A parking space lamp identification and verification method and device, a computer device, and a storage medium

Abstract

The present application relates to a kind of parking space lamp identification verification method, device, computer equipment and storage medium, the method comprises: obtaining the sensing trigger data of all lighting terminals in target garage area;Based on the sensing trigger data, all lighting terminals are initially identified, to obtain initial parking space lamp identification result;Based on the sensing trigger data, the initial parking space lamp identification result is verified, to obtain final verification result;Based on the final verification result, the initial parking space lamp identification result is corrected, to obtain final parking space lamp identification result.The present application is through "data acquisition-initial identification-verification-correction" closed loop process, effectively solve the problem of parking space lamp misidentification in prior art, improve identification accuracy, guarantee garage lighting safety, while reducing the energy waste caused by invalid lighting, provide reliable support for the automation, energy-saving operation of intelligent garage lighting system.

Description

Technical Field

[0001] This invention relates to the field of intelligent garage lighting control technology, and in particular to a parking space light identification and verification method, device, computer equipment, and storage medium. Background Technology

[0002] In intelligent parking garage lighting systems, the accurate differentiation between parking space lights and driveway lights is crucial for achieving intelligent control and energy-saving operation. Existing technologies often suffer from limitations such as sensor cooling periods, data transmission loss, and the inability to effectively distinguish between pedestrian and vehicle activity. This frequently leads to driveway lights being misidentified as parking space lights, causing confusion in lighting control logic. This can result in energy waste or, more seriously, blind spots created by mistakenly turning off driveway lights, compromising driving safety. Currently, there is a lack of a method for efficiently identifying and verifying parking space lights, making it difficult to meet the precise classification requirements of intelligent parking garages. Therefore, there is an urgent need to optimize the identification scheme to address these technical pain points. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a parking light identification and verification method, device, computer equipment and storage medium.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: Firstly, this embodiment provides a parking space light recognition and verification method, including the following steps: Acquire the sensor trigger data of all lighting terminals within the target garage area; Based on the aforementioned sensor-triggered data, all lighting terminals are initially identified to obtain initial parking space light identification results; The initial parking space light recognition result is verified based on the sensor trigger data to obtain the final verification result; The initial parking light recognition result is corrected based on the final verification result to obtain the final parking light recognition result.

[0005] Secondly, this embodiment provides a parking light recognition and verification device, including: an acquisition unit, a recognition unit, a verification unit, and a correction unit; The acquisition unit is used to acquire the sensor trigger data of all lighting terminals within the target garage area; The identification unit is used to perform initial identification of all lighting terminals based on the induction trigger data to obtain initial parking space light identification results; The verification unit is used to verify the initial parking light recognition result based on the sensor trigger data to obtain the final verification result. The correction unit is used to correct the initial parking light recognition result based on the final verification result to obtain the final parking light recognition result.

[0006] Thirdly, this embodiment provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a parking light recognition and verification method as described above.

[0007] Fourthly, this embodiment provides a storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, can implement a parking light recognition and verification method as described above.

[0008] The advantages of this invention compared to existing technologies are as follows: Through a complete closed-loop process, using sensor-triggered data as the core support, it first quickly filters candidate parking space lights through initial identification, then eliminates misidentified objects through targeted verification, and finally outputs accurate results through correction. This effectively avoids the identification errors caused by the cooling period of the sensor terminal, data loss, and difficulties in distinguishing between people and vehicles in existing technologies, significantly improving the accuracy of parking space light identification and ensuring that the garage lighting system can accurately control the working status of both types of lights. This not only avoids the driving safety hazards caused by lane lights being accidentally turned off, but also reduces energy waste caused by ineffective parking space light operation, providing reliable technical support for the automated operation and energy-saving optimization of intelligent garage lighting systems, and adapting to the actual application needs of various garage scenarios.

[0009] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart illustrating a parking light recognition and verification method provided in an embodiment of the present invention. Figure 2 Provided for embodiments of the present invention Figure 1 A flowchart of the S3 process; Figure 3 Provided for embodiments of the present invention Figure 2 A flowchart of the S32 process; Figure 4 Provided for embodiments of the present invention Figure 2 A flowchart of the S34 process; Figure 5 Provided for embodiments of the present invention Figure 2 A flowchart of the S35 process; Figure 6 Provided for embodiments of the present invention Figure 4 A flowchart of S346 in China; Figure 7 This is a schematic block diagram of a parking space light recognition and verification device provided in an embodiment of the present invention; Figure 8 A schematic block diagram of a computer device provided for an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0014] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0015] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations of trajectories, and includes such combinations of trajectories.

[0016] Please see Figures 1 to 6 As shown, a parking space light recognition and verification method includes the following steps: S1. Obtain the sensor trigger data of all lighting terminals within the target garage area.

[0017] Specifically, this step involves acquiring the sensor trigger data of all lighting terminals within the target garage area over a preset statistical period (generally 7 to 31 days). In practical scenarios, the lighting terminals integrate wireless communication modules and microwave radar sensing chips, enabling them to detect pedestrian and vehicle activity within their sensing area in real time and trigger lighting operation and corresponding sensor trigger data. This trigger data includes at least key information such as the lighting terminal's device identifier, trigger timestamp, and trigger duration. The lighting terminals report the sensor trigger data to a data aggregation terminal in real time via their wireless communication modules. The data aggregation terminal preprocesses the received raw data, including deduplicating duplicate data and using adjacent time node interpolation to complete slightly missing data, ensuring data integrity before storage. This provides a reliable data foundation for subsequent initial identification and verification steps. The preprocessed sensor trigger data eliminates redundancy and missing data, providing a high-quality data foundation for subsequent initial identification, sorting, and verification steps, avoiding identification errors caused by data quality issues.

[0018] S2. Based on the induction trigger data, perform initial identification of all lighting terminals to obtain initial parking space light identification results.

[0019] Specifically, the initial identification aims to preliminarily determine which lighting terminals might be parking space lights through simple data filtering and analysis. This process provides a preliminary candidate list of parking space lights for subsequent verification. In practice, by analyzing the trigger frequency, trigger time interval, and other characteristics of each lighting terminal, a preliminary determination can be made of which lighting terminals are likely to be parking space lights, forming a preliminary candidate list of parking space lights as the initial parking space light identification result. For example, parking space lights typically have a high trigger frequency and a relatively short trigger time interval. By setting certain thresholds, lighting terminals that meet the criteria are filtered out as candidate parking space lights and added to the candidate list, forming the initial parking space light identification result. This preliminary filtering step reduces the computational load of subsequent verification and improves the efficiency of the entire identification process.

[0020] S3. Verify the initial parking space light recognition result based on the sensor trigger data to obtain the final verification result; Step S3 includes: S31. Extract all lighting terminals identified as candidate parking lights from the initial parking light identification results and use them as objects to be verified. In the initial parking light recognition results, only candidate parking lights are at risk of misidentification, while non-parking light lights have been excluded through feature screening. This step only verifies candidate parking lights, avoiding invalid verification of non-candidate parking lights, reducing computational load, shortening overall verification time, optimizing resource allocation, and improving verification efficiency.

[0021] S32. Sort all objects to be verified based on the sensor trigger data to obtain the sorting result; Specifically, the installation layout of parking space lights and lane lights is regular, and the induction triggering of adjacent lighting terminals is time-related (when vehicles pass adjacent lighting terminals continuously, the trigger signals will be generated in time sequence). By analyzing the triggering time correlation between the objects to be verified, their adjacent relationship can be determined, laying the foundation for the verification of subsequent adjacent objects.

[0022] Step S32 includes: S3201. Traverse all objects to be verified, and sequentially treat each object to be verified as the first object to be sorted, and the remaining objects to be verified as the second objects to be sorted, and perform the following processing: By traversing through pairs of trajectories, we ensure that all potential adjacency relationships between objects to be verified can be detected, thus avoiding the omission of key associations.

[0023] S3202. Based on the sensor trigger data, extract each trigger timestamp of the first object to be sorted as the first trigger timestamp, and extract each trigger timestamp of the second object to be sorted as the second trigger timestamp. Specifically, the trigger timestamp of each of the first objects to be sorted is extracted as the first trigger timestamp t1i (i=1,2,...,n, where n is the total number of triggers of the first objects to be sorted), and the trigger timestamp of each of the second objects to be sorted is extracted as the second trigger timestamp t2j (j=1,2,...,m, where m is the total number of triggers of the second objects to be sorted).

[0024] S3203. Traverse each first trigger timestamp and find all occurrences of second trigger timestamps within the maximum time length after the first trigger timestamp as associated trigger timestamps. Specifically, the maximum time duration Tmax is set based on the maximum reasonable movement time of a vehicle in the garage. Vehicle speeds in underground garages are typically slow (approximately 5-10 km / h, or 1.4-2.8 m / s), and the distance between adjacent lighting terminals is generally 10-20 meters. Therefore, the movement time between adjacent lighting terminals is approximately 5-15 seconds. In this embodiment, Tmax is set to 25 seconds based on this prediction duration, thus covering the maximum possible movement time and including a certain safety margin to handle scenarios such as vehicle deceleration, turning, or temporary stopping. Simultaneously, Tmax needs to be large enough to ensure the capture of relevant triggering events, but not so large as to introduce irrelevant noise (such as triggering by pedestrians lingering for extended periods). Iterate through each first trigger timestamp t1i, and search for all second trigger timestamps that satisfy t2j∈[t1i,t1i+Tmax] within Tmax after t1i. Use these second trigger timestamps as associated trigger timestamps. This step effectively filters out accidental trigger signals from non-adjacent lighting terminals, ensuring that the selected associated trigger timestamps are all valid associations generated by continuous vehicle movement, thus improving the accuracy of subsequent adjacency determination.

[0025] S3204. Calculate the interval value between each associated trigger timestamp and the corresponding first trigger timestamp, sort all the obtained interval values, and form an interval sequence. Specifically, the time difference between each associated trigger timestamp and its corresponding first trigger timestamp is calculated to obtain the interval value Lij = t2j - t1i. All interval values ​​are sorted in ascending order to form the interval sequence L = {Lij1, Lij2, ..., Lijk} (where k is the total number of associated trigger timestamps). The interval value reflects the time difference between the triggering of two lighting terminals and is the core data for analyzing the distribution pattern of triggering times. The sorted interval sequence facilitates subsequent construction of floating intervals and peak statistics.

[0026] S3205. Traverse each interval value in the interval sequence and construct a corresponding floating interval based on the interval value and a preset time floating value; Specifically, each interval value Lij in the interval sequence L is traversed, and a corresponding floating interval [Lij-ΔT, Lij+ΔT] is constructed based on Lij and ΔT (for example, when Lij=2 seconds, the floating interval is [1.5 seconds, 2.5 seconds]). Given the varying speeds of vehicles in the garage and the response delay of the sensing terminal, a preset time floating value ΔT (e.g., ΔT=0.5 seconds) is set to accommodate these errors, making the algorithm more robust when calculating time intervals and effectively balancing accuracy and fault tolerance.

[0027] S3206. Traverse all floating intervals, count the number of interval values ​​falling within each floating interval in the interval sequence, find the largest number of interval values ​​as the peak frequency, and record the corresponding interval value as the baseline interval value. Specifically, all floating intervals are traversed, and the number of interval values ​​Lnc falling within each floating interval in the interval sequence is counted. The largest number of interval values ​​Lnc is found as the peak frequency Lnc-max, and the interval value Lij corresponding to Lnc-max is recorded as the baseline interval value Lmax. The peak frequency reflects the high-frequency time pattern of the two lighting terminals being continuously triggered, and the baseline interval value represents the time the vehicle travels between the two lighting terminals. Together, they constitute the core quantitative indicator for determining the adjacency relationship.

[0028] S3207. Calculate the adjacent dynamic threshold based on the set of interval values ​​corresponding to all floating intervals; Specifically, in step S3207, the formula for calculating the adjacent dynamic threshold Th0 is as follows: Th0=Snc×Std(Lnc)+punish(Lnc-max), Where Snc is the confidence threshold, ranging from 2 to 5 (generally 3.5 is used). The higher the value, the higher the confidence requirement for determining the adjacency relationship; Std(Lnc) is the standard deviation of the number of interval values ​​corresponding to all floating intervals (i.e., the set L={Lij1,Lij2,...,Lijk}), reflecting the dispersion of the number of interval values; Lnc-max is the peak frequency corresponding to all floating intervals; punish is a penalty nonlinear function used to handle the case where the absolute value of Lnc-max is too small, defined as: punish(Lnc-max)=8, when Lnc-max<8; punish(Lnc-max)=16-Lnc-max, when 8≤Lnc-max<16; punish(Lnc-max)=0, when Lnc-max≥16.

[0029] This step introduces standard deviation to reflect the discrete characteristics of the data, confidence threshold to control the strictness of the judgment, and penalty function to correct the influence of extreme values. This makes the calculation of adjacent dynamic thresholds more in line with the distribution pattern of triggered data in actual scenarios, avoids misjudgment of adjacent relationships due to accidental peaks, and further improves the reliability of the sorting results.

[0030] S3208. Determine whether the reference interval value is less than the preset maximum moving time interval and obtain a first determination result; determine whether the peak frequency is greater than the adjacent dynamic threshold and obtain a second determination result. Specifically, the estimated maximum movement time interval Tneighbor represents the typical upper limit of time for a vehicle to move between adjacent lighting terminals. This value is calculated based on the actual garage layout and vehicle speed, and can be verified through on-site measurements or simulation data to ensure an accurate reflection of the physical distance relationship between adjacent lighting terminals. A smaller Tneighbor (e.g., 3 seconds) helps to eliminate false associations between non-adjacent lighting terminals. A baseline interval value less than the maximum movement time interval ensures that the triggered association conforms to the reasonable time logic of vehicle movement; a peak frequency greater than the adjacent dynamic threshold ensures that the association is statistically significant. This dual-judgment approach significantly improves the reliability of adjacent relationships. This step, through dual-condition screening, eliminates accidental associations and associations that do not conform to the logic of vehicle movement, ensuring the rigor and reliability of adjacent relationship determination.

[0031] S3209. Determine the adjacency relationship between the current first object to be sorted and the second object to be sorted based on the first judgment result and the second judgment result; In step S3209, when the first judgment result is that the baseline interval value is less than the maximum moving time interval, and the second judgment result is that the peak frequency is greater than the adjacent dynamic threshold, the adjacency relationship between the first object to be sorted and the second object to be sorted is as follows: the first object to be sorted is the preceding adjacent object of the second object to be sorted, the second object to be sorted is the following adjacent object of the first object to be sorted, and the estimated time distance between the first object to be sorted and the second object to be sorted is the baseline interval value.

[0032] Specifically, the temporal correlation between the two lighting terminals only meets both the "vehicle movement logic" and "statistical significance" when the first judgment result is that the baseline interval value is less than the maximum movement time interval, and the second judgment result is that the peak frequency is greater than the adjacent dynamic threshold, thus clearly defining their adjacent order and distance; otherwise, it is determined that there is no adjacent relationship. This step clarifies the adjacent relationship and quantified distance between the two lighting terminals, providing a clear spatial correlation basis for subsequent verification of sensing frequency difference and sensing trajectory inference.

[0033] S3210. Repeat the above traversal process until the adjacency relationship between any two objects to be verified is determined, and a sorting result is formed based on the adjacency relationship.

[0034] Specifically, during execution, the first and second objects to be sorted are replaced sequentially, and the processing steps S3201 to S3209 are repeated until the adjacency relationship between any two objects to be verified is determined. Based on the final adjacency relationship, all objects to be verified are linked together according to the logic of "preceding order - current order - subsequent order" to form a sorting result. This step ensures that all potential adjacency relationships between objects to be verified are identified by repeatedly traversing all pairwise trajectory combinations. Based on the complete adjacency relationship, an ordered sequence can be formed, providing a global basis for locating the preceding and subsequent adjacent objects for each object to be verified, and providing a reliable spatial association foundation for subsequent verification steps.

[0035] S33. Sequentially extract each object to be verified as the current verification object, and obtain the preceding and following adjacent objects of the current verification object based on the sorting result; Specifically, the sorting result clearly defines the adjacent order of all objects to be verified. Based on this result, the associated objects of each currently verified object can be directly located without additional calculations, simplifying the operation process. During execution, each object to be verified is treated as the current verification object. From the sorting result, the objects to be verified that precede and are directly adjacent to the current verification object are extracted as its pre-order adjacent objects; the objects to be verified that follow and are directly adjacent to the current verification object are extracted as its post-order adjacent objects. This step eliminates the need for additional association calculations, directly obtaining adjacent object information from the sorting result, simplifying the operation process, improving verification efficiency, and ensuring the accuracy of adjacent objects.

[0036] S34. Based on the preceding adjacent objects, perform a frequency difference verification on the current verification object to obtain a first verification result.

[0037] Specifically, the trigger frequency of a parking space light is reasonably correlated with the trigger frequency of its preceding adjacent object (likely a lane light or another parking space light). Since parking space lights are triggered only when a vehicle is parked, their trigger frequency is relatively low. If the trigger frequency of the currently verified object is significantly higher than that of its preceding adjacent object, it is more likely to be a lane light. This feature can be used to initially eliminate falsely identified objects.

[0038] Step S34 includes: S341. Detect whether the number of the preceding adjacent objects is 0; S342. When the number of preceding adjacent objects is 0, the first verification result is that the verification is passed. Specifically, when the number of preceding adjacent objects is 0, the current verification object has no preceding associated objects and no basis for determining frequency anomalies, so it passes the frequency verification by default. This step can quickly distinguish the applicable scenarios for sensing frequency difference verification, avoid meaningless calculations, and improve process efficiency.

[0039] S343. When the number of preceding adjacent objects is not 0, traverse all preceding adjacent objects and perform the following processing for each preceding adjacent object: S344. Based on the sensing trigger data, obtain the number of sensing triggers of the current verification object as a first calculation parameter, and obtain the number of sensing triggers of the preceding adjacent object as a second calculation parameter; Specifically, based on the preprocessed sensor trigger data, the number of sensor triggers (Triggers(Lparking)) of the current verification object Lparking within a preset statistical period is counted and used as the first calculation parameter C1; the number of sensor triggers (Triggers(Lprev-i)) of each preceding neighboring object Lprev-i within the preset statistical period is counted and used as the second calculation parameter C2. The first calculation parameter C1 and the second calculation parameter C2 reflect the trigger activity levels of the current verification object Lparking and the preceding neighboring object Lprev-i, respectively, and are the core data for frequency difference analysis.

[0040] S345. Based on the sorting result, obtain the estimated time distance between the preceding adjacent object and the current verification object as the third calculation parameter, and obtain the corresponding peak frequency as the fourth calculation parameter.

[0041] Specifically, from the sorting results formed in S32, the estimated time distance Lmax between the current traversed preceding neighbor and the current verification object is extracted and used as the third calculation parameter C3; the peak frequency Lnc-max corresponding to the preceding neighbor and the current verification object determined in S3206 is extracted and used as the fourth calculation parameter C4. The third calculation parameter C3 reflects the spatial correlation between the two lighting terminals, and the fourth calculation parameter C4 reflects the trigger correlation strength between the two lighting terminals. Both can assist in frequency difference analysis and enrich the verification dimensions.

[0042] S346. Calculate the first verification probability based on the first calculation parameter, the second calculation parameter, the third calculation parameter, and the fourth calculation parameter; Step S346 includes: S3461. Calculate the inductive natural logarithm based on the first calculation parameters; Specifically, the formula for calculating the natural logarithm is as follows: Ln(C1)=ln(C1) / ln(1000); This step normalizes the number of times the current verification object is triggered within the preset statistical period, unifies the numerical scale of different trigger counts, and avoids model deviation caused by differences in absolute counts.

[0043] S3462. Calculate the peak frequency ratio based on the first calculation parameter and the fourth calculation parameter; Specifically, the formula for calculating the peak frequency ratio is as follows: Rc = C4 / C1; Where Rc is the peak frequency ratio, C4 is the fourth calculation parameter, and C1 is the first calculation parameter. The ratio of the peak frequencies of the preceding adjacent objects to the current verification object reflects the strength of the triggering association between the two; a higher ratio indicates a more significant association.

[0044] S3463. Calculate the actual number ratio based on the first calculation parameter and the second calculation parameter; Specifically, the formula for calculating the proportion of actual occurrences is as follows: Rf = C1 / C2; Where Rf is the actual number of times ratio, C1 is the first calculation parameter, and C2 is the second calculation parameter. The frequency difference between the two is directly reflected by the ratio of the number of triggers of the preceding adjacent object and the current verification object. The closer the ratio is to 1, the better the frequency characteristics match.

[0045] S3464. Input the third calculation parameter, the inductive natural logarithm, the peak frequency ratio and the actual frequency ratio into the first neural network model with LSTM, and output the first verification probability.

[0046] Specifically, the third calculation parameter C3, the inductive natural logarithm Ln(C1), the peak frequency ratio Rc=C4 / C1, and the actual frequency ratio Rf=C1 / C2 corresponding to each preceding neighboring object Lprev-i are used as feature values. These four feature values ​​are fed into the first neural network model with LSTM for calculation in descending order of the peak frequency Lnc-max corresponding to the preceding neighboring object Lprev-i, thereby obtaining the first verification probability Th1. The first verification probability Th1 reflects the probability that the current verification object Lparking is a parking light. The first neural network model with LSTM is good at capturing the dependencies and patterns in time-series data. Through pre-training, it learns the essential differences between parking lights and lane lights in multi-dimensional features (such as parking lights usually having a lower frequency difference and stable spatial correlation). The input feature combination is transformed into the first verification probability Th1 in the 0-1 interval. This probability directly quantifies the possibility that the current verification object meets the characteristics of a parking light, providing a precise quantitative basis for subsequent threshold comparison and verification result determination.

[0047] S347. Compare the first verification probability with a preset first threshold and obtain a first comparison result; Specifically, a first threshold Snp is preset (Snp is generally 0.6 to 0.8, with 0.75 recommended. The higher the Snp, the more accurate the verification result, but it will reduce the recall rate of parking light recognition). The first comparison result is used as the judgment standard for the first verification result to ensure that the first verification result has consistency and interpretability in different scenarios.

[0048] S348. Confirm the first verification result based on the first comparison result.

[0049] Specifically, when the first comparison result is that the first verification probability Th1 ≥ the first threshold Snp, it means that the sensing frequency of the current verification object matches the frequency characteristics of the parking space light, and the first verification result is that the verification is successful; otherwise, the first verification result is that the verification fails.

[0050] S35. Based on the first verification result, the preceding adjacent object and the following adjacent object, perform sensing trajectory inference verification on the current verification object to obtain the second verification result; Step S35 includes: S3501. Check whether the first verification result is a successful verification; S3502. If the first verification result is a verification failure, then the second verification result is a verification failure. When the verification fails, it means that the current verification object no longer meets the frequency characteristics of the parking light. There is no need to perform subsequent trajectory verification. The verification is directly judged as failed, which can reduce invalid calculations.

[0051] S3503. If the first verification result is successful, then perform the following steps: The first successful verification indicates that the current verification object matches the frequency characteristics of the parking space light. Further verification through trajectory features is required to ensure the accuracy of the recognition results.

[0052] S3504. Determine whether the number of the preceding adjacent objects and the number of the following adjacent objects of the current verification object are both not 0; This step is a prerequisite for checking trajectory analysis, ensuring sufficient trajectory data to support trajectory statistics. It verifies whether the current verification object has both preceding and subsequent adjacent lighting terminals to avoid invalid trajectory analysis due to incomplete data.

[0053] S3505. If the number of preceding adjacent objects or the number of subsequent adjacent objects is 0, then the second verification result is that the verification is passed. Specifically, the purpose of this step is to pass trajectory verification by default when a complete trajectory cannot be constructed, thereby preventing misjudgments due to incomplete data and improving the fault tolerance of the algorithm.

[0054] S3506. If the number of preceding adjacent objects and the number of following adjacent objects are both not 0, then traverse all trajectory combinations of preceding and following adjacent objects, and count the number of complete trajectories with the current verification object as the intermediate node for each trajectory combination. Specifically, this step captures real vehicle movement patterns through comprehensive statistical analysis. For all possible trajectory combinations, the number of consecutive trigger sequences with the current verification object as the intermediate node is counted, and the frequency of the current verification object as a channel node is quantified, providing data support for judgment.

[0055] The method for counting the number of complete trajectories is as follows: After finding the trigger timestamp of a preceding adjacent object Lprev-i, if the current verification object Lparking also has a trigger timestamp within the maximum movement time interval Tneighbor, and immediately after the trigger timestamp of Lparking, if a subsequent adjacent object Lnext-j also has a trigger timestamp within the maximum movement time interval Tneighbor, then it is considered as one trajectory, and the number of complete trajectories is incremented by one.

[0056] S3507. Trajectory combinations with a complete trajectory count greater than the preset trajectory count are considered valid trajectory combinations. Specifically, when the number of complete trajectory occurrences for a trajectory combination is low, it means that these trajectories may be randomly generated. A preset trajectory occurrence count is used to filter out randomly generated trajectories, ensuring that valid trajectory combinations are statistically significant and avoiding misjudgments caused by random trajectories. This step eliminates random noise interference and ensures the reliability of trajectory analysis by filtering out trajectory combinations with a complete trajectory occurrence count greater than the preset count and discarding those with a count less than the preset count. Understandably, the preset trajectory occurrence count can be set to different values ​​according to the actual scenario to make the parking light recognition and verification method more applicable.

[0057] S3508. Detect whether the number of valid trajectory combinations is 0; S3509. If the number of valid trajectory combinations is 0, then the second verification result is that the verification is passed. When the number of valid trajectory combinations is 0, it means that the current verification object does not have frequently occurring continuous trajectories, which meets the characteristics of the parking light as a "parking target", and no further calculation is required.

[0058] S3510. If the number of valid trajectory combinations is not 0, then for each valid trajectory combination, the trajectory feature value is calculated. Specifically, the trajectory feature values ​​include: the logarithm of the preceding number of times ln(Triggers(Lprev-i) / Triggers(Lparking)), the logarithm of the following number of times ln(Triggers(Lnext-j) / Triggers(Lparking)), the logarithm of the midpoint number of times ln(N / Triggers(Lparking)), and the logarithm of the preceding and following number of times.

[0059] Where Triggers(Lprev-i) is the number of times the preceding neighboring object Lprev-i is triggered, Triggers(Lparking) is the number of times the current verification object Lparking is triggered, Triggers(Lnext-j) is the number of times the following neighboring object Lnext-j is triggered, and N is the number of complete trajectory iterations.

[0060] S3511. Input the trajectory feature values ​​of all the effective trajectory combinations into the second neural network model with LSTM, and output the second verification probability; Trajectory feature values ​​(such as the logarithm of the ratio of preceding / following times, the logarithm of the ratio of midpoint times, etc.) quantify the trajectory association patterns between the current verification object and its preceding and following adjacent objects. The second neural network model with LSTM excels at capturing dependencies in time-series data. By inputting the trajectory feature values ​​from all the effective trajectory combinations into the second neural network model with LSTM in ascending order of the complete trajectory count N corresponding to the effective trajectory combination, and outputting the second verification probability as input to the pre-trained second neural network model with LSTM, the essential difference between the "discontinuous trajectory" of parking light and the "continuous trajectory" of lane light can be learned, transforming the feature patterns into a second verification probability, thus achieving quantitative judgment at the trajectory level. This step utilizes the model's nonlinear fitting capability to overcome the limitations of traditional threshold judgment and improve the accuracy of trajectory verification.

[0061] S3512. Compare the second verification probability with a preset second threshold to obtain a second comparison result; Specifically, a second threshold Strace is preset (Strace is generally 0.7 to 0.9, usually 0.8. The higher the Strace, the more accurate the verification result, but it will reduce the recall rate of parking light recognition). The second comparison result is used as the judgment standard for the second verification result to ensure that the second verification result has consistency and interpretability in different scenarios.

[0062] S3513. Confirm the second verification result based on the second comparison result.

[0063] Specifically, when the second comparison result is that the second verification probability Th2 ≥ the second threshold Strace, it means that the trajectory data of the current verification object matches the trajectory data characteristics of the parking light, and the second verification result is that the verification is successful; otherwise, the second verification result is that the verification fails.

[0064] S36. Combining the first verification result and the second verification result, the final verification result of the current verification object is obtained.

[0065] Specifically, the final verification result of the current verification object is a successful verification if and only if both the first verification result and the second verification result are successful, i.e., the current verification object is a parking space light; otherwise, the final verification result of the current verification object is a failed verification, i.e., the current verification object is a lane light.

[0066] S4. Based on the final verification result, the initial parking light recognition result is corrected to obtain the final parking light recognition result.

[0067] Specifically, if the final verification result of the current verification object is a pass, the recognition result of the current verification object as a parking light remains unchanged. If the final verification result of the current verification object is a failure, the current verification object is corrected to a lane light and removed from the parking light candidate list. After each candidate parking light in the initial parking light recognition result has been corrected, the final parking light candidate list is the final parking light recognition result.

[0068] The present invention provides a parking space light recognition and verification method, which completes the initial screening by collecting sensor trigger data, and then performs multi-level verification logic of "sorting of objects to be verified - verification of sensor frequency difference - verification of sensor trajectory inference" to achieve accurate recognition of parking space lights, avoid misjudgment of lane lights, and at the same time take into account the safety, energy saving and automation of garage lighting.

[0069] Please see Figure 7 As shown, the present invention also discloses a parking space light recognition and verification device, comprising: an acquisition unit 10, a recognition unit 20, a verification unit 30, and a correction unit 40; The acquisition unit 10 is used to acquire the sensor trigger data of all lighting terminals in the target garage area; The identification unit 20 is used to perform initial identification of all lighting terminals based on the induction trigger data to obtain initial parking space light identification results; The verification unit 30 is used to verify the initial parking space light recognition result based on the sensing trigger data to obtain the final verification result; The correction unit 40 is used to correct the initial parking light recognition result based on the final verification result to obtain the final parking light recognition result.

[0070] Furthermore, the verification unit 30 is specifically used for: Extract all lighting terminals identified as candidate parking lights from the initial parking light identification results and use them as objects to be verified; Based on the sensor trigger data, all objects to be verified are sorted to obtain the sorting result; Each object to be verified is extracted sequentially as the current verification object, and the preceding and following adjacent objects of the current verification object are obtained based on the sorting result. Based on the preceding adjacent objects, the current verification object is subjected to frequency difference verification to obtain the first verification result; Based on the first verification result, the preceding adjacent object, and the following adjacent object, the sensing trajectory inference verification is performed on the current verification object to obtain the second verification result; By combining the first verification result and the second verification result, the final verification result of the current verification object is obtained.

[0071] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned parking light recognition and verification device and its various units can be referred to the corresponding descriptions in the foregoing method embodiments. For the sake of convenience and brevity, these details will not be repeated here.

[0072] The aforementioned parking space light recognition and verification device can be implemented as a computer program, which can, for example... Figure 8 It runs on the computer device shown.

[0073] Please see Figure 8 , Figure 8 This is a schematic block diagram of a computer device 500 provided in an embodiment of this application; the computer device 500 can be a terminal or a server, wherein the terminal can be an electronic device with communication functions such as a smartphone, tablet computer, laptop computer, desktop computer, personal digital assistant, and wearable device. The server can be a standalone server or a server cluster composed of multiple servers.

[0074] See Figure 8 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0075] The non-volatile storage medium 503 can store a command system 5031 and a computer program 5032. The computer program 5032 includes program instructions that, when executed, cause the processor 502 to perform a parking light recognition and verification method.

[0076] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0077] The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a parking light recognition and verification method.

[0078] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or may combine certain components in a specific pattern, or may have different component arrangements.

[0079] The processor 502 is used to run a computer program 5032 stored in the memory to perform the following steps: Acquire the sensor trigger data of all lighting terminals within the target garage area; Based on the aforementioned sensor-triggered data, all lighting terminals are initially identified to obtain initial parking space light identification results; The initial parking space light recognition result is verified based on the sensor trigger data to obtain the final verification result; The initial parking light recognition result is corrected based on the final verification result to obtain the final parking light recognition result.

[0080] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0081] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0082] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions, which, when executed by a processor, can implement the above-described parking light recognition and verification method. The storage medium stores a computer program, which includes program instructions, which, when executed by a processor, can implement the above-described method. The program instructions include the following steps: Acquire the sensor trigger data of all lighting terminals within the target garage area; Based on the aforementioned sensor-triggered data, all lighting terminals are initially identified to obtain initial parking space light identification results; The initial parking space light recognition result is verified based on the sensor trigger data to obtain the final verification result; The initial parking light recognition result is corrected based on the final verification result to obtain the final parking light recognition result.

[0083] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0084] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0085] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and other division methods may be used in implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or skipped.

[0086] The steps in the method of this invention can be adjusted, merged, or deleted in order as needed. The units in the device of this invention can be merged, divided, or deleted as needed. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0087] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0088] The above embodiments are preferred implementations of the present invention. In addition, the present invention can be implemented in other ways. Any obvious substitutions without departing from the concept of the present technical solution are within the protection scope of the present invention.

Claims

1. A method for identifying and verifying parking space lights, characterized in that, Includes the following steps: Acquire the sensor trigger data of all lighting terminals within the target garage area; Based on the aforementioned sensor-triggered data, all lighting terminals are initially identified to obtain initial parking space light identification results; The initial parking space light recognition result is verified based on the sensor trigger data to obtain the final verification result; The initial parking light recognition result is corrected based on the final verification result to obtain the final parking light recognition result; The step of verifying the initial parking light recognition result based on the sensed trigger data to obtain the final verification result includes: Extract all lighting terminals identified as candidate parking lights from the initial parking light identification results and use them as objects to be verified; Based on the sensor trigger data, all objects to be verified are sorted to obtain the sorting result; Each object to be verified is extracted sequentially as the current verification object, and the preceding and following adjacent objects of the current verification object are obtained based on the sorting result. Based on the preceding adjacent objects, the current verification object is subjected to frequency difference verification to obtain the first verification result; Based on the first verification result, the preceding adjacent object, and the following adjacent object, the sensing trajectory inference verification is performed on the current verification object to obtain the second verification result; By combining the first verification result and the second verification result, the final verification result of the current verification object is obtained.

2. The parking space light recognition and verification method according to claim 1, characterized in that, The step of sorting all objects to be verified based on the sensor trigger data to obtain the sorting result includes: Iterate through all objects to be verified, treating each object as the first object to be sorted, and then treat the remaining objects as the second objects to be sorted, performing the following processing: Based on the sensor trigger data, each trigger node of the first object to be sorted is extracted as the first trigger node, and each trigger node of the second object to be sorted is extracted as the second trigger node; Iterate through each first trigger node and find all the second trigger nodes that appear within the maximum time length after the first trigger node as associated trigger nodes. Calculate the interval value between each associated trigger node and its corresponding first trigger node, sort all the obtained interval values, and form an interval sequence; Iterate through each interval value in the interval sequence and construct a corresponding floating interval based on the interval value and a preset time floating value; Traverse all floating intervals, count the number of interval values ​​falling within each floating interval in the interval sequence, find the largest number of interval values ​​as the peak frequency, and record the corresponding interval value as the baseline interval value. Calculate the adjacent dynamic threshold based on the set of interval values ​​corresponding to all floating intervals; Determine whether the reference interval value is less than the preset maximum movement time interval and obtain a first determination result; determine whether the peak frequency is greater than the adjacent dynamic threshold and obtain a second determination result. Based on the first judgment result and the second judgment result, the adjacency relationship between the current first object to be sorted and the second object to be sorted is determined; Repeat the above traversal process until the adjacency relationship between any two objects to be verified is determined, and a sorting result is formed based on the adjacency relationship.

3. The parking space light recognition and verification method according to claim 2, characterized in that, In the step of determining the adjacency relationship between the first object to be sorted and the second object to be sorted based on the first judgment result and the second judgment result... When the first judgment result is that the baseline interval value is less than the maximum moving time interval, and the second judgment result is that the peak frequency is greater than the adjacent dynamic threshold, the adjacency relationship between the first object to be sorted and the second object to be sorted is as follows: the first object to be sorted is the preceding adjacent object of the second object to be sorted, the second object to be sorted is the following adjacent object of the first object to be sorted, and the estimated time distance between the first object to be sorted and the second object to be sorted is the baseline interval value.

4. The parking space light recognition and verification method according to claim 3, characterized in that, The step of performing frequency difference verification on the current verification object based on the preceding adjacent objects to obtain a first verification result includes: Detect whether the number of the preceding adjacent objects is 0; When the number of preceding adjacent objects is 0, the first verification result is that the verification is passed; If the number of preceding neighbor objects is not 0, then traverse all preceding neighbor objects and perform the following processing for each preceding neighbor object: Based on the sensor trigger data, the number of sensor triggers of the current verification object is obtained as a first calculation parameter, and the number of sensor triggers of the preceding adjacent object is obtained as a second calculation parameter. Based on the sorting results, the estimated time distance between the preceding adjacent object and the current verification object is obtained as the third calculation parameter, and the corresponding peak frequency is obtained as the fourth calculation parameter. The first verification probability is calculated based on the first calculation parameter, the second calculation parameter, the third calculation parameter, and the fourth calculation parameter; The first verification probability is compared with a preset first threshold, and a first comparison result is obtained; The first verification result is confirmed based on the first comparison result.

5. The parking space light recognition and verification method according to claim 4, characterized in that, The step of calculating the first verification probability based on the first calculation parameter, the second calculation parameter, the third calculation parameter, and the fourth calculation parameter includes: The natural logarithm of induction is calculated based on the first calculation parameters; The peak frequency ratio is calculated based on the first calculation parameter and the fourth calculation parameter. The actual number of times ratio is calculated based on the first calculation parameter and the second calculation parameter; The third calculation parameter, the inductive natural logarithm, the peak frequency ratio, and the actual frequency ratio are input into the first neural network model with LSTM, and the first verification probability is output.

6. The parking space light recognition and verification method according to claim 2, characterized in that, The step of performing a sensing trajectory inference verification on the current verification object based on the first verification result, the preceding adjacent object, and the following adjacent object to obtain a second verification result includes: Check whether the first verification result is a successful verification; If the first verification result is a verification failure, then the second verification result is a verification failure. If the first verification result is successful, then perform the following steps: Determine whether the number of the preceding adjacent objects and the number of the following adjacent objects of the current verification object are both non-zero; If the number of preceding adjacent objects or the number of following adjacent objects is 0, then the second verification result is that the verification is passed; If the number of preceding adjacent objects and the number of following adjacent objects are both not 0, then traverse all trajectory combinations of preceding and following adjacent objects, and count the number of complete trajectories with the current verification object as the intermediate node for each trajectory combination. Trajectory combinations with a complete trajectory count greater than the preset trajectory count are considered valid trajectory combinations. Check whether the number of valid trajectory combinations is 0; If the number of valid trajectory combinations is 0, then the second verification result is that the verification is passed; If the number of valid trajectory combinations is not 0, then for each valid trajectory combination, the trajectory feature value is calculated; The trajectory feature values ​​of all the effective trajectory combinations are input into a second neural network model with LSTM, and the second verification probability is output. The second verification probability is compared with a preset second threshold to obtain a second comparison result; The second verification result is confirmed based on the second comparison result.

7. A parking space light recognition and verification device, characterized in that, include: Acquisition unit, identification unit, verification unit and correction unit; The acquisition unit is used to acquire the sensor trigger data of all lighting terminals within the target garage area; The identification unit is used to perform initial identification of all lighting terminals based on the induction trigger data to obtain initial parking space light identification results; The verification unit is used to verify the initial parking light recognition result based on the sensor trigger data to obtain the final verification result. The correction unit is used to correct the initial parking light recognition result based on the final verification result to obtain the final parking light recognition result; The verification unit is specifically used for: Extract all lighting terminals identified as candidate parking lights from the initial parking light identification results and use them as objects to be verified; Based on the sensor trigger data, all objects to be verified are sorted to obtain the sorting result; Each object to be verified is extracted sequentially as the current verification object, and the preceding and following adjacent objects of the current verification object are obtained based on the sorting result. Based on the preceding adjacent objects, the current verification object is subjected to frequency difference verification to obtain the first verification result; Based on the first verification result, the preceding adjacent object, and the following adjacent object, the sensing trajectory inference verification is performed on the current verification object to obtain the second verification result; By combining the first verification result and the second verification result, the final verification result of the current verification object is obtained.

8. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a parking light recognition and verification method as described in any one of claims 1-6.

9. A storage medium, characterized in that, The storage medium stores a computer program, which includes program instructions that, when executed by a processor, implement a parking light recognition and verification method as described in any one of claims 1-6.

Images