of FIG. 7.

If the moving object (1) passes the observation point (p) at a second moving speed (v ₂ ) faster than the reference moving speed (first moving speed, v ₁ ), the second moving speed (v ₂ ) Silhouette information of the moving object 1 may be generated as shown in (c) of FIG. 7.

In other words, when the moving speed v of the moving object 1 is v ₂ >v ₁ , the length of the silhouette information of the moving object 1 moving at the v ₂ speed (that is, horizontally based on the drawing of FIG. 7) The length of the silhouette information in the direction) may appear relatively shorter than the length of the silhouette information of the moving object 1 moving at a velocity v ₁ . That is, when the moving speed v of the moving object 1 is v ₂ > v ₁ , the length of the silhouette information in FIG. 7(c) is relatively larger than the length of the silhouette information in FIG. 7(b). May appear briefly.

On the other hand, when the moving object 1 passes through the observation point (p) at a third moving speed (v ₃ ) slower than the reference moving speed (first moving speed, v ₁ ), it moves at the third moving speed (v3). The silhouette information of the moving object 1 may be generated as shown in (d) of FIG. 7.

In other words, when the moving speed v of the moving object 1 is v ₃ <v ₁ , the length of the silhouette information of the moving object 1 moving at the v ₃ speed (that is, horizontally based on the drawing of FIG. 7) The length of the silhouette information in the direction) may appear relatively longer than the length of the silhouette information of the moving object 1 moving at a velocity v ₁ . That is, when the moving speed v of the moving object 1 is v ₃ <v ₁ , the length of the silhouette information in FIG. 7(d) is relatively larger than the length of the silhouette information in FIG. 7(b). It can appear long.

According to this, in the state in which the type information of the moving object 1 corresponding to the silhouette information d is detected using the deep learning model 3, the detection unit 12 provides the detected type information and silhouette of the moving object 1 By using information (d) (in particular, length information of the silhouette information as time information corresponding to the silhouette information), the current moving speed v of the moving object 1 detected by the device 10 is the reference moving speed It is possible to detect (judgment) whether it is relatively fast or slow compared to (v ₁ ).

That is, the detection unit 12 uses the detected type information of the moving object 1 and silhouette information d, and the moving speed (current moving speed) of the moving object 1 passing through the observation point p is the reference. You can determine whether it is relatively fast or slow compared to the moving speed.

Here, the reference moving speed is, for example, a speed that serves as a reference for determining whether the moving object 1 passing through the observation point p is speeding, and may be preset by a user input or the like.

In other words, suppose that the reference moving speed is set to 60, and the silhouette information of the moving object 1 moving at a moving speed of 60 km/h is the same as in FIG. 7B. At this time, as an example, the silhouette information of the moving object 1 moving at a moving speed of 100 km/h is shown in Fig. 7(c), and the silhouette information of the moving object 1 moving at a moving speed of 40 km/h is shown in Fig. It may be the same as 7(d).

According to this, the detection unit 12 identifies and knows the type information of the moving object 1 corresponding to the silhouette information d generated by the generation unit 11 through the deep learning model 3, and the identified moving object When the silhouette information of the moving object at the preset reference movement speed is known, the detection unit 12 uses only the type information of the detected moving object 1 and the silhouette information d to be relative to the reference movement speed. Whether the moving speed is fast (or slow) can be detected. Even in a situation where the moving speed of the moving object 1 is not accurately known, the present device 10 only provides information on the type and silhouette information of the moving object 1 identified (detected) using the deep learning model 3. You can use it to determine whether you are speeding.

In order to determine whether the moving object 1 is overspeed using only the silhouette information d in a situation where the detection unit 12 does not accurately know the moving speed of the moving object 1, the moving object 1 is added to the database 13 to be described later. Silhouette information for each speed according to the type of) may be previously stored.

According to this, the detection unit 12 primarily determines the type information of the moving object 1 corresponding to the silhouette information d using the deep learning model 3, and then secondly determines the silhouette information d By comparing the silhouette information for each speed according to the type of the moving object 1 previously stored in the database 13, whether the moving speed of the moving object 1 passing through the observation point p is relatively fast, or whether it is overspeed Can be detected (judged).

As described above, the database 13 may pre-store silhouette information for each type of the moving object 1 or silhouette information for each speed according to the type of the moving object 1.

In addition, the database 13 may match and store length information (full length information) for each type of the moving object 1. The database 13 may match length information (full length information) of the moving object 1 corresponding to each type of the moving object 1 and store it in the form of a lookup table. In other words, matching information between the type of the moving object 1 and length information (full length information) of the moving object 1 may be previously stored in the database 13.

At this time, the length information of the moving object 1 stored in the database 13 is the total length information of the actual moving object 1, that is, the total length, which is the horizontal length from the front end to the rear end of the moving object 1 ( overall length) information.

The detection unit 12 may identify the length information of the moving object 1 from the database 13 by using the type information of the moving object 1 detected using the deep learning model 3. The detection unit 12 compares (compares) the type information of the moving object 1 detected using the deep learning model 3 with the lookup table previously stored in the database 13 to identify the length information of the moving object 1. I can.

Thereafter, the detection unit 12 uses length information of the identified moving object 1 and time information corresponding to the silhouette information d, as information about the moving object 1, and the movement corresponding to the silhouette information d Information on the moving speed of the object 1 can be detected (estimated, calculated).

Here, the time information corresponding to the silhouette information (d) corresponds to the time corresponding to the second viewpoint (t ₁ ) (reflection signal reception time at the second viewpoint) and the first viewpoint (t ₀ ), as described above. As time information corresponding to the difference between the time (reception time of the reflected signal at the first time point), it may be differently expressed as time change amount information (t ₁ -t ₀ , Δt).

According to this, the detection unit 12 uses the type information of the moving object 1 identified based on the deep learning model 3 and the time information corresponding to the silhouette information d analyzed based on the silhouette information d. , It is possible to accurately detect (calculate) information on the moving speed of the moving object 1 at the time when the moving object 1 passes through the observation point p.

A more detailed description of the detection process for the moving speed information of the moving object 1 is as follows.

For example, assume that the silhouette of the moving object 1 measured at the observation point p is as shown in FIG. 4.

At this time, the time at which the head of the moving object 1 passes through the observation point p (the first time point) is t _o, and the time when the tail of the moving object 1 passes through the observation point p (second time point) When the time of time 2) is t ₁ , the difference between the time of the first time point and the time of the second time point △t (i.e., time change information, t ₁ -t ₀ = △t) is the same as in Equation 1 below. Can be satisfied.

[Equation 1]

Here, Δt denotes a time difference between the time of the first viewpoint and the time of the second viewpoint, L denotes the length of the moving object (full length), and v denotes the moving speed of the moving object 1.

In the present application, since the observation is made at a single observation point (p) using a single laser sensor (2) in detecting the moving object (1), only the reflected coral-based silhouette information received from the single laser sensor (2) is used. As a result, the length (full length) of the moving object 1 can be calculated (estimated). That is, in this application, since the observation point of the laser sensor 2 is one point, the length of the moving object 1 cannot be known in advance.

Therefore, the detection unit 12 of the device 10 applies the generated silhouette information as an input of the deep learning model 3, thereby using the deep learning model 3 to determine the type of moving object corresponding to the silhouette information d. Can be detected (identified, estimated). Thereafter, the detection unit 12 compares the type information of the moving object 1 detected using the deep learning model 3 with a lookup table previously stored in the database 13 to determine the length (full length) of the moving object 1 Can be identified. Thereafter, the detection unit 12 substitutes the L value, which is the length information of the identified moving object 1, and the Δt value, which is the time change amount information that can be derived from the silhouette information d, into Equation 2, Accurate movement speed can be detected (estimated, calculated).

[Equation 2]

Here, v denotes the moving speed of the moving object 1 passing through the observation point p detected by the detection unit 12.

Exemplarily, when the type information of the moving object 1 detected using the silhouette information d is a medium-sized vehicle (eg, SM 5), the moving object ( The length information (full length information) of 1) can be identified as being 5m. At this time, if the time corresponding to the length of the section having the reflected signal equal to or greater than the preset threshold Th among the silhouette information d is 180 ms (i.e., the time change amount information Δt) is 180 ms, the detection unit 12 uses Equation 2 above. Based on this, the moving speed (passing speed) at which the moving object 1 passes through the observation point p can be detected (calculated) as 100 km/h as shown in Equation 3 below.

[Equation 3]

As such, the device 10 corresponds to a section having a preset threshold value (Th) or more among the silhouette information (d) generated based on the reflected signal obtained for the moving object (1) passing through the observation point (p). Time (Δt) is calculated, and silhouette information (d) is applied as an input of the deep learning model (3) to obtain the type information of the moving object (1), and the acquired type information of the moving object (1) is used. Based on the database 13, the length information of the moving object 1 is acquired, and the obtained length information and the calculated time (Δt) information are substituted into Equation 2, so that the moving object 1 ) Speed can be detected (calculated).

The present system 100 and the present apparatus 10 generate silhouette information d of the moving object 1 related to the total length of the moving object 1 using a single laser sensor 2, and the generated By applying the deep learning model 3 to the silhouette information d, it is possible to accurately detect (identify) the type of the moving object 1. In addition, the system 100 and the device 10 can detect whether the relative moving speed of the moving object 1 is high by using the detected type information of the moving object 1 and silhouette information d. have. In addition, the system 100 and the device 10 are based on the detected type information of the moving object 1 and the length information and silhouette information of the moving object 1 identified using the database 13 (d). ), it is possible to accurately detect (calculate) the moving speed of the moving object 1.

The system 100 and the apparatus 10 may detect the shape, type, and speed of a moving object (eg, a vehicle, etc.) moving at an arbitrary speed.

The laser sensor 2 considered herein is fixed with respect to the direction perpendicular to the moving direction of the moving object 1, and through irradiation of an energy source and reception of a reflected signal corresponding thereto, from the position of the laser sensor 2 The distance to the ground (for example, the surface of the road) and the distance to the upper surface of the moving object 1 as the moving object 1 passes through the observation point p can be measured. The apparatus 10 may generate a silhouette (shape, contour) of the moving object 1 by using the distance measured by the laser sensor 2 (ie, distance information based on reflection information). Thereafter, the device 10 can automatically classify (identify) the type of moving object based on the deep learning model 3 using the generated silhouette information of the moving object, and pass through a single observation point (p). The speed of the moving object 1 can be accurately detected (estimated).

In addition to being able to classify the type of moving object in real time, this application can accurately detect the speed of the moving object passing through a specific point (observation point), so that the moving object is designated for the specific point (p). It is possible to detect (judgment) whether the speed is exceeded or not.

Conventionally, as a method of estimating the moving speed of a moving object, for example, an observer who wants to estimate the moving speed is located in a direction parallel to the moving direction of the moving object (that is, in a horizontal direction parallel to the moving object). There is a method of measuring the speed of a moving object by projecting a laser in a parallel direction and using a change in a reflected signal. However, with such a conventional method, there is a problem in that it is difficult to distinguish (identify) the types of moving objects.

Accordingly, the present application irradiates (projects) an energy source (eg, light, ultrasonic waves, electromagnetic waves, etc.) vertically with respect to the moving direction of the moving object 1, and is based on the reflected signal changed according to the speed and shape of the moving object. Silhouette information is generated, and the type and speed according to the shape of the moving object 1 can be accurately detected using this.

Unlike the existing loop sensor method for detecting vehicle speeding and vehicle passing, the company does not damage the road at all and does not need to install additional structures (facilities) for measurement, It is possible to effectively detect information on a moving object while using the same existing structure (facility) as it is (ie, maintaining the state of the existing structure intact).

In addition, the present application not only detects the type of moving object (e.g., vehicle type), but also detects (measures) the moving speed of the moving object or the traffic condition on a specific road in real time. Can be utilized as a high technology.

In addition, when the laser sensor 2 is installed in a wide area such as an urban area or a highway with dense road network, the company acquires traffic volume-related big data information from information on a plurality of moving objects detected using the laser sensor 2 You can (collect). The traffic volume-related big data information collected by the present application can be used for city planning and the like.

In addition, according to the present application, it is possible to monitor traffic conditions in real time, and the present application can be easily utilized for destination search technology that links real-time traffic conditions to vehicle navigation.

In other words, the present application relates to a technology capable of determining the moving speed and type of a moving object. The present application uses a single laser sensor positioned in a direction perpendicular to the moving direction of a moving object (for example, a vehicle), unlike the conventional method using a sensor positioned in the horizontal direction with respect to the moving direction of the vehicle. You can determine the type and speed of movement.

For example, in the present application, the laser sensor may be provided in a place such as a traffic light to irradiate an energy source in a direction perpendicular to a moving direction of a moving object and receive a reflected signal corresponding to the irradiated energy source. In this case, silhouette information of the moving object 1 may be generated by using a reflected signal (a changed reflected signal) reflected from the moving object and received by the laser sensor 2. Since the silhouette information generated by the generating unit 11 is different according to the moving speed of the moving object (that is, the contour shape of the generated silhouette information is deformed), the present application uses this silhouette information to determine the relative moving speed of the moving object. Whether it is fast and the speed of a moving object can be determined.

In the present application, a deep learning model can be applied when classifying (identifying) the type of moving object, and it can be easily used in technologies such as traffic information, road information, and vehicle classification using big data.

Further, in the present application, the silhouette information d may be more accurately corrected by using environmental data (eg, temperature, humidity, fine dust concentration, etc.) around the laser sensor 2 as an example.

In one example of the present application, the moving object 1 is an object that moves on the ground (for example, a road, etc.) and is exemplified only as a vehicle (car), but is not limited thereto, and various movable objects may be considered. have. For example, the moving object 1 may be a motorcycle, train, subway (train), KTX, etc. as an object moving on the ground, and a ship, cargo ship, yacht, boat as an object moving on water (sea, etc.) Etc.

Hereinafter, based on the details described above, the operation flow of the present application will be briefly described.

8 is a flowchart illustrating a method for detecting a moving object according to an exemplary embodiment of the present disclosure.

The moving object detection method illustrated in FIG. 8 may be performed by the apparatus 10 described above. Therefore, even if omitted below, the description of the apparatus 10 may be equally applied to the description of the method of detecting a moving object.

Referring to FIG. 8, in step S11, the generator may generate silhouette information of the moving object by receiving a reflected signal changed by the moving object as the moving object passes through the observation point on the ground from the laser sensor.

In addition, the reflected signal changed in step S11 may be a signal received by reflecting an energy source irradiated in a vertical direction toward the upper surface of the moving object passing through the observation point.

The laser sensor may irradiate an energy source in a direction orthogonal to a moving direction of a moving object passing through the observation point and receive a reflected signal corresponding to the irradiated energy source.

Also, the silhouette information generated in step S11 may be information expressed by a relationship between a reflected signal measured over time.

In addition, the silhouette information generated in step S11 is from the first point in time when the reflected signal having a preset threshold value is first measured among the reflected signals transmitted from the laser sensor, and the reflected signal having a preset threshold value is measured second. It may be information generated by using a reflected signal belonging to two viewpoints as a changed reflected signal.

Next, in step S12, the detection unit may detect information on the moving object through analysis using the silhouette information generated in step S11. Here, the information on the moving object may include type (type) information and moving speed information of the moving object.

In addition, in step S12, the detection unit takes a plurality of training silhouette information as an input value and uses the type information of a moving object matched with each of the plurality of training silhouette information as an output value, based on a deep learning item based on a deep learning model. By applying the silhouette information to the model as input, it is possible to detect (estimate) type information of the moving object corresponding to the silhouette information as information about the moving object.

In addition, although not shown in the drawings, the method of detecting a moving object according to an exemplary embodiment of the present disclosure may include matching length information (full length information) for each type of a moving object before step S12 and storing it in a database.

At this time, in step S12, the detection unit may identify length information of the moving object from the database using the type information of the moving object detected using the deep learning model. Thereafter, the detection unit may detect movement speed information of the moving object corresponding to the silhouette information by using length information of the identified moving object and time information corresponding to the silhouette information.

In addition, in step S12, the detection unit may detect whether the moving object is relatively fast compared to the reference moving speed as information on the moving object, using the detected moving object type information and silhouette information.

In the above description, steps S11 and S12 may be further divided into additional steps or may be combined into fewer steps, according to an embodiment of the present disclosure. In addition, some steps may be omitted as necessary, and the order between steps may be changed.

The method of detecting a moving object according to an exemplary embodiment of the present disclosure may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -A hardware device specially configured to store and execute program instructions such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine language codes such as those produced by a compiler but also high-level language codes that can be executed by a computer using an interpreter or the like. The above-described hardware device may be configured to operate as one or more software modules to perform the operation of the present invention, and vice versa.

Further, the above-described moving object detection method may be implemented in the form of a computer program or application executed by a computer stored in a recording medium.

The foregoing description of the present application is for illustrative purposes only, and those of ordinary skill in the art to which the present application pertains will be able to understand that it is possible to easily transform it into other specific forms without changing the technical spirit or essential features of the present application. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as being distributed may also be implemented in a combined form.

The scope of the present application is indicated by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present application.

100: moving object detection system
1: moving object
2: laser sensor
10: moving object detection device
11: generation
12: detection unit
13: database

Claims (12)

In the moving object detection device,
A generator configured to generate silhouette information of the moving object by receiving the changed reflection signal by the moving object from the laser sensor as the moving object passes through the observation point on the ground; And
A detection unit for detecting information on the moving object through analysis using the silhouette information,
The information on the moving object includes type information and moving speed information of the moving object. The method of claim 1,
The laser sensor,
The moving object detection apparatus, wherein the energy source is irradiated in a direction orthogonal to the moving direction of the moving object passing through the observation point, and a reflection signal corresponding to the irradiated energy source is received. The method of claim 2,
The changed reflected signal,
The apparatus for detecting a moving object, wherein the laser sensor is a signal received by reflecting an energy source irradiated in a vertical direction toward an upper surface of the moving object passing through the observation point. The method of claim 1,
The silhouette information,
It is information expressed by the relationship of the reflected signal measured over time,
Among the reflected signals received from the laser sensor, a reflected signal belonging to a second point in time from a first point in time when a reflected signal having a preset threshold value is first measured and a second point in time when the reflected signal having the preset threshold value is measured second is the The information generated by using the changed reflected signal, the moving object detection device. The method of claim 1,
The detection unit,
The silhouette information is input to the deep learning model based on a deep learning model-based pre-learning item in which a plurality of training silhouette information is used as an input value and type information of a moving object matching each of the plurality of training silhouette information is used as an output value. To detect the type information of the moving object corresponding to the silhouette information by applying to. The method of claim 5,
Further comprising a database for matching and storing length information for each type of moving object,
The detection unit,
Identifying length information of the moving object from the database using the type information of the moving object detected using the deep learning model,
And detecting movement speed information of a moving object corresponding to the silhouette information by using length information of the identified moving object and time information corresponding to the silhouette information. The method of claim 5,
The detection unit,
Using the detected type information of the moving object and the silhouette information, as information on the moving object, detecting whether a relative moving speed is faster than a reference moving speed of the moving object. In the moving object detection system,
A laser sensor for irradiating an energy source in a direction orthogonal to a moving direction of a moving object passing through an observation point on the ground and receiving a reflected signal corresponding to the irradiated energy source; And
When the moving object passes through the observation point, a changed reflection signal by the moving object is received from the laser sensor to generate silhouette information of the moving object, and information on the moving object is detected through analysis using the silhouette information. Including a moving object detection device,
The information on the moving object includes type information and moving speed information of the moving object. In the moving object detection method,
Generating silhouette information of the moving object by receiving a reflected signal changed by the moving object as the moving object passes through the observation point on the ground from the laser sensor; And
And detecting information on the moving object through analysis using the silhouette information,
The information on the moving object includes type information and moving speed information of the moving object. The method of claim 9,
In the generating step, the changed reflected signal,
The method of detecting a moving object, wherein the energy source irradiated in a vertical direction toward the upper surface of the moving object passing through the observation point by the laser sensor is reflected and received. The method of claim 9,
The detecting step,
The silhouette information is input to the deep learning model based on a deep learning model-based pre-learning item in which a plurality of training silhouette information is used as an input value and type information of a moving object matching each of the plurality of training silhouette information is used as an output value. To detect the type information of the moving object corresponding to the silhouette information by applying to the moving object detection method. A computer-readable recording medium storing a program for executing the method of claim 9 on a computer.

Images