A method and system for generating three-dimensional holographic intersection based on laser point cloud technology

Abstract

The present invention provides a method and system for generating three-dimensional holographic intersections based on laser point cloud technology, and relates to the technical field of intelligent traffic management. The method for generating three-dimensional holographic intersections comprises the following steps: establishing a three-dimensional intersection model corresponding to the overall structure of the intersection; obtaining the position information of a mobile target in the overall structure of the intersection and the scanning data corresponding to the mobile target; judging whether the scanning data exists in each adjacent intersection, if so, controlling the mobile target history model corresponding to the adjacent intersection to be used as the mobile target model of the current intersection, and if not, controlling the mobile target model of the mobile target to be created based on the scanning data at the current intersection; importing the mobile target model into the corresponding position in the three-dimensional intersection model to synthesize and obtain the three-dimensional model of the holographic intersection. The present invention can effectively reduce the operating load during the three-dimensional synthesis of the holographic intersection, and can accurately realize the analysis and prediction of the turning situation of the intersection.

Description

Technical Field

[0001] The present invention relates to the field of intelligent traffic management technology, and in particular to a method and system for generating three-dimensional holographic intersections based on laser point cloud technology. Background Art

[0002] Laser point cloud is a collection of scan points. The laser radar system scans the ground to obtain the three-dimensional coordinates of the ground reflection point. Each ground reflection point is distributed in the three-dimensional space in the form of a point according to the three-dimensional coordinates, which is called a scan point. Laser point cloud technology can also be used to generate a three-dimensional holographic intersection model to display the traffic conditions of the road intersection in an all-round way and conduct timely and convenient intersection traffic management.

[0003] When existing mobile targets such as four-wheeled vehicles, three-wheeled vehicles, two-wheeled vehicles and pedestrians are traveling at road intersections, it is difficult to accurately know the corresponding positions of the mobile targets at each traffic intersection, and it is difficult to accurately understand the vehicle queues and possible turns at the intersection in the same lane only through roadside traffic flow monitoring means, resulting in relatively low commuting efficiency for vehicles and pedestrians due to inaccurate control of traffic light waiting time. Summary of the invention

[0004] In view of the shortcomings of the prior art mentioned above, the purpose of the present invention is to provide a three-dimensional generation method and system for holographic intersections based on laser point cloud technology, which is used to solve the problem that it is difficult to accurately obtain the corresponding position of a moving target at each traffic intersection in the prior art, and it is difficult to accurately understand the vehicle queuing situation and possible turning situations at the intersection of the same lane in the same direction only through roadside traffic flow monitoring means, thereby resulting in relatively low commuting efficiency of vehicles and pedestrians due to inaccurate control of traffic light waiting time.

[0005] To achieve the above-mentioned and other related purposes, the present invention provides a method for generating three-dimensional holographic intersections based on laser point cloud technology, comprising the following steps:

[0006] Establish a three-dimensional intersection model corresponding to the overall structure of the intersection;

[0007] Obtaining location information of a mobile target within the overall structure of the intersection and scanning data corresponding to the mobile target;

[0008] Determine whether the scan data exists at each adjacent intersection, if yes, control the mobile target history model corresponding to the adjacent intersection as the mobile target model of the current intersection, if no, control the mobile target model of the mobile target based on the scan data at the current intersection;

[0009] The mobile target model is imported into the corresponding position of the three-dimensional intersection model to synthesize the holographic intersection three-dimensional model, and the behavioral characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed to obtain the path prediction data of each mobile target and upload it to the traffic light control system.

[0010] In one embodiment of the present invention, a three-dimensional intersection model corresponding to the overall structure of the intersection is established, including: regularly scanning the overall framework of the intersection, comparing data with the three-dimensional intersection model of the previous cycle, and adjusting the changed coordinate areas to update the adjusted three-dimensional intersection model.

[0011] In one embodiment of the present invention, obtaining the position information of a mobile target in the overall structure of the intersection and the scanning data corresponding to the mobile target includes: obtaining the control signal of a mobile target i newly appearing in the overall structure of the intersection, obtaining the position data (x i ,y i ), control the position data (x i ,y i ) to perform laser scanning to obtain a laser point cloud data set, and process the laser point cloud data set to obtain scanning data.

[0012] In one embodiment of the present invention, the laser point cloud data set is processed to obtain scanning data, including: searching for a feature data set reflecting the unique features of the moving target i in the laser point cloud data set, processing the feature data set according to an algorithm corresponding to the unique features, and obtaining feature data of each unique feature as scanning data S i ; Among them, the unique features include: maximum height h max , maximum width w max and a specific area reflection profile L.

[0013] In one embodiment of the present invention, before determining whether the scan data exists at each adjacent intersection, it also includes: determining the accurate classification of the current moving target based on the interval range of each unique feature, wherein the accurate classification includes: people, four-wheeled vehicles and two-wheeled vehicles.

[0014] In one embodiment of the present invention, it is determined whether the scanned data exists at each adjacent intersection. If so, the historical model of the mobile target at the corresponding adjacent intersection is controlled to be used as the mobile target model at the current intersection, including: according to the newly appeared position data (x i ,y i ), determine the position F of the mobile target i at the intersection in the three-dimensional intersection model x ∈[F 1 , F 2 , …, F n ], where n is the number of branches at the current intersection, and the approaching direction F is retrievedx The scan data set [S 1 , S 2 , S 3 , …, S j ], analyze the scan data S i Is there a scanned dataset [S 1 , S 2 , S 3 , …, S j ], if the scan data set of the previous intersection [S 1 , S 2 , S 3 , …, S j ] contains the scan data S corresponding to the moving target a a With scan data S i If the same, the control calls the scan data S a The corresponding historical model M La The moving target model M as the moving target i Yi .

[0015] In one embodiment of the present invention, it is determined whether the scanned data exists at each adjacent intersection. If not, the mobile target model of the mobile target is created based on the scanned data at the current intersection, including: according to the newly appeared position data (x i ,y i ), determine the position F of the mobile target i at the intersection in the three-dimensional intersection model x ∈[F 1 , F 2 , …, F n ], where n is the number of branches at the current intersection, and the approaching direction F is retrieved x The scan data set [S 1 , S 2 , S 3 , …, S j ], analyze the scan data S i Is there a scanned dataset [S 1 , S 2 , S 3 , …, S j ], if the scan data set of the previous intersection [S 1 , S 2 , S 3 , …, S j ] does not exist in the scan data S i The same corresponding data, then control according to the feature data set and scanning data S of the mobile target i i Draw the moving target model M of moving target iYi .

[0016] In one embodiment of the present invention, the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed to obtain the path prediction data of each mobile target, including:

[0017] The moving target is a four-wheeled vehicle in the same direction motor vehicle lane. C When the four-wheeled vehicle i C Is it in the solid line area of ​​the intersection? If so, the four-wheeled vehicle i is placed in the solid line area of ​​the intersection according to the distribution of the turning lanes corresponding to the solid line area of ​​the intersection. C Classified as left-turning vehicles i CL 、Straight-moving vehicles CS and right turning vehicles CR , and get with the left-turning vehicle i CL The corresponding actual left turn amount N CL , and straight-moving vehicles i CS The corresponding actual straight distance N CS and right turning vehicles CR The corresponding actual right turn amount N CR If not, then control and count the four-wheeled vehicles outside the solid line area of ​​the intersection i C Total N Z1 , according to the left turn prediction factor δ for four-wheel vehicle steering prediction CL , straight line prediction factor δ CS and the right turn prediction factor δ CR , calculate the predicted left turn amount N CL =N Z1 *δ CL , predicted straight travel volume N CS =N Z1 *δ CS and predicted right turn amount N CR =N Z1 *δ CR , and get the four-wheeled vehicle i C The total number of left turn predictions N ZL1 =N CL +N Z1 *δ CL 、Total straight line forecast N ZS1 =N CS +N Z1 *δ CS and the total right turn prediction N ZR1 =N CR +N Z1 *δ CR , where turning lanes include left-turn and U-turn lanes, through lanes and right-turn lanes;

[0018] The moving target is a non-motorized target i in the non-motorized vehicle lane in the same direction.F When each non-maneuverable target i F Whether it exists in the signal light waiting area, if so, the non-maneuverable target i is placed in the signal light waiting area according to the distribution of the turning areas corresponding to the signal light waiting area. F Divided into left turn target i ML 、Straight target i MS and right turn target i MR , and get the left turn target i ML The corresponding actual left turn target N ML , and the straight target i MS The corresponding actual straight target amount N MS and right turn target i MR The corresponding actual right turn target N MR If not, then control the non-mobile target i outside the signal light waiting area F Total N Z2 , according to the target left turn prediction factor δ for the non-maneuvering target turning prediction ML , target straight line prediction factor δ MS and the target right turn prediction factor δ MR , calculate the predicted left turn target value N ML =N Z2 *δ ML , predicted straight target quantity N MS =N Z2 *δ MS and predicted right turn target N MR =N Z2 *δ MR , and get the non-maneuvering target i F The total number of left turn prediction targets N ZL2 =N ML +N Z2 *δ ML , the total amount of straight line forecast target N ZS2 =N MS +N Z2 *δ MS and the total right turn prediction target N ZR2 =N MR +N Z2 *δ MR , where the turning area includes a left-turn area, a right-turn area and a straight-ahead area.

[0019] In one embodiment of the present invention, the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed to obtain the path prediction data of each mobile target, and the following further includes: obtaining the four-wheeled vehicle i in the same direction motor vehicle lane in the previous cycle C The actual total number of left turns N ZLS1 、Actual straight total N ZSS1and the actual total right turn N ZRS1 , and get the left turn prediction factor Straight Predictors and right turn predictors Get the non-motorized target i in the same direction non-motorized lane in the previous cycle F The actual total left turn target N ZLS2 , Actual total straight travel target N ZSS2 and the actual total right turn target N ZRS2 , get the target left turn prediction factor Target Straight Predictor and target right turn predictor

[0020] The present invention also provides a 3D generation system for holographic intersections based on laser point cloud technology, comprising:

[0021] An intersection modeling unit, the intersection modeling unit establishes a three-dimensional intersection model corresponding to the overall structure of the intersection;

[0022] A data acquisition unit, the data acquisition unit acquires the position information of the mobile target in the overall structure of the intersection and the scanning data corresponding to the mobile target;

[0023] A judgment query unit, wherein the judgment query unit judges whether the scan data exists at each adjacent intersection, and if so, controls the mobile target history model corresponding to the adjacent intersection to be used as the mobile target model of the current intersection; if not, controls the mobile target model of the mobile target to be created at the current intersection based on the scan data; and

[0024] The synthesis and analysis unit imports the mobile target model into the corresponding position in the three-dimensional intersection model to synthesize the holographic intersection three-dimensional model, analyzes the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model, obtains the path prediction data of each mobile target, and uploads it to the traffic light control system.

[0025] As described above, a method and system for generating three-dimensional holographic intersections based on laser point cloud technology of the present invention has the following beneficial effects: by pre-constructing a three-dimensional intersection model of an overall structure that is not easy to change, the modeling operation amount of the three-dimensional model of the holographic intersection can be preliminarily and effectively reduced. Then, according to the location information of the mobile target in the intersection area and the scanning data of the mobile target, it is further determined whether the mobile target has appeared in each adjacent intersection corresponding to the intersection. If it has appeared, the historical model of the mobile target corresponding to the previous adjacent intersection can be directly used as the mobile target model of the current intersection, and if it does not exist, the mobile target model is constructed again. Thereby, the construction amount of the mobile target model of the mobile target can be effectively reduced, thereby reducing the load of the system to construct the mobile target model and improving the generation efficiency of the three-dimensional model of the holographic intersection. Moreover, after the mobile target model and the three-dimensional intersection model are synthesized into a three-dimensional model of the holographic intersection, the distribution of the mobile target model in the three-dimensional intersection model can be used to achieve a reasonable prediction of the next turning path of the mobile target. This allows the waiting time of traffic lights to be further regulated based on the prediction results, thereby improving the rationality and accuracy of regulating the waiting time of vehicles and pedestrians at intersections when commuting. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 The figure is a flow chart of the three-dimensional generation method of the present invention.

[0027] Figure 2 Shown is a schematic diagram of the intersection layout of a preferred embodiment of the present invention.

[0028] Figure 3 This is a three-dimensional architecture diagram of the present invention. DETAILED DESCRIPTION

[0029] See also Figure 1 The present invention provides a method for generating three-dimensional holographic intersections based on laser point cloud technology, comprising the following steps:

[0030] Establish a three-dimensional intersection model corresponding to the overall structure of the intersection;

[0031] Obtaining location information of a mobile target within the overall structure of the intersection and scanning data corresponding to the mobile target;

[0032] Determine whether the scan data exists at each adjacent intersection, if yes, control the mobile target history model corresponding to the adjacent intersection as the mobile target model of the current intersection, if no, control the mobile target model of the mobile target based on the scan data at the current intersection;

[0033] The mobile target model is imported into the corresponding position of the three-dimensional intersection model to synthesize the holographic intersection three-dimensional model, and the behavioral characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed to obtain the path prediction data of each mobile target and upload it to the traffic light control system.

[0034] It is not difficult to find from the above content that in the process of constructing a three-dimensional holographic intersection according to the laser point cloud technology, in order to simplify the complexity of constructing the model, a three-dimensional intersection model of the overall structure of the intersection is pre-established. In other words, according to the distribution of the overall structure that is not easy to change at the intersection, the immovable overall structure is first built, and then the corresponding three-dimensional intersection model is obtained. For example, the pre-construction of the three-dimensional intersection model can be directly completed by laser point cloud technology. Before the model is constructed for the movable target appearing at the intersection, that is, when the mobile target appears in the overall structure area of ​​the intersection within the preset range, the location information of the corresponding mobile target appearance area will be controlled to be obtained, and the scanning data of the mobile target will be obtained through laser point cloud processing. The overall characteristics of the mobile target can be obtained by scanning, and this overall feature can be understood as a feature that can determine the individuality or uniqueness of the mobile target. After obtaining the scanning data of the mobile target, it is further determined whether the scanning data appears in the previous adjacent intersection with the opposite moving direction. If it exists, the historical model of the mobile target in the previous adjacent intersection can be directly used as the mobile target model of the current intersection, thereby saving the time required to reconstruct the mobile target model using, for example, laser point cloud technology, and reducing the system operation load. If the mobile target is a newly appeared target, for example, when there are residential areas, offices, etc. near the entire road where the mobile target moves toward the intersection, the historical model of the mobile target may not be left at other intersections. Therefore, a mobile target model corresponding to the mobile target can be newly created based on the data set corresponding to the scanning data and the laser point cloud. After determining the mobile target model, the mobile target model is imported into the corresponding position in the three-dimensional intersection model according to the corresponding position information obtained, thereby obtaining the entire holographic intersection three-dimensional model. In addition, after the holographic intersection three-dimensional model is generated, the behavioral characteristics of the mobile target in each waiting area of ​​the intersection will be further analyzed after the three-dimensional intersection model and the mobile target model are combined. The turning path composed of all mobile targets is obtained by analysis and real-time prediction, and is sent to the signal light control system after the prediction is completed. After obtaining relevant prediction data in the background, you can choose to redesign the traffic lights and adjust the waiting time at the intersection through intelligent or manual control.

[0035] Among them, establishing a three-dimensional intersection model corresponding to the overall structure of the intersection includes: regularly scanning the overall framework of the intersection, and comparing the data with the three-dimensional intersection model of the previous cycle, and adjusting the changed coordinate area to update the adjusted three-dimensional intersection model.

[0036] In one embodiment of the present invention, in the process of establishing a three-dimensional intersection model, in order to ensure that the overall structure of the intersection does not differ greatly from the actual situation, it is also necessary to set a preset period to scan the overall structure of the intersection. After scanning, the data is compared with the previous three-dimensional intersection model of the previous cycle to determine the changed area, and the changed area is updated to obtain an updated three-dimensional intersection model.

[0037] Further, obtaining the position information of the mobile target in the overall structure of the intersection and the scanning data corresponding to the mobile target includes: obtaining the control signal of the mobile target i newly appearing in the overall structure of the intersection, obtaining the position data (x i ,y i ), control the position data (x i ,y i ) to perform laser scanning to obtain a laser point cloud data set, and process the laser point cloud data set to obtain scanning data.

[0038] In one embodiment of the present invention, when determining the location information and scanning data of a mobile target, when a mobile target i enters the corresponding area of ​​the overall structure of the intersection, a control signal of the newly appeared mobile target i entering the corresponding area of ​​the intersection is scanned. And the location data (x i ,y i After obtaining the position data, further based on the position data (x i ,y i ), the laser radar scans the mobile target i to obtain a laser point cloud data set. Since the laser point cloud data set contains a large amount of data, the simplified scan data can be obtained by filtering, converting, calculating, etc. the laser point cloud data set, which can be used as the personalized or unique features of the mobile target. This can realize the analysis query of the situation where the mobile target i may appear at the previous intersection.

[0039] Specifically, the laser point cloud data set is processed to obtain the scanning data, including: searching for a feature data set reflecting the unique features of the mobile target i in the laser point cloud data set, processing the feature data set according to an algorithm corresponding to the unique features, and obtaining feature data of each unique feature as the scanning data S i ; Among them, the unique features include: maximum height h max , maximum width w max and a specific area reflection profile L.

[0040] In one embodiment of the present invention, in the process of obtaining scanning data, when the mobile target i enters the area of ​​the overall structure of the intersection, a laser point cloud data set can be obtained by laser scanning. By searching the laser point cloud data set for a feature data set that can reflect the mobile target i. For example, a coordinate area reflecting the height (including the highest point and the lowest point of the boundary along the height direction), a coordinate area reflecting the width (including two points of the boundary along the width direction), and a feature data set of a contour area corresponding to a specified purpose or target (such as a face contour, a license plate contour, etc.). After obtaining the highest point and the lowest point of the boundary along the height direction, further calculate the distance between the highest point and the lowest point of the boundary to obtain the maximum height h max After obtaining the two boundary points along the width direction, we further calculate the maximum width w based on the two boundary points. max Similarly, after obtaining the characteristic data set of the contour area, the reflection contour L of the specific area can be obtained according to the laser reflection situation.

[0041] Preferably, before determining whether the scanned data exists at each adjacent intersection, the method further includes: determining the accurate classification of the current moving target according to the interval range of each unique feature, wherein the accurate classification includes: people, four-wheeled vehicles and two-wheeled vehicles. In this embodiment, after obtaining the scanned data, according to the maximum height h of the four-wheeled vehicle, for example max , maximum width w max If both are greater than a person and a two-wheeled vehicle, it can be accurately determined that it is a four-wheeled vehicle. max , maximum width w max There are also large differences, so it is also possible to more accurately determine whether the moving target is a person or a two-wheeled vehicle. And after determining that it is a four-wheeled vehicle or a two-wheeled vehicle, by scanning the license plate outline of the vehicle, the unique information corresponding to the four-wheeled vehicle or the two-wheeled vehicle can be obtained. When it is confirmed to be a pedestrian, by scanning the face, different pedestrian individuals can be accurately determined. In this way, the segmentation of people, four-wheeled vehicles and two-wheeled vehicles is accurately completed. It is worth noting that four-wheeled vehicles correspond to driving in motor vehicle lanes, and pedestrians and two-wheeled vehicles correspond to driving in non-motor vehicle lanes.

[0042] Further, it is determined whether the scanned data exists at each adjacent intersection. If so, the historical model of the mobile target at the corresponding adjacent intersection is controlled to be the mobile target model at the current intersection, including: according to the newly appeared position data (x i ,y i ), determine the position F of the mobile target i at the intersection in the three-dimensional intersection model x ∈[F 1 , F 2 , …, F n], where n is the number of branches at the current intersection, and the approaching direction F is retrieved x The scan data set [S 1 , S 2 , S 3 , …, S j ], analyze the scan data S i Is there a scanned dataset [S 1 , S 2 , S 3 , …, S j ], if the scan data set of the previous intersection [S 1 , S 2 , S 3 , …, S j ] contains the scan data S corresponding to the moving target a a With scan data S i If the same, the control calls the scan data S a The corresponding historical model M La The moving target model M as the moving target i Yi .

[0043] In one embodiment of the present invention, when analyzing and judging the moving target model of each adjacent intersection based on the scanning data, it is necessary to first analyze and judge the moving target model based on the position data (x i ,y i ), determine the corresponding side of the mobile target i relative to the intersection. For example, when the intersection is a four-way intersection and the corresponding lanes include the east, south, west, and north directions, the mobile target i can come towards the intersection from the east, south, west, and north. If it comes from the east, determine the orientation F of the mobile target i at the intersection in the three-dimensional intersection model x Then, the direction F is further determined. x The corresponding scan data set [S 1 , S 2 , S 3 , …, S j ]. Then determine the scan data S of the current intersection i Is there a scanned dataset [S 1 , S 2 , S 3 , …, S j ]. If it exists, it can be directly based on the corresponding scan data set [S 1 , S 2 , S 3 , …, S j ] in the scan data S a , query its corresponding historical model M La, and the historical model M La Directly use the moving target model M of the moving target i at the current intersection Yi . Using the historical model M corresponding to the previous intersection La Directly use the moving target model M of the moving target i at the current intersection Yi , which can effectively reduce the amount of calculation in the system modeling process and reduce the system load.

[0044] Further, it is determined whether the scanned data exists at each adjacent intersection. If not, the mobile target model of the mobile target is created based on the scanned data at the current intersection, including: according to the newly appeared position data (x i ,y i ), determine the position F of the mobile target i at the intersection in the three-dimensional intersection model x ∈[F 1 , F 2 , …, F n ], where n is the number of branches at the current intersection, and the approaching direction F is retrieved x The scan data set [S 1 , S 2 , S 3 , …, S j ], analyze the scan data S i Is there a scanned dataset [S 1 , S 2 , S 3 , …, S j ], if the scan data set of the previous intersection [S 1 , S 2 , S 3 , …, S j ] does not exist in the scan data S i The same corresponding data, then control according to the feature data set and scanning data S of the mobile target i i Draw the moving target model M of moving target i Yi .

[0045] In one embodiment of the present invention, when analyzing and judging the moving target model of each adjacent intersection based on the scanning data, it is necessary to first analyze and judge the moving target model based on the position data (x i ,y i ), determine which side of the intersection the moving target i is on. For example, when the intersection is a four-way intersection and includes lanes in the east, south, west, and north directions, the moving target i can approach the intersection from the east, south, west, or north. If it approaches from the east, determine the orientation F of the intersection of the moving target i in the three-dimensional intersection model. x Then, the direction F is further determined.x The corresponding scan data set [S 1 , S 2 , S 3 , …, S j ]. Then determine the scan data S of the current intersection i Is there a scanned dataset [S 1 , S 2 , S 3 , …, S j ]. If it does not exist, it means that the scan data Therefore, there may be a situation where the mobile target i has just started driving from home or office. i The moving target model M for moving target i Yi Perform new modeling process.

[0046] Specifically, the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed to obtain the path prediction data of each mobile target, including:

[0047] The moving target is a four-wheeled vehicle in the same direction motor vehicle lane. C When the four-wheeled vehicle i C Is it in the solid line area of ​​the intersection? If so, the four-wheeled vehicle i is placed in the solid line area of ​​the intersection according to the distribution of the turning lanes corresponding to the solid line area of ​​the intersection. C Classified as left-turning vehicles i CL 、Straight-moving vehicles CS and right turning vehicles CR , and get with the left-turning vehicle i CL The corresponding actual left turn amount N CL , and straight-moving vehicles i CS The corresponding actual straight distance N CS and right turning vehicles CR The corresponding actual right turn amount N CR If not, then control and count the four-wheeled vehicles outside the solid line area of ​​the intersection i C Total N Z1 , according to the left turn prediction factor δ for four-wheel vehicle steering prediction CL , straight line prediction factor δ CS and the right turn prediction factor δ CR , calculate the predicted left turn amount N CL =N Z1 *δ CL , predicted straight travel volume N CS =N Z1 *δ CS and predicted right turn amount N CR =NZ1 *δ CR , and get the four-wheeled vehicle i C The total number of left turn predictions N ZL1 =N CL +N Z1 *δ CL 、Total straight line forecast N ZS1 =N CS +N Z1 *δ CS and the total right turn prediction N ZR1 =N CR +N Z1 *δ CR , where turning lanes include left-turn and U-turn lanes, through lanes and right-turn lanes;

[0048] The moving target is a non-motorized target i in the non-motorized vehicle lane in the same direction. F When each non-maneuverable target i F Whether it exists in the signal light waiting area, if so, the non-maneuverable target i is placed in the signal light waiting area according to the distribution of the turning areas corresponding to the signal light waiting area. F Divided into left turn target i ML 、Straight target i MS and right turn target i MR , and get the left turn target i ML The corresponding actual left turn target N ML , and the straight target i MS The corresponding actual straight target amount N MS and right turn target i MR The corresponding actual right turn target N MR If not, then control the non-mobile target i outside the signal light waiting area F Total N Z2 , according to the target left turn prediction factor δ for the non-maneuvering target turning prediction ML , target straight line prediction factor δ MS and the target right turn prediction factor δ MR , calculate the predicted left turn target value N ML =N Z2 *δ ML , predicted straight target quantity N MS =N Z2 *δ MS and predicted right turn target N MR =N Z2 *δ MR , and get the non-maneuvering target i F The total number of left turn prediction targets N ZL2 =N ML +N Z2 *δ ML, the total amount of straight line forecast target N ZS2 =N MS +N Z2 *δ MS and the total right turn prediction target N ZR2 =N MR +N Z2 *δ MR , where the turning area includes a left-turn area, a right-turn area and a straight-ahead area.

[0049] In one embodiment of the present invention, when analyzing the behavior characteristics of each moving target model to predict the path of the moving target, there is a moving target that is a four-wheeled vehicle traveling in the same direction of the motor vehicle lane. C and the non-motorized target i traveling in the non-motorized lane in the same direction F Among them, the non-maneuvering target i F Including pedestrians and two-wheeled vehicles. C When conducting the analysis, two parts of the analysis are required, namely, the situation inside the turning lane corresponding to the solid line area of ​​the intersection and the situation outside the solid line area of ​​the intersection. C When the vehicle is in the turning lane corresponding to the solid line area of ​​the intersection, further analysis shows that the vehicle i CL 、Straight-moving vehicles CS and right turning vehicles CR Thus, the actual straight-through volume N corresponding to each lane can be further obtained by statistics. CS and right turning vehicles CR The corresponding actual right turn amount N CR . And four-wheeled vehicles i C If the vehicle is not in the turning lane corresponding to the solid line area of ​​the intersection, it is necessary to count the number of four-wheeled vehicles outside the solid line area of ​​the intersection. C Total N Z1 . And according to the four-wheeled vehicles outside the solid line area of ​​the intersection C Corresponding to the possible situations of turning left, going straight or turning right, the left turn prediction factor δ is preset CL , straight line prediction factor δ CS and the right turn prediction factor δ CR Then we get the predicted left turn amount N CL =N Z1 *δ CL , predicted straight travel volume N CS =N Z1 *δ CS and predicted right turn amount N CR =N Z1 *δ CR Combined with the four-wheeled vehicles outside the solid line area of ​​the intersection C Total N Z1 , further predict the total number of left turns NZL1 =N CL +N Z1 *δ CL 、Total straight line forecast N ZS1 =N CS +N Z1 *δ CS and the total right turn prediction N ZR1 =N CR +N Z1 *δ CR . In the case of non-maneuvering targets F When conducting the analysis, two parts are also analyzed. That is, it includes the turning area corresponding to the signal light waiting area and the area outside the signal light waiting area. F When the vehicle is within the turning area corresponding to the signal light waiting area, further analysis can be performed to obtain the left turn target i ML 、Straight target i MS and right turn target i MR Then, through statistics, we can get the left turn target i ML The corresponding actual left turn target N ML , and the straight target i MS The corresponding actual straight target amount N MS and right turn target i MR The corresponding actual right turn target N MR If it is not a mobile target, F If the target is not in the signal light waiting area, it is necessary to first count the non-motorized targets outside the signal light waiting area. F Total N Z2 Then according to the preset target left turn prediction factor δ ML , target straight line prediction factor δ MS and the target right turn prediction factor δ MR , and further calculate the predicted left turn target value N ML =N Z2 *δ ML , predicted straight target quantity N MS =N Z2 *δ MS and predicted right turn target N MR =N Z2 *δ MR . Then according to the non-motorized target i outside the signal light waiting area F Total N Z2 , predict the non-maneuvering target i F The total number of left turn prediction targets N ZL2 =N ML +N Z2 *δ ML , the total amount of straight line forecast target N ZS2 =N MS+N Z2 *δ MS and the total right turn prediction target N ZR2 =N MR +N Z2 *δ MR By further utilizing the above data in the traffic light management system, it is possible to understand the distribution of moving targets and turning requirements at the intersection in real time and accurately, thereby completing the analysis and regulation of the waiting time of traffic lights in an intelligent or manual manner.

[0050] Preferably, the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed to obtain the path prediction data of each mobile target, and the following further includes: obtaining the four-wheeled vehicle i in the same-direction motor vehicle lane in the previous cycle C The actual total number of left turns N ZLS1 、Actual straight total N ZSS1 and the actual total right turn N ZRS1 , and get the left turn prediction factor Straight Predictors and right turn predictors Get the non-motorized target i in the same direction non-motorized lane in the previous cycle F The actual total left turn target N ZLS2 , Actual total straight travel target N ZSS2 and the actual total right turn target N ZRS2 , get the target left turn prediction factor Target Straight Predictor and target right turn predictor

[0051] In one embodiment of the present invention, in order to ensure the accuracy and adjustability of the path prediction, after the path prediction of the mobile target is completed, the four-wheeled vehicle i C The actual total number of left turns N ZLS1 、Actual straight total N ZSS1 and the actual total right turn N ZRS1 Monitor and count to get the left turn prediction factor Straight Predictors and right turn predictors In order to achieve corresponding real-time adjustment of each corresponding prediction factor at the same time point of the next day or in the next traffic light cycle. F The actual total left turn target N ZLS2 , Actual total straight travel target N ZSS2 and the actual total right turn target N ZRS2 , get the target left turn prediction factor Target Straight Predictor and target right turn predictor This enables corresponding real-time adjustment of each corresponding prediction factor at the same time point of the next day or in the next traffic light cycle, for example.

[0052] See also Figure 2 In a preferred embodiment, the current intersection includes lanes in four directions. There are motor vehicle lanes and pedestrian lanes in the direction toward the intersection. Motor vehicle lanes are used for four-wheeled vehicles, and pedestrian lanes are used for pedestrians and two-wheeled vehicles, or small three-wheeled or four-wheeled vehicles. When a four-wheeled vehicle is traveling on the motor vehicle lane, it can turn left or make a U-turn, turn right, or go straight. The icon corresponding to turning left or making a U-turn is "←", the icon corresponding to turning right is "→", and the icon corresponding to going straight is "↓", which are respectively turning in three directions. There are "A", "B", and "C" on the pedestrian lane. Among them, "A" represents turning left, "B" represents going straight, and "C" represents turning right, which are respectively turning in three directions.

[0053] like Figure 3 As shown, the present invention also provides a 3D generation system for holographic intersections based on laser point cloud technology, comprising:

[0054] An intersection modeling unit, the intersection modeling unit establishes a three-dimensional intersection model corresponding to the overall structure of the intersection;

[0055] A data acquisition unit, the data acquisition unit acquires the position information of the mobile target in the overall structure of the intersection and the scanning data corresponding to the mobile target;

[0056] A judgment query unit, wherein the judgment query unit judges whether the scan data exists at each adjacent intersection, and if so, controls the mobile target history model corresponding to the adjacent intersection to be used as the mobile target model of the current intersection; if not, controls the mobile target model of the mobile target to be created at the current intersection based on the scan data; and

[0057] The synthesis and analysis unit imports the mobile target model into the corresponding position in the three-dimensional intersection model to synthesize the holographic intersection three-dimensional model, analyzes the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model, obtains the path prediction data of each mobile target, and uploads it to the traffic light control system.

[0058] In one embodiment of the present invention, when a holographic three-dimensional intersection is generated using laser point cloud technology, a three-dimensional intersection model of the overall structure of the intersection is pre-established through an intersection modeling unit. The data acquisition unit is used to obtain the position information of the mobile target and the scanning data of the mobile target when the mobile target enters the current intersection. The judgment query unit is used to determine whether the scanning data exists in the corresponding previous adjacent intersection in the opposite direction of the mobile target. If the scanning data exists in the previous adjacent intersection, the historical model of the mobile target in the previous adjacent intersection will be directly used as the mobile target model of the current intersection. If the previous adjacent intersection does not exist, the mobile target model is constructed based on the data set related to the scanning data and the laser point cloud. Finally, the mobile target model and the three-dimensional intersection model are fused through the synthesis analysis unit to generate a holographic intersection three-dimensional model. And in order to realize the analysis of the current state of the holographic intersection three-dimensional model, the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model are analyzed, so as to predict the possible turning path of the mobile target. And by uploading it to the traffic light control system, the waiting time of the traffic light can be analyzed and adjusted.

[0059] In summary, the present invention can preliminarily and effectively reduce the modeling operation amount of the holographic intersection three-dimensional model by pre-constructing a three-dimensional intersection model with an overall structure that is not easy to change. Then, according to the location information of the mobile target in the intersection area and the scanning data of the mobile target, it is further determined whether the mobile target has appeared in each adjacent intersection corresponding to the intersection. If it has appeared, the historical model of the mobile target corresponding to the previous adjacent intersection can be directly used as the mobile target model of the current intersection, and if it does not exist, the mobile target model is constructed again. Thereby, the construction amount of the mobile target model of the mobile target can be effectively reduced, thereby reducing the load of the system to construct the mobile target model and improving the generation efficiency of the holographic intersection three-dimensional model. Moreover, after the mobile target model and the three-dimensional intersection model are synthesized into a holographic intersection three-dimensional model, the distribution of the mobile target model in the three-dimensional intersection model can be used to reasonably predict the next turning path of the mobile target. In order to further adjust the waiting time of the signal light according to the prediction results, the rationality and accuracy of the waiting time adjustment of vehicles and pedestrians at the intersection when commuting at the intersection are improved. Therefore, the present invention effectively overcomes various shortcomings in the prior art and has high industrial utilization value.

[0060] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A three-dimensional generation method for holographic intersections based on laser point cloud technology, characterized in that: The steps include: Establish a three-dimensional intersection model corresponding to the overall structure of the intersection; Acquire location information of a mobile target within the overall structure of the intersection and scan data corresponding to the mobile target; Determine whether the scanned data exists at each adjacent intersection, if so, control the mobile target history model corresponding to the adjacent intersection as the mobile target model of the current intersection, if not, control the mobile target model of the mobile target based on the scanned data at the current intersection; Importing the mobile target model into the corresponding position of the three-dimensional intersection model to synthesize the three-dimensional model of the holographic intersection, analyzing the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the three-dimensional model of the holographic intersection, obtaining the path prediction data of each mobile target, and uploading it to the traffic light control system; Analyzing the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model to obtain path prediction data of each mobile target, including: The moving target is a four-wheeled vehicle in the same direction motor vehicle lane. C When the four-wheeled vehicle i C Whether it exists in the solid line area of ​​the intersection, if so, the four-wheeled vehicle i is placed in the solid line area of ​​the intersection according to the distribution of the turning lanes corresponding to the solid line area of ​​the intersection. C Classified as left-turning vehicles i CL 、Straight-moving vehicles CS and right turning vehicles CR , and get with the left-turning vehicle i CL The corresponding actual left turn amount N CL , and straight-moving vehicles i CS The corresponding actual straight distance N CS and right turning vehicles CR The corresponding actual right turn amount N CR If not, then control and count the four-wheeled vehicles outside the solid line area of ​​the intersection i C Total N Z1 , according to the left turn prediction factor δ for four-wheel vehicle steering prediction CL , straight line prediction factor δ CS and the right turn prediction factor δ CR , calculate the predicted left turn amount N CL =N Z1 *δ CL , predicted straight travel volume N CS =N Z1 *δ CS and predicted right turn amount N CR =N Z1 *δ CR , and obtain the four-wheel vehicle i C The total number of left turn predictions N ZL1 =N CL +N Z1 *δ CL 、Total straight line forecast N ZS1 =N CS +N Z1 *δ CS and the total right turn prediction N ZR1 =N CR +N Z1 *δ CR , where turning lanes include left-turn and U-turn lanes, through lanes and right-turn lanes; The moving target is a non-motorized target i in the non-motorized vehicle lane in the same direction. F When each non-maneuverable target i F Whether it exists in the signal light waiting area, if so, the non-maneuverable target i is placed in the signal light waiting area according to the distribution of the turning areas corresponding to the signal light waiting area. F Divided into left turn target i ML 、Straight target i MS and right turn target i MR , and get the left turn target i ML The corresponding actual left turn target N ML , and the straight target i MS The corresponding actual straight target amount N MS and right turn target i MR The corresponding actual right turn target N MR If not, then control the non-mobile target i outside the signal light waiting area F Total N Z2 , according to the target left turn prediction factor δ for the non-maneuvering target turning prediction ML , target straight line prediction factor δ MS and the target right turn prediction factor δ MR , calculate the predicted left turn target value N ML =N Z2 *δ ML , predicted straight target quantity N MS =N Z2 *δ MS and predicted right turn target N MR =N Z2 *δ MR , and obtain the non-maneuverable target i F The total number of left turn prediction targets N ZL2 =N ML +N Z2 *δ ML , the total amount of straight line forecast target N ZS2 =N MS +N Z2 *δ MS and the total right turn prediction target N ZR2 =N MR +N Z2 *δ MR , where the turning area includes a left-turn area, a right-turn area and a straight-ahead area.

2. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 1 is characterized in that: Establish a 3D intersection model corresponding to the overall structure of the intersection, including: The overall framework of the intersection is scanned regularly, and data is compared with the three-dimensional intersection model of the previous cycle, and the changed coordinate areas are adjusted to update the adjusted three-dimensional intersection model.

3. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 1, characterized in that: Acquiring the position information of a mobile target within the overall structure of the intersection and the scanning data corresponding to the mobile target, including: Obtain the control signal of the newly appeared mobile target i in the overall structure of the intersection, and obtain the position data (x i ,y i ), controls the position data (x i ,y i ) to perform laser scanning to obtain a laser point cloud data set, and process the laser point cloud data set to obtain the scanning data.

4. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 3 is characterized in that: Processing the laser point cloud data set to obtain the scanning data includes: Find a feature data set reflecting the unique features of the mobile target i in the laser point cloud data set, process the feature data set according to an algorithm corresponding to the unique features, and obtain feature data of each unique feature as the scanning data S i ; Among them, the unique features include: maximum height h max , maximum width w max and a specific area reflection profile L.

5. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 4 is characterized in that: Before determining whether the scanned data exists at each adjacent intersection, the method further includes: According to the interval range of each of the unique features, the accurate classification of the current moving target is determined, wherein the accurate classification includes: people, four-wheeled vehicles and two-wheeled vehicles.

6. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 5 is characterized in that: Determining whether the scanned data exists at each adjacent intersection, and if so, controlling the moving target history model corresponding to the adjacent intersection as the moving target model of the current intersection, including: According to the newly appeared position data (x i ,y i ), determine the position F of the mobile target i at the intersection in the three-dimensional intersection model x ∈[F1,F2,…,F n ], where n is the number of branches at the current intersection, and the branch road near the position F is retrieved. x The scanned data sets [S1, S2, S3, …, S j ], analyze the scan data S i Whether it exists in the scan data set [S1, S2, S3, ..., S j ], if the scan data set of the previous intersection [S1, S2, S3, ..., S j ] contains the scan data S corresponding to the moving target a. a With the scan data S i If the same, the control calls the scanning data S a The corresponding historical model M La As the moving target model M of the moving target i Yi .

7. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 5, characterized in that: Determining whether the scan data exists at each adjacent intersection, and if not, controlling to create the mobile target model of the mobile target based on the scan data at the current intersection, including: According to the newly appeared position data (x i ,y i ), determine the position F of the mobile target i at the intersection in the three-dimensional intersection model x ∈[F1,F2,…,F n ], where n is the number of branches at the current intersection, and the branch road near the position F is retrieved. x The scanned data sets [S1, S2, S3, …, S j ], analyze the scan data S i Whether it exists in the scan data set [S1, S2, S3, ..., S j ], if the scan data set of the previous intersection [S1, S2, S3, ..., S j ] does not exist in the scan data S i The same corresponding data, then control according to the feature data set and scanning data S of the mobile target i i Draw the moving target model M of the moving target i Yi .

8. The method for generating three-dimensional holographic intersections based on laser point cloud technology according to claim 1, characterized in that: Analyzing the behavior characteristics of each mobile target model in each waiting area of ​​the intersection of the holographic intersection three-dimensional model to obtain path prediction data of each mobile target, further comprising: Get the four-wheeled vehicle i in the same direction motor vehicle lane in the previous cycle C The actual total number of left turns N ZLS1 、Actual total straight travel N ZSS1 and the actual total right turn N ZRS1 , and get the left turn prediction factor Straight Predictors and right turn predictors Get the non-motorized target i in the same direction non-motorized lane in the previous cycle F The actual total left turn target N ZLS2 、Actual total straight target amount N ZSS2 and the actual total right turn target N ZRS2 , get the target left turn prediction factor Target Straight Predictor and target right turn predictor 9. A holographic intersection 3D generation system for the holographic intersection 3D generation method based on laser point cloud technology as described in any one of claims 1 to 8, characterized in that: include: An intersection modeling unit, wherein the intersection modeling unit establishes a three-dimensional intersection model corresponding to the overall structure of the intersection; A data acquisition unit, the data acquisition unit acquires position information of a mobile target within the overall structure of the intersection and scan data corresponding to the mobile target; a judgment query unit, wherein the judgment query unit judges whether the scan data exists at each adjacent intersection, and if so, controls the mobile target history model corresponding to the adjacent intersection to be used as the mobile target model of the current intersection; if not, controls the mobile target model of the mobile target to be created at the current intersection based on the scan data; and A synthesis and analysis unit, wherein the synthesis and analysis unit imports the mobile target model into the corresponding position in the three-dimensional intersection model to synthesize a holographic intersection three-dimensional model, analyzes the behavior characteristics of each mobile target model in each waiting area at the intersection of the holographic intersection three-dimensional model, obtains path prediction data of each mobile target, and uploads it to the traffic light control system.

Images