TWI773175B - Method and system for violation identifying - Google Patents

Abstract

A violation identifying method being implemented using a violation identifying system includes steps of: (A) Setting one of a plural violation identifying model as a target violation identifying model according to a current location. (B) Operating the target violation identifying model to perform image analysis to a dynamic image, and determining whether a trigger condition is met, wherein the trigger condition includes the position or movement of a vehicle being presented in the dynamic image matches a violation behavior feature corresponding to the target violation identifying model. (C) After determining the trigger condition is met, extracting a key part from the dynamic image according to the time point of the trigger condition is met, generating and storing violation data including the key part.

Description

Violation identification method and system

The present invention relates to a violation identification method, in particular to a violation identification method suitable for identifying vehicle violations. The present invention also relates to a violation identification system capable of implementing the violation identification method.

In the modern society with high vehicle density, if the driver of the vehicle violates the regulations for the sake of convenience, it will not only easily damage the rights and interests of other passers-by, but may even lead to accidents.

Although law enforcement officers sometimes crack down on illegal hotspots, vehicle violations occur anytime, anywhere, and the preventive effect of only relying on law enforcement officers to ban is actually quite limited. On the other hand, although the general public can use the video recorded by the dash cam to report, it is necessary to edit the video in the process of preparing the report material, which is time-consuming and labor-intensive. Therefore, how to make the report more convenient for the general public? Illegal vehicles have become a topic worthy of study.

Therefore, one of the objectives of the present invention is to provide a violation identification method that can assist the general public to report violation vehicles.

The violation identification method of the present invention is implemented by a violation identification system. The violation identification system stores a plurality of violation identification models, and the violation identification models respectively correspond to a variety of vehicle-related violation behavior characteristics. The violation identification method includes: (A ) The violation identification system sets at least one of the violation identification models related to the current geographic location as a target violation identification model according to a current geographic location, wherein the violation behavior characteristic corresponding to the target violation identification model Characterized as a target violation. (B) The violation identification system performs image identification on a dynamic image by running the target violation identification model and determines whether a trigger condition is met, wherein the trigger condition includes the position or movement of a vehicle displayed in the dynamic image in a manner consistent with the target violation characteristics. (C) After judging that the trigger condition is met, the violation identification system extracts a key segment from the moving image at least according to the time point when the trigger condition is met, and generates and stores a violation data including the key segment.

In some implementation aspects of the violation identification method of the present invention, the violation identification system further stores a model selection rule, and the model selection rule includes a plurality of non-overlapping geographic location ranges, and a plurality of geographic location ranges corresponding to the same geographic location ranges. model parameters, and each model parameter indicates at least one of the violation identification models. In step (A), the violation identification system sets the target violation identification model by first determining the target violation identification model according to the current geographic location. Whether there is a target geographic location range covering the current geographic location in the same geographic location range, and if the judgment result is yes, set each violation identification model indicated by the model parameters of the target geographic location range as the target geographic location range. Target Violation Identification Model.

In some implementation aspects of the violation identification method of the present invention, in step (B), the method of performing image identification on the dynamic image by the violation identification system is to first identify one or more images displayed by the dynamic image. license plate, and mark each object with the license plate in the dynamic image as a vehicle, and then determine whether the trigger condition is met by identifying whether the position and movement mode of each vehicle conform to the characteristics of the target violation.

In some implementation aspects of the violation identification method of the present invention, in step (C), the time point when the trigger condition is met is taken as a key time point, and the violation identification system is based on the key time point and at least one preset The time length determines a capture start time point and a capture end time point after the capture start time point, and then extracts from the dynamic image according to the capture start time point and the capture end time point Extract the key segment.

In some implementation aspects of the violation identification method of the present invention, in step (C), the violation identification system further extracts a plurality of key images from the dynamic image or the key segment at least according to the key time point, and, The offending material also contains the key images.

In some implementation aspects of the violation identification method of the present invention, in step (C), the violation data further includes a license plate number corresponding to the vehicle, a violation time related to the key time point, and a related violation time. The location of the violation at a critical point in time.

In some implementation aspects of the violation identification method of the present invention, in step (C), after generating the violation data, the violation identification system also transmits the violation data to an external electronic device, so that the external electronic device can identify the violation data. The offending data is stored.

In some implementation aspects of the violation identification method of the present invention, in step (A), when the violation identification system receives an activation instruction input by the user and indicating one of the violation identification models, The violation identification system sets the violation identification model indicated by the enabling instruction as another target violation identification model, and when the violation identification system receives a user input that indicates the target violation identification model When one of the deactivation instructions is issued, the violation identification system sets the target violation identification model indicated by the deactivation instruction as a non-target violation identification model.

In some implementation aspects of the violation identification method of the present invention, in step (C), after judging that the trigger condition is met, the violation identification system also outputs a warning instruction corresponding to the characteristics of the target violation behavior, and the warning instruction It is used for the external vehicle-mounted device to receive and output a pair of warning notices corresponding to the warning command according to the warning command.

Another object of the present invention is to provide a violation identification system capable of implementing the violation identification method.

The violation identification system of the present invention includes a storage unit and a processing unit electrically connected to the storage unit. The storage unit stores a plurality of violation identification models, and the violation identification models respectively correspond to a variety of vehicle-related violation behavior characteristics. The processing unit sets at least one of the violation identification models related to the current geographic location as a target violation identification model according to a current geographic location, wherein the violation behavior feature corresponding to the target violation identification model is used as A target violation characteristic. The processing unit performs image recognition on a dynamic image by running the target violation recognition model and determines whether a trigger condition is met, wherein the trigger condition includes that the position or movement pattern of a vehicle displayed in the dynamic image meets the target Violation characteristics. After judging that the trigger condition is satisfied, the processing unit extracts a key segment from the moving image at least according to the time point when the trigger condition is satisfied, and generates a violation data including the key segment and stores the violation data in the storage unit.

In some implementation aspects of the violation identification system of the present invention, the storage unit further stores a model selection rule, and the model selection rule includes a plurality of geographic location ranges that do not overlap with each other, and a plurality of models corresponding to the same geographic location ranges respectively. parameters, and each model parameter indicates at least one of the violation identification models. The processing unit sets the target violation identification model by first judging whether there is a target geographic location range covering the current geographic location in the geographic location ranges according to the current geographic location, and if the judgment result is yes, Each violation identification model indicated by the model parameter corresponding to the target geographic location range is set as the target violation identification model.

In some implementation aspects of the violation recognition system of the present invention, the processing unit performs image recognition on the dynamic image by first recognizing one or more license plates displayed by the dynamic image, and then using the dynamic image to identify one or more license plates. Each object with a license plate is marked as a vehicle, and then it is determined whether the trigger condition is met by identifying whether the position and movement mode of each vehicle conform to the characteristics of the target violation behavior.

In some implementation aspects of the violation identification system of the present invention, the time point when the trigger condition is satisfied is regarded as a key time point, and the processing unit determines a capture start according to the key time point and at least a predetermined time length a start time point and a capture end time point after the capture start time point, and then extract the key segment from the moving image according to the capture start time point and the capture end time point.

In some implementation aspects of the violation identification system of the present invention, the processing unit further extracts a plurality of key images from the dynamic image or the key segment at least according to the key time point, and the violation data further includes the key images image.

In some implementation aspects of the violation identification system of the present invention, the violation data further includes a license plate number corresponding to the vehicle, a violation time related to the critical time point, and a violation location related to the critical time point.

In some implementation aspects of the violation identification system of the present invention, after generating the violation data, the processing unit also transmits the violation data to an external electronic device for the external electronic device to store the violation data.

In some implementation aspects of the violation identification system of the present invention, when the processing unit receives an activation instruction input by the user and indicating one of the violation identification models, the processing unit applies the activation instruction to The indicated violation identification model is set as another target violation identification model, and when the processing unit receives a deactivation instruction input by the user and indicating one of the target violation identification models, the The processing unit sets the target violation identification model indicated by the deactivation instruction as a non-target violation identification model.

In some implementation aspects of the violation identification system of the present invention, after judging that the trigger condition is satisfied, the processing unit further outputs a warning instruction corresponding to the characteristics of the target violation behavior, and the warning instruction is used for the external vehicle-mounted device Receive and output a pair of alert notifications corresponding to the alert instruction according to the alert instruction.

The effect of the present invention is that: the violation identification system can set a corresponding target violation identification model according to the current geographical location, and perform image recognition on the dynamic image by running the target violation identification model to determine whether there is a position of someone else's vehicle. Or the movement method conforms to the target violation behavior characteristics corresponding to the target violation identification model (that is, judging whether the trigger conditions are met), whereby the violation identification system can automatically target the most likely local vehicle violations according to the current geographic location. It is not necessary to continuously detect all vehicle violation patterns at the same time, so it can be more efficient in the utilization of computing resources of image recognition. In addition, when it is determined that the trigger condition is met, the violation identification system can extract the key segment from the moving image according to the time point when the trigger condition is met, and then generate and store the violation data including the key segment, thereby , the user can directly use the violation data as evidence of the violation of the vehicle by others to report to the traffic violation adjudication authority without editing the dynamic image by himself. Therefore, the violation identification system can effectively reduce the time cost for the user to prepare reporting materials, and allows the user to report the illegal vehicle more conveniently, thereby improving the inconvenience of the prior art.

Before the present invention is described in detail, it should be noted: if not specifically defined, the "electrical connection" described in this patent specification generally refers to the "wired connection" between multiple electronic devices/devices/elements connected to each other through conductive materials. electrical connection” and “radio connection” for one-way/two-way wireless signal transmission through wireless communication technology. In addition, unless otherwise defined, the "electrical connection" described in this patent specification also generally refers to a "direct electrical connection" formed by directly connecting multiple electronic devices/devices/elements to each other, as well as multiple electronic devices/devices/components. The devices/components are also indirectly connected to each other through other electronic devices/devices/components.

Referring to FIG. 1 and FIG. 2 , an embodiment of the violation identification system 1 of the present invention is, for example, suitable for being installed in a vehicle (eg, a first vehicle 51 shown in FIG. 2 ), and the violation identification system 1 is mainly It is used to generate and retain evidence of violations of other vehicles (such as a second vehicle 52 shown in FIG. 2 ) by means of image recognition, so that users (such as the owner of the first vehicle 51 ) can more easily The traffic violation adjudication authority proposes a report on the violating vehicle.

In this embodiment, the violation identification system 1 is, for example, implemented as a digital video recording device (Digital Video Recorder in English, DVR for short) with satellite positioning function, that is, a driving recording device. In other words, the violation identification system 1 of the present embodiment can be independently manufactured and sold, and can be installed on various vehicles. However, in other embodiments, the violation identification system 1 is not limited to be implemented as a driving record device. For example, the violation identification system 1 can also be built into a vehicle before it leaves the factory. , which is equivalent to a set of driving assistance systems with image recording and satellite positioning functions. Therefore, the specific implementation of the violation identification system 1 is not limited to this embodiment.

More specifically, the violation identification system 1 of this embodiment includes, for example, a storage unit 11 , a photographing unit 12 , a positioning unit 13 , an input/output unit 14 , and an electrical connection between the storage unit 11 and the photographing unit 12 . , the positioning unit 13 and the processing unit 15 of the I/O unit 14 .

The storage unit 11 is implemented, for example, as a flash memory for storing digital data in this embodiment. However, in other embodiments, the storage unit 11 can also be implemented as a conventional hard disk, a solid-state hard disk, or other types of computer-readable recording media, or a plurality of computer-readable recording media of different types. combination, but not limited to this embodiment.

In this embodiment, the photographing unit 12 may, for example, be implemented as a photographing lens module including a lens group and a photosensitive element, and, taking the first vehicle 51 as an example, the photographing unit 12 is suitable for setting The purpose of photographing is toward the front side of the first vehicle 51, but it is not limited to this.

The positioning unit 13 is, for example, implemented as a satellite positioning module based on satellite positioning technology. More specifically, the positioning unit 13 can receive satellite signals to perform real-time positioning of the current location of the positioning unit 13 . More specifically, in this embodiment, the satellite signal may come from, for example, a global positioning system (Global Positioning System, GPS for short), that is, the positioning unit 13 in this embodiment is, for example, A GPS satellite positioning module. However, in other embodiments, the satellite signal may also come from other systems that utilize satellite signals to achieve real-time positioning, such as Beidou, which is generally referred to as a satellite navigation system (Global Navigation Satellite System, GNSS for short). Satellite navigation system (BDS), Galileo positioning system (GALILEO), GLONASS system (GLONASS), etc. Therefore, the specific implementation of the positioning unit 13 is not limited to this embodiment.

The I/O unit 14 may include, for example, at least one of a display screen, a button combination, a prompt light combination, and a speaker, and the display screen may be implemented as a touch display screen, but not as a limit.

The processing unit 15 can be implemented, for example, as a central processing unit. However, in other embodiments, the processing unit 15 can also be implemented, for example, as a control circuit board with a central processing unit, instead of the present embodiment. limit.

In this embodiment, the storage unit 11 stores, for example, a plurality of violation identification models M that can be run by the processing unit 15 for image recognition, and the storage unit 11 also stores a data related to the violation identification models M. The model uses the rule R.

In this embodiment, each violation identification model M is pre-trained by, for example, machine learning, and each violation identification model M enables the processing unit 15 to identify a vehicle with a license plate in a dynamic image. In addition, the violation identification models M respectively correspond to a variety of vehicle-related violation features, and each violation feature is, for example, a vehicle violation pattern defined based on traffic regulations (such as road traffic management penalties). .

More specifically, for example, the violation identification models M include, but are not limited to, a red light running identification model, an illegal line crossing identification model, and an illegal parking identification model.

Wherein, the red light recognition model not only enables the processing unit 15 to recognize the vehicle with the license plate, but also enables the processing unit 15 to further recognize the light signal of the traffic light in the dynamic image, and recognize the dynamic image. Check if a vehicle is passing the intersection at a red light. In other words, the characteristic of the violation behavior corresponding to the red light running recognition model is, for example, "red light running". Moreover, the data used to train the red light recognition model may include, but are not limited to, images of traffic lights, images of intersections with traffic lights, and images of vehicles passing through intersections with traffic lights. Videos, etc., but not limited to this.

On the other hand, the illegal crossing identification model can, for example, enable the processing unit 15 to further identify the prohibited lane changing lines (such as double white lines and double yellow lines) and trough lines in the moving image, and identify whether there is any lane change in the moving image. The vehicle crosses the said no-change lane line or trough line while moving. In other words, the illegal behavior characteristic corresponding to the illegal line crossing identification model is, for example, "crossing the prohibited lane line or the trough line". In addition, the data used to train the illegal line crossing recognition model may include, but are not limited to, images of prohibited lane changing and trough lines, and images or videos of vehicles crossing the prohibited lane changing or trough lines, etc., But not limited to this.

On the other hand, the illegal parking recognition model can, for example, enable the processing unit 15 to further recognize the no-parking line and the no-parking sign in the dynamic image, and identify whether there are vehicles parked side by side in the dynamic image, or stop at the No Parking Lines or No Parking Signs. In other words, the illegal behavior feature corresponding to the illegal parking identification model is, for example, "illegal parking". In addition, the data used for training the illegal parking recognition model may include, but not limited to, images of no-parking markings and no-parking signs, images of vehicles parked side by side, and the like.

It should be added that the aforementioned red light running recognition model, illegal line crossing recognition model and illegal parking recognition model are only exemplary implementations used to illustrate the illegal recognition models M in this embodiment. It should be understood that, The implementation of the violation identification models M is not limited to this embodiment, and the data used for training the violation identification models M is not the focus of this patent specification, so it will not be repeated.

In addition, in a further implementation aspect of this embodiment, the violation identification models M may, for example, further include a reverse driving identification model for the processing unit 15 to identify the characteristics of the violation behavior of “driving in the wrong direction”, and a reverse driving identification model for A light signal violation identification model for the processing unit 15 to identify the feature of the violation behavior of "turning or changing lanes without using the turn signal", but not limited thereto.

In this embodiment, the model selection rule R includes, for example, a plurality of non-overlapping geographic location ranges, and a plurality of model parameters corresponding to the same geographic location ranges, and each model parameter indicates the violation identification models M at least one of them. More specifically, in this embodiment, each geographic location range may include, for example, a plurality of geographic location coordinates, and the geographic location range may, for example, be jointly defined by the geographic location coordinates included in itself, Moreover, each geographic location range may, for example, represent a specific road segment on the map, a specific intersection formed by a plurality of specific road segments, or a specific area covering a plurality of specific road segments/specific intersections, and each The specific road segment may, for example, represent a complete specific road, or a part of a specific road, but is not limited thereto.

Further, in this embodiment, the model selection rule R may be generated by, for example, the processing unit 15 performing statistical analysis on a violation history case database. The historical case database summarizes the most frequently occurring illegal driving patterns in each geographic location range, and thereby determines which illegal identification models M are to be indicated by the model parameters corresponding to the geographic location range. For example, if a right turn at a red light occurs most frequently in one geographic location range, the model parameter corresponding to the geographic location range may, for example, indicate the red light running recognition model in the violation recognition models M. For another example, if the frequency of illegal crossing of marking lines and illegal parking in another geographic location range is very high, the model parameters corresponding to the other geographic location range may, for example, simultaneously indicate that the violation identification model M has a high frequency. The illegal crossing identification model and the illegal parking identification model. It should be added that, in a similar implementation aspect of this embodiment, the model selection rule R can also be generated by an analysis server after analyzing the violation history case database in the same way, and then providing it through the network Stored in the violation identification system 1, therefore, the manner in which the model selection rule R is generated is not limited to this embodiment.

Referring to FIGS. 1 and 2 together with FIG. 3 , the following takes the first vehicle 51 as an example to describe how the violation identification system 1 of the present embodiment implements a violation identification method during the driving of the first vehicle 51 .

First, in step S1, when the processing unit 15 receives an activation command generated by the I/O unit 14 according to the user's operation, the processing unit 15 controls the shooting unit 12 to continuously record video according to the activation command, In order to obtain a dynamic image generated by the photographing unit 12 from the photographing unit 12 , in other words, the dynamic image is a real-time video generated by the photographing unit 12 continuously recording in this embodiment. Next, the flow proceeds to step S2.

In step S2 , the processing unit 15 controls the positioning unit 13 to perform real-time positioning, so as to obtain a current geographic location from the positioning unit 13 . It should be noted that the current geographic location is, for example, implemented as a GPS positioning coordinate, and the current geographic location is not only equivalent to the current location of the violation identification system 1 itself, but also equivalent to the current location of the first vehicle 51 . . In addition, optionally, the processing unit 15, for example, will confirm whether the current time counted by the violation identification system 1 is accurate at regular intervals, and correct when the time error is too large. Next, the flow proceeds to step S3.

In step S3, the processing unit 15 sets at least one of the violation identification models M related to the current geographic location as a target violation identification model M' according to the current geographic location (exemplarily shown in FIG. 1 ). ). For the convenience of description, it is assumed here that the number of target violation identification models M' set by the processing unit 15 in this embodiment is one, and here, the violation behavior characteristics corresponding to the target violation identification model M' are taken as this A target violation feature in an embodiment.

More specifically, in this embodiment, the processing unit 15 sets the target violation identification model M' by first judging, according to the current geographic location, whether there is one of the geographic location ranges of the model selection rule R that covers the target violation identification model M'. The target geographic location range of the current geographic location, and if the determination result of the processing unit 15 is yes, the processing unit 15 sets the violation identification model M indicated by the model parameter corresponding to the target geographic location range as the target Violation identification model M'.

It is added that, if the model parameter corresponding to the target geographic location range indicates more than one (eg, two or more) of the violation identification models M, the processing unit 15 will set the model parameter to The indicated violation identification models M are respectively set as a plurality of target violation identification models M', that is to say, in an actual implementation, the number of target violation identification models M' set by the processing unit 15 can also be The number is more than one, and is not limited to this embodiment. On the other hand, if the processing unit 15 determines that the current geographic location is not located within any geographic location range, the process may start from step S2 again when a predetermined period ends, for example.

After the processing unit 15 sets the target violation identification model M' according to the current geographic location, the flow proceeds to step S4.

In step S4, the processing unit 15 performs image recognition on the moving image by running the target violation recognition model M', and determines whether a trigger condition is met according to the result of the image recognition. If the judgment result of the processing unit 15 is NO, the flow proceeds to step S5, and if the judgment result of the processing unit 15 is YES, the flow proceeds to step S6.

Specifically, in this embodiment, the triggering condition, for example, represents that the position or movement method of another vehicle (here, the second vehicle 52 in FIG. 2 is taken as an example) displayed in the dynamic image complies with the target violation Identify the characteristics of the target violation behavior corresponding to the model M'.

For example, assuming that the processing unit 15 sets the red light running recognition model as the target violation recognition model M', if the processing unit 15 recognizes that the movement mode of the second vehicle 52 meets the When the violation characteristic of "running a red light" is found, the processing unit 15 will determine that the triggering condition is met. For another example, assuming that the processing unit 15 sets the illegal parking identification model as the target illegal parking identification model M', if the processing unit 15 recognizes that the second vehicle 52 is the second vehicle 52 by running the illegal parking identification model The processing unit 15 will determine that the triggering condition is met when the vehicle stops at the no-parking line or the no-parking sign and meets the illegal behavior characteristic of "illegal parking".

It is worth mentioning that, in this embodiment, the processing unit 15 performs image recognition on the dynamic image by first recognizing one or more license plates displayed by the dynamic image, and then in the dynamic image. Each object with a license plate is marked as a vehicle (such as the second vehicle 52 ), and then, the processing unit 15 identifies the target corresponding to the target violation identification model M' by identifying the position and movement of each vehicle. Violation characteristics to determine whether the trigger condition is met. With the aforementioned image recognition method in this embodiment, the processing unit 15 only regards the object with the license plate in the dynamic image as a vehicle, that is to say, the processing unit 15 in this embodiment first recognizes the object with the license plate from the dynamic image Vehicles with license plates, and then identify whether the position and movement mode of each vehicle conform to the characteristics of the target violation. Therefore, the processing unit 15 of this embodiment will not target objects without license plates in the dynamic image (such as bicycles and Electric bicycles) to identify the characteristics of target violations, so it can avoid wasting computing resources on objects (objects) that are difficult to report. However, in other embodiments, the processing unit 15 may, for example, first recognize that the position or movement of an object conforms to the characteristics of the target violation behavior, and then determine whether the object is a vehicle according to whether the object has a license plate or not. Therefore, the manner in which the processing unit 15 performs image recognition on the moving image is not limited to this embodiment.

In step S5 following step S4, once the processing unit 15 determines that the trigger condition is not met, the processing unit 15, for example, controls the positioning unit 13 again to perform real-time positioning, and judges according to the positioning result of the positioning unit 13 Whether the violation identification system 1 itself is still within the target geographic location range (equivalent to judging whether the first vehicle 51 is still within the target geographic location range). If the result of the judgment of the processing unit 15 is yes, the process starts to execute again from step S4, for example, so as to continue to judge whether the trigger condition is met. And if the result of the judgment of the processing unit 15 is no, the process starts to execute again from step S2, for example, to re-set the target violation identification model M', but not limited to this.

In step S6 following step S4, after the processing unit 15 determines that the trigger condition is met, the processing unit 15 extracts a key segment and a plurality of key segments from the moving image at least according to the time point when the trigger condition is met key image.

More specifically, the processing unit 15 determines that the time point at which the trigger condition is met is taken as a key time point in this embodiment, that is, the key time point is equivalent to the time when the second vehicle 52 is in the dynamic image. The point in time when the behavior matches the behavior characteristic of this target violation. Moreover, in this embodiment, the processing unit 15 retrieves the key segment by, for example, firstly determining a capture start time point and a and the capture end time point after the start time point, and then the key segment is retrieved from the moving image according to the capture start time point and the capture end time point. Moreover, in this embodiment, the capture start time point may be determined by, for example, the processing unit 15 pushing back the key time point by the preset time length, and the capture end time point may be determined by, for example, The processing unit 15 is determined by pushing the critical time point back by the preset time length, but not limited thereto.

For example, assuming that the key time point is at the 5th minute and 30th second of the dynamic video, and assuming that the preset time length is 10 seconds, in this example, the capture start time point is, for example, from The 5th minute and 30th second of the dynamic video is pushed back by 10 seconds (that is, the preset time length is pushed back), and it will be determined as the 5th minute and 20th second of the dynamic video. On the other hand, the capture termination time point is, for example, pushed back by 10 seconds from the 5th minute and 30 seconds of the dynamic video (that is, pushed back by the preset time length), and is determined to be the fifth time point of the dynamic video 40 seconds. Therefore, in this example, the key segment will be the portion of the moving image from the 5th minute and 20th to the 5th minute and 40th.

By means of the processing unit 15 retrieving the key segment in this embodiment, the key segment can present at least a part of the process that the behavior of the second vehicle 52 conforms to the characteristics of the target violation behavior, in other words, the Key segments can be used as film evidence that the second vehicle 52 constitutes a violation.

On the other hand, the processing unit 15 captures the key images, for example, the dynamic image at the key time point, one second before the key time point, and one second after the key time point. Three frames (Frames) are respectively used as the key images, but not limited thereto. In other words, the key images are equivalent to photos taken from the dynamic image, and may be used as photo evidence that the second vehicle 52 constitutes a violation. It should be added that in other embodiments, the number of key images captured by the processing unit 15 may also be one, two, four or more, and the time between the key images The interval length is not limited to the aforementioned one second, and the processing unit 15 can also extract the key images from the key segment after extracting the key segment from the moving image. , the manner in which the processing unit 15 retrieves the key images is not limited to this embodiment. Moreover, in some embodiments, the processing unit 15 may only capture the key segment without capturing the key images in step S6.

After the processing unit 15 captures the key segment and the key images, the process proceeds to step S7.

In step S7 , the processing unit 15 generates a violation data, and stores the violation data in the storage unit 11 . In this embodiment, the violation data includes, for example, the key segment, the key images, a license plate number corresponding to the second vehicle 52 , a violation time related to the key time point, and a related key to the key The location of the violation at the time point, but not limited thereto. More specifically, the license plate number is obtained, for example, by the processing unit 15 in the process of performing image recognition on the dynamic image. On the other hand, the violation time is, for example, the result of converting the critical time point into the actual environment time by the processing unit 15 according to the electronic clock function built in the violation identification system 1 (for example, 3 pm on a certain day in a certain year). 20:00). On the other hand, the illegal location is, for example, implemented as an address, and the illegal location is generated by, for example, the processing unit 15 converting the positioning result of the positioning unit 13 at the key time point according to an electronic map, but not Not limited to this. Since the violation data contains the necessary information that the general public needs to provide when reporting the violation vehicle, the user (for example, the owner of the first vehicle 51 ) can conveniently use the violation data to report the violation to the traffic violation adjudicating authority. Two violations of the vehicle 52, without the need to edit the dynamic image.

After the processing unit 15 generates the violation data or stores the violation data in the storage unit 11, optionally, the process proceeds to step S8.

In step S8, the processing unit 15, for example, transmits the illegal data to an external electronic device (not shown) for the external electronic device to store the illegal data. More specifically, in this embodiment, the external electronic device may be, for example, a cloud server that provides a cloud hard disk service, and the processing unit 15, for example, automatically transmits the illegal data after generating the illegal data. The network is sent to the cloud server for storage, so that the user can conveniently download the illegal data from the cloud server through a mobile phone or a computer without needing to perform the operation of capturing the data by the illegal identification system 1 .

It should be added that, in step S8 of another embodiment, the external electronic device may also be, for example, a mobile electronic device (such as a mobile phone) held by the user, and the processing unit 15 is, for example, generating the violation After the data, the illegal data is automatically transmitted to the mobile electronic device through bluetooth or network for storage, so that the user does not need to transfer the illegal data from the illegal identification system 1 to his mobile phone by himself. Further, in step S8 of another embodiment, the external electronic device may also be, for example, a traffic violation report acceptance server managed by a traffic violation adjudication authority and providing an online report service, and the processing unit 15 may be, for example, It is to transmit the violation data to the traffic violation report acceptance server through the network under the condition of confirmation by the user, so that the user can directly operate the violation identification system 1 to complete the detection of the second vehicle 52. Report without having to transfer the offending data to your own mobile phone or computer in order to report.

In addition, in step S3 of a further implementation aspect of this embodiment, after the processing unit 15 completes the setting of the target violation identification model M' according to the current geographic location, the processing unit 15 can, for example, control the input/output unit 14 A violation mode setting interface is displayed, and the violation mode setting interface can, for example, indicate all the violation identification models M, and which violation identification models M are currently set as the target violation identification model M'. Moreover, when the processing unit 15 receives an activation instruction generated by the I/O unit 14 according to the user's input and indicating one of the violation identification models M, the processing unit 15 will activate according to the activation instruction. instruction to set the violation identification model M indicated by the enable instruction as another target violation identification model M'. On the other hand, when the processing unit 15 receives a deactivation instruction generated by the I/O unit 14 according to the user's input and indicating one of the target violation identification model(s) M', the The violation identification system 1 sets the target violation identification model M' indicated by the deactivation instruction as a non-target violation identification model according to the deactivation instruction. More specifically, the enabling command and the disabling command are, for example, generated by the I/O unit 14 according to the user's operation of the violation mode setting interface, that is, although the processing unit 15 The geographic location automatically sets the target violation identification model M', but the user can still freely decide which of the violation identification models M to set as the target violation identification model M' through the violation aspect setting interface, but this is not the case. limit.

On the other hand, in step S6 of a further implementation aspect of this embodiment, when the processing unit 15 determines that the trigger condition is met, optionally, the processing unit 15 further determines, for example, the target violation identification model M' Whether the corresponding target violation behavior feature belongs to a preset high-risk violation behavior feature (for example, driving in the opposite direction, but not limited thereto), and, in the case that the judgment result is yes, the processing unit 15, for example The I/O unit 14 is controlled to output a warning reminder, thereby prompting the driver of the first vehicle 51 to have a high-risk illegal driving behavior on the current road section, so as to increase the alertness of the driver of the first vehicle 51 . Moreover, in the step S6 of the further implementation aspect, when the processing unit 15 determines that the trigger condition is met, it can also output a warning instruction corresponding to the characteristics of the target violation behavior, for example, by wireless transmission, and , the warning instruction is, for example, used for one or more external vehicle-mounted devices (not shown) to receive wirelessly, and make the external vehicle-mounted device(s) output a corresponding warning instruction according to the warning instruction. A warning notice, and the warning notice, for example, indicates the vehicle violation state represented by the target violation behavior characteristic (such as "running a red light"). To be more specific, each external vehicle-mounted device is, for example, disposed on a neighboring vehicle whose distance from the first vehicle 51 does not exceed a distance threshold, that is to say, the warning command output by the processing unit 15 is used for Each external vehicle-mounted device is used to output the warning notification, thereby reminding the driver of the adjacent vehicle that there is a violation of the vehicle in the vicinity, thereby helping the driver of the adjacent vehicle to be more alert to reduce the possibility of accident.

The above is an example description of the violation identification system 1 and the violation identification method implemented by the violation identification system 1 of the present embodiment. It should be understood that the above-mentioned steps S1 to S8 and the flowchart in FIG. 2 are only used to illustrate one of the possible implementations of the violation identification method, and are not used to limit the implementation scope of the present invention. For example, in other embodiments, the order in which step S1 and step S2 are performed may be reversed to that of the present embodiment, or may be performed simultaneously. Therefore, the order of steps in the method for identifying violations does not vary. It is not limited to this embodiment.

To sum up, by implementing the violation identification method, the violation identification system 1 can set a corresponding target violation identification model M' according to the current geographic location, and execute the target violation identification model M' to perform a dynamic image analysis. Image recognition, to determine whether the position or movement mode of other people's vehicles conform to the target violation behavior characteristics corresponding to the target violation identification model M' (ie, to determine whether the trigger conditions are met), whereby the violation identification system 1 can According to the current geographic location, it automatically detects the most likely vehicle violations in the local area, instead of continuously detecting all vehicle violations at the same time, so it can be more efficient in the utilization of computing resources for image recognition. In addition, when it is determined that the trigger condition is met, the violation identification system 1 can extract the key segment and the key images from the dynamic image according to the time point when the trigger condition is met, and then generate and store the key segment including the key The video clips and the illegal data of these key images, so that the user can directly use the illegal data as evidence of the violation of other vehicles to report to the traffic violation adjudication authority without editing the dynamic image by himself. Therefore, the violation identification system 1 can effectively reduce the time cost for the user to prepare the report data, and allow the user to report the violation vehicle more conveniently, and can indeed achieve the purpose of the present invention.

However, the above are only examples of the present invention, and should not limit the scope of the present invention. Any simple equivalent changes and modifications made according to the scope of the application for patent of the present invention and the content of the patent specification are still within the scope of the present invention. within the scope of the invention patent.

1: Violation identification system 11: Storage unit 12: Shooting unit 13: Positioning unit 14: I/O unit 15: Processing unit M: Violation Identification Model M': target violation identification model R: model selection rules 51: The first vehicle 52: Second vehicle S1~S8: Steps

Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, wherein: 1 is a schematic block diagram of an embodiment of a violation identification system of the present invention; FIG. 2 is a schematic diagram exemplarily illustrating a first vehicle and a second vehicle provided with the embodiment; and FIG. 3 is a flowchart illustrating how the embodiment implements a violation identification method.

S1~S8: Steps

Claims (10)

A violation identification method is implemented by a violation identification system, the violation identification system stores a plurality of violation identification models, and the violation identification models respectively correspond to a variety of vehicle-related violation behavior characteristics, and the violation identification method includes: (A) The violation identification system sets at least one of the violation identification models related to the current geographic location as a target violation identification model according to a current geographic location, wherein the violation corresponding to the target violation identification model The behavioral feature is used as a target violation behavioral feature; (B) The violation identification system performs image identification on a dynamic image by running the target violation identification model and determines whether a trigger condition is met, wherein the trigger condition includes the position or movement of a vehicle displayed in the dynamic image in a manner consistent with the characteristics of the targeted violation; and (C) After judging that the trigger condition is met, the violation identification system extracts a key segment from the moving image at least according to the time point when the trigger condition is met, and generates and stores a violation data including the key segment. The violation identification method according to claim 1, wherein the violation identification system further stores a model selection rule, and the model selection rule includes a plurality of geographic location ranges that do not overlap with each other, and a plurality of geographic location ranges respectively corresponding to the same geographic location range. model parameters, and each model parameter indicates at least one of the violation identification models. In step (A), the violation identification system sets the target violation identification model by first determining the target violation identification model according to the current geographic location. Whether there is a target geographic location range covering the current geographic location in the same geographic location range, and if the judgment result is yes, set each violation identification model indicated by the model parameters of the target geographic location range as the target geographic location range. Target Violation Identification Model. The method for identifying violations of claim 1, wherein, in step (B), the method for performing image identification on the dynamic image by the illegal identification system is to first identify one or more images displayed by the dynamic image. license plate, and mark each object with the license plate in the dynamic image as a vehicle, and then determine whether the trigger condition is met by identifying whether the position and movement mode of each vehicle conform to the characteristics of the target violation. The violation identification method according to claim 1, wherein, in step (C), a time point when the trigger condition is met is taken as a key time point, and the violation identification system is based on the key time point and at least one preset The time length determines a capture start time point and a capture end time point after the capture start time point, and then extracts from the dynamic image according to the capture start time point and the capture end time point Extract the key segment. The violation identification method according to claim 4, wherein, in step (C), the violation identification system further extracts a plurality of key images from the moving image or the key segment at least according to the key time point, and, The offending material also contains the key images. The violation identification method according to claim 1, wherein, in step (C), the violation data further includes a license plate number corresponding to the vehicle, a violation time related to the key time point, and a related violation time. The location of the violation at a critical point in time. The violation identification method according to claim 1, wherein, in step (C), after the violation identification system generates the violation data, the violation identification system also transmits the violation data to an external electronic device, so that the external electronic device can identify the violation data. The offending data is stored. The violation identification method according to claim 1, wherein, in step (A), when the violation identification system receives an activation instruction input by the user and indicating one of the violation identification models, The violation identification system sets the violation identification model indicated by the enabling instruction as another target violation identification model, and when the violation identification system receives a user input that indicates the target violation identification model When one of the deactivation instructions is issued, the violation identification system sets the target violation identification model indicated by the deactivation instruction as a non-target violation identification model. The violation identification method according to claim 1, wherein, in step (C), the violation identification system outputs a warning instruction corresponding to the characteristics of the target violation behavior after judging that the trigger condition is met, and the warning instruction It is used for the external vehicle-mounted device to receive and output a pair of warning notices corresponding to the warning command according to the warning command. A violation identification system includes a storage unit and a processing unit electrically connected to the storage unit, the storage unit stores a plurality of violation identification models, and the violation identification models respectively correspond to a variety of vehicle-related violation behavior characteristics, and , the processing unit can implement the violation identification method described in any one of claim items 1 to 9.

Images