WO2020001263A1

WO2020001263A1 - Parking space recognition

Info

Publication number: WO2020001263A1
Application number: PCT/CN2019/090635
Authority: WO
Inventors: 高海珂
Original assignee: 杭州海康威视数字技术股份有限公司
Priority date: 2018-06-25
Filing date: 2019-06-11
Publication date: 2020-01-02
Also published as: CN110633606A

Abstract

The present invention provides a parking space recognition method, apparatus, and device, and storage medium, said parking space recognition method comprising: recognizing whether vehicle information and a parking space identifier are present in a parking space area image block in a captured image; said parking space area image block refers to an image block, in said image, corresponding to a parking space area comprising a plurality of vertically overlapping parking spaces; the parking space identifier is used for identifying the parking space; if it is recognized that the parking space area image block has vehicle information and the parking space identifier is not present, then determining that the current state of the parking space area is occupied; if it is recognized that the parking space area image block does not have vehicle information and the parking space identifier is present, then determining that the current state of the parking space area is unoccupied.

Description

Parking space recognition

Cross-reference to related applications

This patent application claims priority from a Chinese patent application filed on June 25, 2018 with an application number of 201810663986.9 and an invention name of "a method, device, device, and storage medium for parking space identification". The entire text of the application is incorporated by reference. The way is incorporated in this article.

Technical field

The invention relates to the technical field of intelligent parking, in particular to a method, a device and a device for identifying parking spaces, and a storage medium.

Background technique

In the field of intelligent parking, it is usually to automatically identify whether there is a car in the parking space to determine the occupancy of the parking space, so as to reasonably manage the parking space.

At present, in a flat garage, the image collected by the camera is used to determine whether there is a car in the parking space. Furthermore, the position of the camera used to collect the parking space image is used to determine which parking space the vehicle is parked in or which parking space is occupied.

However, for a mechanical garage, such as a multi-level circular three-dimensional garage, referring to FIG. 1, the camera 1 'can only be installed on the first floor 23' of the mechanical garage, so the camera 1 'can only see the first floor 23 of the mechanical garage. ', And the other layers such as the second layer 22' and the third layer 21 'are outside the field of view of the camera. Therefore, the image collected by the camera 1 'can only reflect the parking occupancy of the first layer 23'. When the car moves out of the field of view of the camera 1 ', it cannot be determined which parking space the vehicle is parked in or whether it is occupied. . Therefore, the method of determining the occupancy of the parking space in combination with the position of the camera used to collect the parking space image has certain limitations on the mechanical garage.

Summary of the invention

In view of this, the present invention provides a parking space identification method, device, equipment and storage medium, which can be applied to determining the parking space occupation situation in a mechanical garage.

A first aspect of the present invention provides a parking space identification method, including:

Identifying whether there is vehicle information and a parking space identifier in a parking area tile in the collected image, wherein the parking area tile refers to a tile in the image corresponding to a parking space area including a plurality of vertically overlapping parking spaces, the A parking space identifier is used to identify the parking space;

When it is identified that vehicle information exists in the parking space area tile and the parking space identifier does not exist, it is determined that the current status of the parking space area is occupied;

When it is identified that no vehicle information exists in the parking space area tile and the parking space identifier exists, it is determined that the current state of the parking space area is unoccupied.

According to an embodiment of the present invention, the method further includes:

When it is identified that no vehicle information exists in the parking space area tile and the parking space identifier does not exist, it is determined that the current state of the parking space area is unoccupied.

According to an embodiment of the present invention, the parking space identification is a two-dimensional code identification, and / or a number, and / or a character, and / or a barcode identification; the parking space identification provided on different parking spaces is different.

According to an embodiment of the present invention, identifying whether a parking space area tile in the captured image has a parking space identifier includes:

The acquired image is input into a preset neural network, and the neural network recognizes a parking space identifier in the image, and outputs the identified parking space identifier.

Comparing the current state of the parking space area with the previous state of the parking space area, wherein the previous state of the parking space area is based on the previous frame for the parking space area relative to the captured image Determined by image;

If the current state is different from the previous state and the current state is unoccupied, updating the state of the parking space corresponding to the parking space identifier to unoccupied;

If the current state is different from the previous state and the current state is occupied, the state of the parking space corresponding to the parking space identifier identified from the previous frame image is updated to be occupied.

If the state of the parking space is occupied, query whether there is a binding relationship between the parking space and a vehicle occupying the parking space, and if the binding relationship does not exist, establish the binding relationship;

If the state of the parking space is unoccupied, query whether there is a binding relationship between the parking space and the vehicle, and if the binding relationship exists, release the binding relationship.

A second aspect of the present invention provides a parking space identification device, including:

A recognition unit, configured to identify whether there is vehicle information and a parking space identifier in a parking area tile in the collected image, wherein the parking area tile refers to a parking space area in the image corresponding to a parking space area including a plurality of vertically overlapping parking spaces; A tile, the parking space identifier is used to identify the parking space;

A first determining unit, configured to determine that the current state of the parking space area is occupied when it is recognized that the parking space area tile has vehicle information and the parking space identifier does not exist;

A second determining unit is configured to determine that the current state of the parking space area is unoccupied when it is recognized that the parking space area tile does not have vehicle information and the parking space identifier exists.

According to an embodiment of the present invention, the apparatus further includes:

A third determining unit is configured to determine that the current state of the parking space area is unoccupied when it is identified that the parking area area tile does not have vehicle information and the parking space identifier does not exist.

According to an embodiment of the present invention, the identification unit further includes:

A neural network is configured to recognize a parking space identifier in the image based on the inputted captured image, and output the recognized parking space identifier.

A comparison unit is configured to compare the current state of the parking space area with a previous state of the parking space area, wherein the previous state of the parking space area is based on Determined by the previous frame of the image;

A first updating unit, configured to update the state of a parking space corresponding to the parking space identifier to be unoccupied if the current state is different from the previous state and the current state is unoccupied;

A second updating unit, configured to update the status of the parking space corresponding to the parking space identification identified from the previous frame image to, if the current status is different from the previous status and the current status is occupied; Occupied.

A binding unit is configured to query whether there is a binding relationship between the parking space and a vehicle occupying the parking space if the status of the parking space is occupied; if the binding relationship does not exist, establish the binding relationship Fixed relationship

The unbinding unit is configured to query whether there is a binding relationship between the parking space and the vehicle if the state of the parking space is not occupied, and release the binding relationship if the binding relationship exists.

According to a third aspect of the present invention, there is provided an electronic device including a processor and a memory; the memory stores a program that can be called by the processor; wherein when the processor executes the program, it implements any one of the foregoing embodiments. The parking space identification method described in the item.

A fourth aspect of the present invention provides a machine-readable storage medium on which a program is stored, and when the program is executed by a processor, the parking space identification method according to any one of the foregoing embodiments is implemented.

Compared with the prior art, the embodiments of the present invention have at least the following beneficial effects:

The parking space is marked with a parking space identifier, so it can be determined which parking space is identified when the parking space identification is identified. After the identification of the parking space area is determined once, the parking space area's occupancy can be obtained later based on the parking space identification. The determination and management of the situation can of course also be used in a flat garage, which has a wider scope of application; the presence of a parking space identifier and vehicle information in the parking area of the image at the same time as a condition for determining whether the parking space is occupied can also improve the determination result of the parking space occupation situation Accuracy.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of a field of view of an image collected by a camera in a mechanical garage in the prior art; FIG.

2 is a schematic flowchart of a parking space identification method according to an embodiment of the present invention;

3 is a schematic diagram of an acquired image according to an embodiment of the present invention;

4 is a schematic flowchart of a parking space identification method according to another embodiment of the present invention;

5 is a structural block diagram of a parking space identification device according to an embodiment of the present invention;

6 is a structural block diagram of a parking space identification device according to another embodiment of the present invention;

FIG. 7 is a structural block diagram of an electronic device according to an embodiment of the present invention.

detailed description

Exemplary embodiments will be described in detail here, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present invention. Rather, they are merely examples of devices and methods consistent with some aspects of the invention as detailed in the appended claims.

The terminology used in the present invention is for the purpose of describing particular embodiments and is not intended to limit the present invention. The singular forms "a," "the," and "the" as used in this invention and the appended claims are also intended to include the majority, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

In order to make the description of the embodiments of the present invention clearer and more concise, some technical terms are explained below.

Parking guidance: provide users with guidance on empty parking spaces. For example, in a large parking lot with a large passenger flow, it is inconvenient to find an empty parking space after the user's vehicle enters the parking lot. The parking guidance function can more quickly determine the location of the empty parking space and guide the user's vehicle to the empty parking space Parking makes it easier to find empty parking spaces. The device for parking induction may include an intersection guidance screen and / or an LED (Light-Emitting Diode) indicator provided above the parking space. For example, green may be used to indicate that there is an available parking space and red to indicate that there is no available parking space; or An LED indicator shows the status of 3 parking spaces; or an LED indicator shows whether a parking space is occupied, etc. An empty space means that the parking space is not occupied by the vehicle, and a full space means that the parking space is occupied by the vehicle.

Reverse car search: The user can query the specific location of the car in the parking lot by entering the car information, such as the license plate number. For example, in a large parking lot with a large passenger flow, the user parks the vehicle into the parking space of the parking lot and leaves. When the user returns to pick up the vehicle, he often cannot find the vehicle or forgets where the vehicle is parked. And through the reverse car search function, users can query the parking position of the vehicle, so as to find the vehicle accurately. The device for realizing reverse car search may be a fixed query device in a parking lot, or a smart terminal such as a mobile phone that installs or accesses a specific application program. This disclosure does not limit this.

The following describes the parking space identification method in the embodiment of the present invention, but it should not be limited to this.

In one embodiment, referring to FIG. 2, a method for identifying a parking space is shown. The method includes the following steps.

S101: Identify whether there is vehicle information and a parking space identifier in the parking area tile in the acquired image.

S102: When it is identified that vehicle information exists in the parking space area tile and the parking space identifier does not exist, it is determined that the current state of the parking space area is occupied.

S103: When it is identified that no vehicle information exists in the parking space area tile and the parking space identifier exists, it is determined that the current state of the parking space area is unoccupied.

There is no sequence between the above steps S102 and S103, and it depends on the recognition result.

The parking space recognition method according to the embodiment of the present invention may be applied to an image processing device. The image processing device may be any device capable of processing acquired images, such as a DSP (Digital Signal Processor) device. Of course, the image processing device may have an image acquisition function itself, for example, a smart camera; or it may not have an image acquisition function, but instead receives and processes the real-time image acquisition input from the image acquisition device through a connection with the image acquisition device. Is not limited.

In step S101, it is identified whether there is vehicle information and a parking space identifier in the parking area tile in the acquired image. The parking area tile refers to a tile in the image corresponding to a parking space area including a plurality of vertically overlapping parking spaces, and the parking space identifier is used to identify the parking space.

The acquired image may be an image acquired in real time by an image processing device, or may be an image acquired from a device used to acquire an image in real time. For the sake of simplicity, the present disclosure is described below by taking an image acquired by an image acquisition device as an example. The image contains at least one parking area tile. Each parking space is provided with a unique parking space identification, and the corresponding parking space can be uniquely determined by identifying the parking space identification. It should be noted that, for the mechanical garage shown in FIG. 1, in the present disclosure, each parking space (car board) on each floor has a unique parking space identification. Real-time processing of the collected images can provide services such as parking guidance and reverse car search in a timely manner, avoiding situations where there are vacancies but no vacancy information can be provided.

For example, as shown in Figure 3, the captured image can include parking area tile 10 and parking area tile 20. Generally speaking, the image collected for a mechanical garage will include a parking area tile, and for a plane The image collected by the garage will include more than one parking area tile, of course, it is not limited to this. For example, if a mechanical garage is extended so that two parking spaces can be lowered to the first floor at the same time, the camera can capture images of the two parking spaces at the same time, or the image collected by the camera in the flat garage can have only one parking area tile.

When there is more than one parking area tile in the image, the subsequent steps can be performed for each parking area tile separately, in the same manner as the image includes only one parking area tile. The position of the parking area tile in the image can be determined by image recognition of the parking space boundary, or it can be set in advance (when the camera's shooting orientation is fixed), which is not specifically limited.

When the image contains more than one parking area tile, you can first identify one of the parking area tiles, and then perform the subsequent steps on the parking area tile; you can also perform the same operation on two or more parking area tiles at the same time. Recognition, the recognition result is distinguished according to the parking area block, and then the subsequent steps are performed for the recognition result of each parking area block, without specific restrictions.

The purpose of identification is to find out the vehicle information and parking space identification in the parking space area tile of the image. And, as the vehicle in the parking space is parked, the vehicle leaves, and the person or object is blocked, the recognition result is dynamically changed. The recognition result includes the following possibilities: there is a parking space identifier and no vehicle information, there is no parking space identifier and vehicle information, there is no parking space identifier and vehicle information, and there is both a parking space identifier and vehicle information.

The feature recognition method of the image may be implemented by using an object recognition technology in image processing that is well known to those skilled in the art, and is not specifically limited. Of course, it is also possible to perform pre-processing on the image and then perform feature recognition to ensure the accuracy of the recognition result.

If there is no parking space identification but vehicle information on the parking area tile of the image, it means that the current state of the parking area is occupied, that is, it is full. If there is a parking space logo and no vehicle information on the parking area tile of the image, it means that the current state of the parking space area is unoccupied, that is, it is empty.

A parking space identifier is set on each parking space to identify the corresponding parking space, thereby managing information about the parking space. The information about the parking space may include parking space occupation information, parked vehicle information, parking space position information, and the like.

For example, in a mechanical garage, vehicles are driven in and out on the first floor, and the camera's field of view is also on the first floor. When the parking space is outside the field of view of the camera, the corresponding parking space identification of the parking space can be queried to determine the parking space occupancy. Or, the corresponding parking space can be lowered to the first floor according to the parking space identification, and the parking space occupancy can be identified based on the vehicle information and the parking space identification. In this way, even when the parking space is not in the field of view of the camera, the occupancy situation of the parking space can be known, which facilitates the realization of the guidance of the vacant position in parking guidance and the guidance of the vehicle position in reverse car searching.

In the embodiment of the present invention, a parking space identifier is marked on each parking space, so when a parking space identifier is identified, which parking space can be determined. After the occupancy of the parking space is determined, the occupancy of the parking space can be obtained later according to the parking space identification. Applicable to the determination and management of parking spaces in mechanical garages. In addition, the presence of the parking space identifier and the vehicle information in the parking area block of the image at the same time as the conditions for determining whether the parking space is occupied can improve the accuracy of the determination result of the parking space occupation situation.

As shown in FIG. 5, the method flow of steps S101 to S103 may be executed by the parking space identification device 200. The parking space identification device 200 mainly includes an identification unit 201, a first determination unit 202, and a second determination unit 203. The identification unit 201 is configured to perform the foregoing step S101, the first determination unit 202 is configured to perform the foregoing step S102, and the second determination unit 203 is configured to perform the foregoing step S103.

In one embodiment, the parking space identification method further includes the following step S104 (not shown in FIG. 2): when it is identified that the parking space area tile does not have vehicle information and the parking space identifier does not exist, determining the parking space The area's current status is unoccupied.

Based on the recognition in step S101, in step S104, it can be determined whether the parking space area is occupied by the vehicle according to the recognition result. When the recognition result is that no vehicle information exists in the parking area tile and no parking space identifier exists, it is determined that the current state of the parking space area is unoccupied.

If the judgment is made only on the presence of a parking area tile in the image, it is determined that the current status of the parking area is unoccupied when the parking space identification exists, and it is determined as the parking area area if the parking area identification does not exist. The current status is occupied, and it is only guaranteed to determine the status of the parking area in the absence of abnormal conditions. If the obstruction obscures the parking space mark, the state of the accurate parking area cannot be obtained. For example, when there is no parking mark in the parking area tile, the status of the parking area is determined to be occupied. However, it may be that the parking space is blocked by a block. At this time, the status of the parking area is unoccupied and an error occurs. Sentence.

In the embodiment of the present invention, whether the parking space area and the vehicle information are present in the parking area block of the image is used as a condition for judging whether the parking area area is occupied. It is determined that the parking space area is not occupied by the vehicle, thereby avoiding the problem of erroneously judging that the parking space area is occupied by the vehicle because the parking space mark is blocked by the obstruction.

Of course, there may also be a case where the vehicle information and the parking space identification are identified in the parking area block of the image at the same time. This situation is a small probability anomalous event, but in order to avoid misjudgment in this case, The parking space identification method may further include: when it is recognized that the vehicle information exists and the parking space identifier exists, determining that the current state of the parking space area is unoccupied.

As shown in FIG. 6, the method flow of steps S101 to S104 may be executed by the parking space identification device 200. The parking space identification device 200 mainly includes an identification unit 201, a first determination unit 202, a second determination unit 203, and a third determination unit 204. The identification unit 201 is configured to execute the above step S101, the first determination unit 202 is used to perform the above step S102, the second determination unit 203 is used to perform the above step S103, and the third determination unit 204 is used to execute the above step S104.

In one embodiment, the parking space identification may be a two-dimensional code identification, and / or a number, and / or a character, and / or a bar code identification, as long as it can be used to uniquely identify the parking space. The parking space identifiers set on different parking spaces must be different, so that each parking space has a unique parking space identifier, which will not cause the information to be confused with each other. By identifying the parking space identifier, the corresponding parking space can be uniquely determined.

For example, the parking space identification is a two-dimensional code identification, and the identification of the two-dimensional code identification can obtain the parking space number of the parking area. Of course, the information obtained by identifying the two-dimensional code can also include other required information, such as the location information of the parking space area (flat garage), or the location information of the parking area pre-allocated or allocated (mechanical garage) ); Or may also include fixed parking space information. A fixed parking space refers to a vehicle that can only park in a specific license plate or plates. Other vehicles are not allowed to park in. These license plate information can be set in the parking space sign. When other vehicles with a license plate are parked, a warning and corresponding Follow-up processing.

Referring to FIG. 3, the image shown includes a parking area block 10 and a parking area block 20. In order to make the description more concise, taking the parking area block 10 as an example, the parking space block 10 can be identified in the parking area block 10 . The parking space identification 101 is a result of identifying a physical mark provided on the parking space. The physical mark can be attached, painted, or connected through a connection structure to mark a position in the parking space that can be blocked by a vehicle parked on it. For example, at the middle position of the parking space, it is ensured that there is no parking space identifier in the image when the parking space is occupied by the vehicle.

In one embodiment, the vehicle information includes at least one of the following information: vehicle body characteristic information and license plate characteristic information. The vehicle information is information indicating a vehicle parked in the parking space area, as long as it can determine whether there is a vehicle in the parking space area. The vehicle information may be, for example, the characteristics of the license plate. After the license plate is identified in the parking area block of the image, it can be determined that the vehicle is parked in the parking space, and the license plate number on the license plate can be further identified to uniquely determine the corresponding vehicle.

Preferably, the vehicle information may include both vehicle body characteristic information and license plate characteristic information. In this way, when the vehicle license plate characteristic information is blocked by an obstruction, the vehicle body characteristic information may also be used to determine that there is indeed a vehicle on the parking area tile in the image. The volume is large, and it is usually not completely blocked, while the license plate features are small, and it is easy to be blocked by passing vehicles or people.

In one embodiment, the step S101 of identifying whether a parking space tile in the captured image has a parking space identifier includes:

The acquired image is input into a preset neural network to recognize the parking space identifier in the image by the neural network, and output the identified parking space identifier.

The neural network is already trained. By inputting the image into the neural network, the neural network can realize the identification of the parking space mark. The neural network may be integrated in the recognition unit 201 as a part of the parking space recognition device 200, or may be provided outside the parking space recognition device 200, and may be called by the recognition unit 201.

In this disclosure, the specific form of the neural network is not limited, and the neural network may be a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), or a Deep Neural Network (DNN) Wait.

Because there may or may not be a parking space identification in the image, the neural network outputs a specific parking space identification when the parking space identification is recognized, and may output no information or output information indicating that there is no parking space identification when the parking space identification is not recognized.

Taking the parking space identification as a two-dimensional code identification as an example, a neural network can be used to identify the two-dimensional identification code from the image, and then scan the code to analyze the information of the two-dimensional identification code, such as the parking space number. In order to improve the clarity of the two-dimensional code identification and increase the scanning accuracy, various situations need to be considered when training the neural network.

In an example, identifying whether a parking space area tile in the acquired image has a parking space identifier may include the following steps: inputting the acquired image into at least two preset neural networks corresponding to different imaging environments, so that each The neural network separately recognizes whether there is a parking space identifier in the parking area tile in the image and outputs the recognized parking space identifier; compares the parking space identifiers output by each of the neural networks, and selects the optimal result as the parking area tile of the image Parking space sign.

The object of comparison may be, for example, unlimited clarity, completeness, uniformity of brightness, etc. of the parking space mark, and the optimal result may correspondingly be the parking space mark with the highest definition and / or completeness and / or the most uniform brightness.

Because the environment of the garage is different, it may happen that it is too bright or too dark. Therefore, you can perform deep learning for different imaging environments to obtain neural networks suitable for different imaging environments, and make the collected images can be input to In these neural networks, the parking space identifiers of the parking area tiles in the image are identified by these neural networks, respectively. Next, the results are compared. Of course, assuming that the neural network does not recognize the parking space identification, it is not necessary to perform the comparison, but the result of directly identifying the parking space identification is not determined.

Since different neural networks are deep-learned in different environments, they are adapted to different imaging environments. Assuming that the images are collected in a dark environment, the images are input to the neural network obtained by deep learning in a dark environment. The results obtained are of higher quality, resulting in higher accuracy of the final parking space identification information.

The neural networks corresponding to these different imaging environments are deep learning neural networks trained through image training sets corresponding to different imaging environments. The image training set may include sample images with parking space identification and corresponding parking space identification.

Different imaging environments are, for example, a darker environment and a brighter environment. A darker environment may refer to an image environment with a brightness lower than a set minimum brightness threshold, and a brighter environment may refer to an image environment with a brightness higher than a set maximum brightness threshold. Of course, there can be more imaging environments, for example, it can be set to environmental conditions that change in a gradient manner.

In another example, identifying whether the parking area tile in the acquired image includes a parking space identifier may include the following steps: inputting the acquired image into a preset neural network to identify the neural network by the neural network; The parking space identifier in the image, and the identified parking space identifier is output. This preset neural network can be adapted to different imaging environments, that is, the collected image may have multiple conditions, such as an image that is too bright, an image that is too dark, etc., but after inputting this image into this preset neural network, Both can get a relatively good result.

Therefore, when training the neural network, various situations need to be considered. The image training set of the neural network may include multiple sample images of parking spaces collected at different locations in the garage at different times. In each sample image, the vehicle may or may not be parked in the parking space, and the parked vehicle may be blocked by a pedestrian or other vehicles. Similarly, in each sample image, the parking space may or may not have a parking space identification, and the parking space identification may be blocked. Similarly, in each sample image, the background brightness may be bright or dim. Of course, there can be more imaging environments, for example, it can be set to environmental conditions that change in a gradient manner.

In one embodiment, when the supervised learning method is used, the algorithm can also artificially correct the recognition errors and compare the results of the parking space identification in the neural network recognition image. The comparison object can be, for example, the clarity of the parking space identification, There is no limit to completeness, uniformity of brightness, etc., and the optimal result may be the parking space marking with the highest definition and / or the highest completeness and / or the most uniform brightness.

In step S101, the vehicle information in the parking area block of the collected image can be identified through image processing methods such as traditional license plate recognition and feature recognition, or can also be identified through a neural network.

Since the parking space of the mechanical garage can be moved up and down, an image acquisition device may correspond to multiple parking spaces. Therefore, a specific parking space area tile in the image collected by the acquisition device may correspond to multiple parking spaces. In order to improve the accuracy of the determination of the occupied state of the parking space, multiple frames of consecutively collected images can be used for determination. As shown in FIG. 4, in combination with the above FIG. 2, in one embodiment, the parking space recognition method further includes the following steps.

For the foregoing steps S101-S103, reference may be made to the related description in FIG. 2, and details are not described herein again. After the current state of the parking space area is determined according to the acquired image, the previous state of the parking space area may also be determined according to a previous frame relative to the acquired image. If the previous state is different from the current state, after determining that the current state of the parking space area is occupied in step S102, step S204 may be further performed to determine the state of the parking space; in step S103, the state of the parking space area is determined. After the current state is not occupied, step S205 may be further performed to determine the state of the parking space.

S204: When it is determined that the previous state of the parking space area is not occupied according to the previous frame image, the state of the parking space corresponding to the parking space identification identified from the previous frame image is updated to be occupied.

S205: When it is determined that the previous state of the parking space area is occupied according to the previous frame image, update the state of the parking space corresponding to the parking space identifier to unoccupied.

In this example, the status of each parking space in the mechanical garage is determined by combining several frames of the parking space area. According to steps S102 and S103, the current state of the parking space area can be obtained from the acquired images. At the same time, for the previous frame image of the parking space area relative to the acquired image, the previous state of the parking space area can be determined, and the current state of the parking space area is compared with the previous state. If the current state is the same as the previous state, the state is not updated. If the current state and the previous state are different, there are two cases of the state of the parking space area. In the first case, if the current status is unoccupied, the status of the parking space corresponding to the parking space identification is updated to unoccupied according to the parking space identification identified by the acquired image. In the second case, if the current status is occupied, the status of the parking space corresponding to the parking space identification identified from the previous frame image is updated to be occupied. Furthermore, the parking space can be determined based on the acquired image Corresponding vehicle information is associated with parking spaces.

Specifically, before the vehicle A enters a certain parking space on the first floor of the mechanical garage, the image acquisition device performs image acquisition on the empty parking space to obtain a first frame image (that is, a previous frame image). The image processing device can recognize that there is no vehicle information and a parking space identifier exists in the parking space area according to the first frame image. The previous state of the parking space area is unoccupied, and the parking space identification may be 101. at this time. When the vehicle enters the parking space with the parking space identification 101, the image acquisition device performs image acquisition on the parking space to obtain a second frame image (that is, the captured image). The image processing device can identify the parking space based on the second frame image. There is vehicle information in the area and no parking space identifier exists. The current status of the parking space area is occupied, and the vehicle information may be vehicle A. At this time, the image processing device associates the vehicle information A with the parking space identification 101 obtained from the first frame image, and updates the status of the parking space corresponding to the parking space identification 101 to be occupied.

After parking the vehicle, the vehicle A may rise to another floor with the panel. When the user picks up the vehicle, the vehicle A needs to be lowered to the first floor first. At this time, the image acquisition device can perform image acquisition on the parking space to obtain the first frame image (that is, the previous frame image). The image processing device can recognize that the parking area has vehicle information A and no parking space identifier according to the first frame image. At this time, the previous state of the parking area is occupied. After vehicle A drives out of the parking space, the image acquisition device performs image acquisition on the parking space to obtain a second frame image (that is, the captured image), and the image processing device can identify that there is no vehicle in the parking space area according to the second frame image Information and a parking space identification 101 exists, the current state of the parking space area is unoccupied. At this time, the image processing device may update the status of the parking space corresponding to the parking space identification 101 to be unoccupied.

It can be seen that the information of the previous frame is used to obtain the parking space identifier by combining the information of the previous frame. If the current state of the parking space area is occupied, the vehicle will put the parking space on the parking lot file, and it will not know which specific board the vehicle is occupying. At this time, there is a parking space in the previous frame image, so the previous frame is combined. Image, you can know the status of the parking space corresponding to the parking space mark in the previous image. By using at least two frames of images, the state of the parking space area can be determined to a specific parking space.

It should be noted that the time interval between the previous frame image and the acquired image described in this example can be set according to experience values, which is not limited in this application. In addition, the previous frame image and the captured image described in this example are just examples. In actual use, due to the specific speed of the vehicle entering and exiting the parking space, multiple frames of image may be collected to determine the status of the parking space. For example, the first frame of the collected image has a parking space identification. Due to the slow entry of the vehicle into the warehouse, the second frame of the image cannot identify the vehicle information (such as the vehicle number cannot be determined). At this time, the status of the parking space corresponding to the parking space identifier may be updated using the information of the first frame and the i-th frame image. The specific implementation manner is similar to the use of two consecutive frames of images, and details are not described herein again.

The method described in this example can more accurately identify each parking space in a mechanical garage, and provide a basis for subsequent parking induction and reverse car search.

In one embodiment, the parking space identification method may further include the following steps:

If the state of the parking space is occupied, query whether there is a binding relationship between the parking space and the vehicle occupying the parking space, and if the binding relationship does not exist, establish the binding relationship.

Because in the mechanical garage, vehicles enter and exit are on the first floor, the vehicle information can be identified when there is a vehicle in the parking space, and the parking space identification can be identified when there is no vehicle, so the parking space can be identified by the parking space identification and vehicle information. , When the vehicle leaves, the vehicle information and parking space identification can be identified successively. No matter the vehicle parks in or leaves, the parking space identification and vehicle information can be obtained according to at least two frames of the collected images to obtain the occupied status of the parking space. When the status of the parking space changes, the parking space identifier and vehicle information can be bound or unbound. The same goes for flat garages.

When a parking space is occupied by a vehicle, it means that the vehicle has been parked in the parking space. Check whether there is a binding relationship between the vehicle occupying the parking space and the parking space. If there is no binding relationship, establish a binding relationship between the vehicle and the parking space. , The identified parking space identifier and vehicle information can be bound, for example, the parking space identifier carried by the parking space identifier and the license plate information can be bound. If there is a binding relationship, it means that the parking space is occupied by the vehicle and is full. The binding relationship can be used to perform reverse car search processing, for example, the corresponding parking space label can be queried according to the license plate information, so that the specific location of the parking space can be located.

When the parking space is not occupied by the vehicle, it means that there is no parked vehicle in the parking space. You can query whether there is a binding relationship between the parking space and the vehicle. If there is a binding relationship, the binding relationship between the vehicle and the parking space is released, indicating that the vehicle is Has left the parking space. When the parking space does not have a binding relationship, it means that the parking space area is not occupied by the vehicle and is empty. Other vehicles can be induced to stop, such as providing vacancy information for a parking space, which can further guide the direction of the vacancy.

In order to make subsequent parking induction and reverse car searching easier, a binding relationship identifier may be set for each parking space identifier, and the relationship identifier may be updated during binding and unbinding.

For example, when parking guidance is performed, the actual number of parking spaces in a parking area minus the number of parking spaces identified by the binding relationship as a binding state can quickly obtain the remaining parking spaces in the area. For another example, when searching for a car in the reverse direction, the search can be performed only in the parking spaces whose binding relationship is identified as a binding state, thereby increasing the searching efficiency.

In one embodiment, referring to FIG. 5, a parking space identification device 200 may include:

A recognition unit 201 is configured to identify whether there is vehicle information and a parking space identifier in a parking area tile in an acquired image, where the parking area tile corresponds to a parking space area including a plurality of vertically overlapping parking spaces in the image. Block, the parking space identifier is used to identify the parking space;

A first determining unit 202, configured to determine that the state of the parking space area is occupied when it is recognized that the parking space area tile has vehicle information and the parking space identifier does not exist;

The second determining unit 203 is configured to determine that the state of the parking space area is unoccupied when it is recognized that the parking space area tile does not have vehicle information and the parking space identifier exists.

In one embodiment, referring to FIG. 6, the parking space identification device 200 further includes:

The third determining unit 204 is configured to determine that the state of the parking space area is unoccupied when it is recognized that the parking space area tile does not have vehicle information and the parking space identifier does not exist.

In one embodiment, the parking space identification is a two-dimensional code identification, and / or a number, and / or a character, and / or a barcode identification; the parking space identifications set on different parking spaces are different.

In one embodiment, the identification unit 201 further includes:

In one embodiment, a parking space identification device 200 further includes:

In one embodiment, the parking space identification device 200 further includes:

For details about the implementation process of the functions and functions of the units in the above device, refer to the implementation process of the corresponding steps in the foregoing method for details, and details are not described herein again.

As for the device embodiment, since it basically corresponds to the method embodiment, the relevant part may refer to the description of the method embodiment. The device embodiments described above are only schematic, and the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units.

With the device provided by the present invention, a parking space identifier is marked on each parking space, so it can be determined which parking space is when the parking space identifier is recognized. Once the occupancy of the parking space is identified and determined, the occupancy of the parking space can be learned and the subsequent management can be performed according to the parking space identification. Using this method, it is relatively easy to realize the occupation management of parking spaces, such as reverse car search, parking induction, etc.

The present invention also provides an electronic device including a processor and a memory; the memory stores a program that can be called by the processor; wherein when the processor executes the program, the program is implemented as in any one of the foregoing embodiments. The parking space identification method described above.

The embodiments of the parking space identification device of the present invention can be applied to electronic equipment. Taking software implementation as an example, as a device in a logical sense, it is formed by reading the corresponding computer program instructions in the non-volatile memory into the memory through the processor of the electronic device where it is located. In terms of hardware, as shown in FIG. 7, FIG. 7 is a hardware structure diagram of an electronic device in which the parking space identification device 200 is located according to an exemplary embodiment of the present invention, except for the processor 510 and the memory shown in FIG. 7. In addition to 530, network interface 520, bus 550, and non-volatile memory 540, the electronic device in which device 200 is located in the embodiment may generally include other hardware according to the actual function of the electronic device, and details are not described herein again.

The present invention also provides a machine-readable storage medium on which a program is stored, and when the program is executed by a processor, the image processing method according to any one of the foregoing embodiments is implemented.

The present invention may take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code therein. Machine-readable storage media includes permanent and non-permanent, removable and non-removable media, and information can be stored by any method or technology. Information may be computer-readable instructions, data structures, modules of a program, or other data. Examples of machine-readable storage media include, but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only Memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, read-only disc read-only memory (CD-ROM), digital versatile disc (DVD), or other optical storage , Magnetic tape cartridges, magnetic tape storage or other magnetic storage devices or any other non-transmission media can be used to store information that can be accessed by computing devices.

The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the present invention. Within the scope of protection.

Claims

A parking space identification method includes:

Identifying whether there is vehicle information and a parking space identifier in a parking area tile in the collected image, wherein the parking area tile refers to a tile in the image corresponding to a parking space area including a plurality of vertically overlapping parking spaces, the A parking space identifier is used to identify the parking space;

When it is identified that vehicle information exists in the parking space area tile and the parking space identifier does not exist, it is determined that the current status of the parking space area is occupied;

When it is identified that no vehicle information exists in the parking space area tile and the parking space identifier exists, it is determined that the current state of the parking space area is unoccupied.
The method for identifying a parking space according to claim 1, further comprising:

When it is identified that no vehicle information exists in the parking space area tile and the parking space identifier does not exist, it is determined that the current state of the parking space area is unoccupied.
The parking space identification method according to claim 1, wherein the parking space identification is a two-dimensional code identification, and / or a number, and / or a character, and / or a bar code identification; parking space identifications set on different parking spaces are different.
The method for identifying a parking space according to claim 1, wherein identifying whether the parking space area tile in the acquired image exists for the parking space identifier comprises:

The acquired image is input into a preset neural network, so that the neural network recognizes a parking space identifier in the image, and outputs the identified parking space identifier.
The method for identifying a parking space according to claim 1, further comprising:

Comparing the current state of the parking space area with the previous state of the parking space area, wherein the previous state of the parking space area is based on the previous frame for the parking space area relative to the captured image Determined by image;

If the current state is different from the previous state and the current state is unoccupied, updating the state of the parking space corresponding to the parking space identifier to unoccupied;

If the current state is different from the previous state and the current state is occupied, the state of the parking space corresponding to the parking space identifier identified from the previous frame image is updated to be occupied.
The method for identifying a parking space according to claim 5, further comprising:

If the state of the parking space is occupied, query whether there is a binding relationship between the parking space and a vehicle occupying the parking space, and if the binding relationship does not exist, establish the binding relationship;

If the state of the parking space is unoccupied, query whether there is a binding relationship between the parking space and the vehicle, and if the binding relationship exists, release the binding relationship.
A parking space identification device includes:

A recognition unit, configured to identify whether there is vehicle information and a parking space identifier in a parking area tile in the collected image, wherein the parking area tile refers to a parking space area in the image corresponding to a parking space area including a plurality of vertically overlapping parking spaces; A tile, the parking space identifier is used to identify the parking space;

A first determining unit, configured to determine that the current state of the parking space area is occupied when it is recognized that the parking space area tile has vehicle information and the parking space identifier does not exist;

A second determining unit is configured to determine that the current state of the parking space area is unoccupied when it is recognized that the parking space area tile does not have vehicle information and the parking space identifier exists.
The parking space identification device according to claim 7, wherein the device further comprises:

A third determining unit is configured to determine that the current state of the parking space area is unoccupied when it is identified that the parking area area tile does not have vehicle information and the parking space identifier does not exist.
The parking space identification device according to claim 7, wherein the parking space identification is a two-dimensional code identification, and / or a number, and / or a character, and / or a bar code identification; parking space identifications set on different parking spaces are different.
The parking space recognition device according to claim 7, wherein the recognition unit further comprises:

A neural network is configured to recognize a parking space identifier in the image based on the inputted captured image, and output the recognized parking space identifier.
The parking space identification device according to claim 7, further comprising:

A comparison unit is configured to compare the current state of the parking space area with a previous state of the parking space area, wherein the previous state of the parking space area is based on Determined by the previous frame of the image;

A first updating unit, configured to update the state of a parking space corresponding to the parking space identifier to be unoccupied if the current state is different from the previous state and the current state is unoccupied;

A second updating unit, configured to update the status of the parking space corresponding to the parking space identification identified from the previous frame image to, if the current status is different from the previous status and the current status is occupied; Occupied.
The parking space identification device according to claim 11, further comprising:

A binding unit is configured to query whether there is a binding relationship between the parking space and a vehicle occupying the parking space if the status of the parking space is occupied; if the binding relationship does not exist, establish the binding relationship Fixed relationship

The unbinding unit is configured to query whether there is a binding relationship between the parking space and the vehicle if the state of the parking space is not occupied, and release the binding relationship if the binding relationship exists.
An electronic device includes a processor and a memory; the memory stores a program that can be called by the processor; wherein when the processor executes the program, the device according to any one of claims 1-7 is implemented The parking space identification method described above.
A machine-readable storage medium stores a program thereon. When the program is executed by a processor, the method for identifying a parking space according to any one of claims 1-7 is implemented.