Intelligent identification method and system based on computer vision

Abstract

The invention is suitable for the technical field of computer vision, and provides an intelligent identification method and system based on computer vision. According to the invention, triggering monitoring is carried out on two target monitoring directions, and an early warning monitoring direction is selected; determining whether there is an early warning suspicious object; according to a preset shooting interval, carrying out interval binocular shooting on the trigger monitoring object and the early warning suspicious object; calculating trigger arrival time and early warning arrival time; and according to the trigger arrival time and the early warning arrival time, an arrival time difference value is calculated, and when the arrival time difference value is smaller than a preset standard time difference value, early warning prompting is carried out. Interval binocular shooting can be performed on a trigger monitoring object and an early warning suspicious object, trigger arrival time and early warning arrival time are calculated, and when an arrival time difference value is smaller than a standard time difference value, early warning prompting is performed on two target monitoring directions, so that early warning prompting away from an intersection is realized, avoidance time is effectively prolonged, and the safety of a road is improved. And safe avoidance when the intersection has a collision risk is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of computer vision, and in particular relates to an intelligent recognition method and system based on computer vision. Background Art

[0002] Computer vision is the use of cameras and computers to replace the human eye to perform machine vision tasks such as target identification, tracking and measurement, and further perform graphic processing so that the computer processing becomes an image more suitable for human eye observation or transmission to instrument detection. It can achieve various complex visual tasks such as object recognition, scene analysis, motion tracking, 3D reconstruction, image segmentation, behavior recognition, etc.

[0003] Intelligent recognition based on computer vision has wide applications in traffic management.

[0004] In the prior art, computer vision technology has not been used for reasonable and effective anti-collision intelligent recognition applications at intersections. It can usually only perform photography and recognition at intersections and issue early warning prompts when the target approaches the intersection, resulting in limited avoidance time. If the target is too fast, it is difficult to achieve safe avoidance. Summary of the invention

[0005] The purpose of the embodiments of the present invention is to provide an intelligent recognition method and system based on computer vision, aiming to solve the technical problems existing in the prior art mentioned in the background technology.

[0006] The embodiment of the present invention is implemented as follows:

[0007] An intelligent recognition method based on computer vision, the method specifically comprises the following steps:

[0008] Trigger monitoring is performed on two target monitoring directions, the trigger monitoring object is determined, the early warning monitoring direction is selected, and initial binocular shooting is performed to obtain initial shooting data;

[0009] Based on computer vision technology, the initial shooting data is identified and analyzed to determine whether there is a warning suspicious object;

[0010] According to the preset shooting interval, the trigger monitoring object is photographed by binocular interval to obtain trigger shooting data, and the warning suspicious object is photographed by binocular interval to obtain warning shooting data;

[0011] Analyze the initial shooting data, the triggered shooting data and the early warning shooting data, and calculate the trigger arrival time and the early warning arrival time;

[0012] The arrival time difference is calculated according to the trigger arrival time and the warning arrival time, and when the arrival time difference is less than a preset standard time difference, a warning prompt is issued for the two target monitoring directions.

[0013] As a further limitation of the technical solution of the embodiment of the present invention, the triggering monitoring of two target monitoring directions, determining the triggering monitoring object, selecting the early warning monitoring direction, and performing initial binocular shooting, and obtaining the initial shooting data specifically include the following steps:

[0014] Perform trigger monitoring on two target monitoring directions to obtain trigger monitoring data;

[0015] Analyzing the trigger monitoring data to determine the trigger monitoring object;

[0016] From the two target monitoring directions, select the target monitoring direction that does not have the trigger monitoring object, and mark it as the early warning monitoring direction;

[0017] An initial binocular shooting is performed on the early warning monitoring direction to obtain initial shooting data.

[0018] As a further limitation of the technical solution of the embodiment of the present invention, the identification and analysis of the initial shooting data based on computer vision technology to determine whether there is a warning suspicious object specifically includes the following steps:

[0019] Based on computer vision technology, feature recognition is performed on the initial shooting data to determine whether it has warning appearance features;

[0020] Based on computer vision technology, the initial shooting data is analyzed for direction to determine whether the warning direction is met;

[0021] Based on computer vision technology, the distance analysis is performed on the initial shooting data to determine whether it is within the warning suspicious range;

[0022] If the object has the warning appearance characteristics, meets the warning suspicious direction and is within the warning suspicious range, it is determined to be a warning suspicious object;

[0023] When the object does not have the warning appearance features, does not meet the warning suspicious direction and / or is not within the warning suspicious range, it is determined that there is no warning suspicious object.

[0024] As a further limitation of the technical solution of the embodiment of the present invention, the trigger monitoring object is photographed at intervals by binocular according to a preset shooting interval to obtain trigger shooting data, and the warning suspicious object is photographed at intervals by binocular, and obtaining the warning shooting data specifically includes the following steps:

[0025] Generate interval shooting instructions according to the preset shooting interval;

[0026] According to the interval shooting instruction, the trigger monitoring object is photographed by binocular at intervals to obtain trigger shooting data;

[0027] According to the interval shooting instruction, the warning suspicious object is photographed by binocular at intervals to obtain warning shooting data.

[0028] As a further limitation of the technical solution of the embodiment of the present invention, the analyzing of the initial shooting data, the trigger shooting data and the warning shooting data to calculate the trigger arrival time and the warning arrival time specifically includes the following steps:

[0029] Analyzing the initial shooting data to determine an initial warning distance;

[0030] Analyzing the triggered shooting data to determine the trigger interval distance;

[0031] Analyze the warning shooting data to determine the warning interval distance;

[0032] The trigger arrival time and the warning arrival time are calculated according to the warning initial distance, the trigger interval distance and the warning interval distance.

[0033] As a further limitation of the technical solution of the embodiment of the present invention, the calculation formula of the trigger arrival time is:

[0034]

[0035] Among them, T a is the trigger arrival time, T z For the shooting interval, is the preset trigger distance, is the trigger interval distance, H is the preset binocular shooting height;

[0036] The calculation formula of the warning arrival time is:

[0037]

[0038] Among them, T b To warn of arrival time, To warn the initial distance, The warning interval distance.

[0039] As a further limitation of the technical solution of the embodiment of the present invention, the calculation of the arrival time difference according to the trigger arrival time and the warning arrival time, and issuing a warning prompt for two target monitoring directions when the arrival time difference is less than a preset standard time difference specifically includes the following steps:

[0040] Calculate the arrival time difference according to the trigger arrival time and the warning arrival time;

[0041] Compare the arrival time difference with a preset standard time difference;

[0042] When the arrival time difference is less than the standard time difference, generating an early warning signal;

[0043] According to the two target monitoring directions, two early warning prompt locations are selected;

[0044] At the two early warning prompting positions, early warning prompts are given to the two target monitoring directions according to the early warning prompting signals.

[0045] An intelligent recognition system based on computer vision, the system includes a trigger monitoring processing module, an initial recognition judgment module, an interval binocular shooting module, an arrival time calculation module and an early warning prompt processing module, wherein:

[0046] The trigger monitoring processing module is used to trigger monitoring of two target monitoring directions, determine the trigger monitoring object, select the early warning monitoring direction, perform initial binocular shooting, and obtain initial shooting data;

[0047] An initial recognition and judgment module is used to recognize and analyze the initial shooting data based on computer vision technology to determine whether there is a warning suspicious object;

[0048] An interval binocular shooting module is used to perform interval binocular shooting on the trigger monitoring object according to a preset shooting interval to obtain trigger shooting data, and to perform interval binocular shooting on the warning suspicious object to obtain warning shooting data;

[0049] An arrival time calculation module is used to analyze the initial shooting data, the trigger shooting data and the warning shooting data to calculate the trigger arrival time and the warning arrival time;

[0050] The early warning processing module is used to calculate the arrival time difference according to the trigger arrival time and the early warning arrival time, and to issue early warning prompts for the two target monitoring directions when the arrival time difference is less than a preset standard time difference.

[0051] As a further limitation of the technical solution of the embodiment of the present invention, the initial recognition and judgment module specifically includes:

[0052] A feature recognition unit, used to perform feature recognition on the initial shooting data based on computer vision technology to determine whether it has a warning appearance feature;

[0053] A direction analysis unit, used to perform direction analysis on the initial shooting data based on computer vision technology to determine whether the warning direction is met;

[0054] A distance analysis unit, used to perform distance analysis on the initial shooting data based on computer vision technology to determine whether it is within a warning suspicious range;

[0055] The identification and judgment unit is used to determine that there is a warning suspicious object when it has the warning appearance characteristics, meets the warning suspicious direction and is within the warning suspicious range; when it does not have the warning appearance characteristics, does not meet the warning suspicious direction and / or is not within the warning suspicious range, it is used to determine that there is no warning suspicious object.

[0056] As a further limitation of the technical solution of the embodiment of the present invention, the early warning prompt processing module specifically includes:

[0057] A difference calculation unit, used for calculating the arrival time difference according to the trigger arrival time and the warning arrival time;

[0058] A difference comparison unit, used for comparing the arrival time difference with a preset standard time difference;

[0059] A signal generating unit, configured to generate a warning signal when the arrival time difference is less than a standard time difference;

[0060] A position selection unit, used to select two early warning prompt positions according to two target monitoring directions;

[0061] The early warning prompt unit is used to provide early warning prompts for two target monitoring directions at two early warning prompt positions according to the early warning prompt signals.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] The embodiment of the present invention triggers monitoring of two target monitoring directions, selects an early warning monitoring direction; determines whether there is an early warning suspicious object; performs interval binocular shooting of the trigger monitoring object and the early warning suspicious object according to a preset shooting interval; calculates the trigger arrival time and the early warning arrival time; calculates the arrival time difference based on the trigger arrival time and the early warning arrival time, and issues an early warning prompt when the arrival time difference is less than the preset standard time difference. It is able to perform interval binocular shooting of the trigger monitoring object and the early warning suspicious object, calculate the trigger arrival time and the early warning arrival time, and issue an early warning prompt for the two target monitoring directions when the arrival time difference is less than the standard time difference, thereby realizing an early warning prompt away from the intersection, effectively extending the avoidance time, and ensuring safe avoidance when there is a collision risk at the intersection. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A flowchart of an intelligent recognition method based on computer vision provided by an embodiment of the present invention is shown;

[0065] Figure 2 A schematic diagram of an application scenario provided by an embodiment of the present invention is shown;

[0066] Figure 3 A flowchart of triggering monitoring and initial binocular shooting in the method provided by an embodiment of the present invention is shown;

[0067] Figure 4 A flowchart of initial shooting data recognition and analysis in the method provided by an embodiment of the present invention is shown;

[0068] Figure 5 A flow chart of performing interval binocular shooting in the method provided by an embodiment of the present invention is shown;

[0069] Figure 6 A flowchart showing the calculation of the trigger arrival time and the warning arrival time in the method provided by the embodiment of the present invention is shown;

[0070] Figure 7 A flow chart showing time difference comparison and early warning prompting in the method provided by an embodiment of the present invention is shown;

[0071] Figure 8 An application architecture diagram of an intelligent recognition system based on computer vision provided by an embodiment of the present invention is shown;

[0072] Fig. 9 An application architecture diagram of an initial identification and judgment module in a system provided by an embodiment of the present invention is shown;

[0073] Fig.10 The application architecture diagram of the early warning prompt processing module in the system provided by the embodiment of the present invention is shown. DETAILED DESCRIPTION

[0074] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0075] It is understandable that in the prior art, computer vision technology has not been applied in a reasonable and effective manner for collision avoidance intelligent recognition at intersections. It is usually only capable of photographing and identifying at intersections, and issuing early warnings when a target approaches an intersection, resulting in limited avoidance time. If the target is moving too fast, safe avoidance is difficult to achieve.

[0076] To solve the above problems, an intelligent recognition method and system based on computer vision disclosed in an embodiment of the present invention is provided. The method and system trigger monitoring of two target monitoring directions, determine the trigger monitoring object, select the warning monitoring direction, and perform initial binocular shooting to obtain initial shooting data; based on computer vision technology, the initial shooting data is identified and analyzed to determine whether there is a warning suspicious object; according to the preset shooting interval, the trigger monitoring object is shot at interval binocular shooting to obtain trigger shooting data, and the warning suspicious object is shot at interval binocular shooting to obtain warning shooting data; the initial shooting data, the trigger shooting data and the warning shooting data are analyzed to calculate the trigger arrival time and the warning arrival time; according to the trigger arrival time and the warning arrival time, the arrival time difference is calculated, and when the arrival time difference is less than the preset standard time difference, the two target monitoring directions are warned. It is possible to perform interval binocular shooting on the trigger monitoring object and the warning suspicious object, calculate the trigger arrival time and the warning arrival time, and when the arrival time difference is less than the standard time difference, the two target monitoring directions are warned, so as to realize the warning prompt away from the intersection, effectively extend the avoidance time, and ensure safe avoidance when there is a collision risk at the intersection.

[0077] Specifically, Figure 1 A flow chart of an intelligent recognition method based on computer vision provided by an embodiment of the present invention is shown.

[0078] In a preferred embodiment of the present invention, a computer vision-based intelligent recognition method comprises the following steps:

[0079] Step S101: trigger monitoring of two target monitoring directions, determine the trigger monitoring object, select the early warning monitoring direction, perform initial binocular shooting, and obtain initial shooting data.

[0080] In an embodiment of the present invention, infrared detection points are provided at the trigger distances of the two target monitoring directions. At the two infrared detection points, the two target monitoring directions are triggered and monitored to obtain trigger monitoring data. By analyzing the trigger monitoring data, it is determined whether there is a trigger signal change. When there is a trigger signal change, it is determined that a triggered monitoring object has passed by. Then, from the two target monitoring directions, the target monitoring direction with the triggered monitoring object is selected and marked as the triggered monitoring direction, and the target monitoring direction without the triggered monitoring object is selected and marked as the early warning monitoring direction. Initial shooting data is obtained by performing initial binocular shooting on the early warning monitoring direction.

[0081] It can be understood that the two target monitoring directions are the directions of the two roads intersecting at the intersection, such as Figure 2A schematic diagram of an application scenario provided by an embodiment of the present invention is shown. In the embodiment of the present invention, pedestrians, motor vehicles and / or non-motor vehicles on roads in target monitoring directions A and target monitoring directions B are monitored, wherein the two intersecting roads may be motor vehicle lane A1 and motor vehicle lane B1, motor vehicle lane A1 and non-motor vehicle lane B2, non-motor vehicle lane A2 and motor vehicle lane B1, or non-motor vehicle lane A2 and non-motor vehicle lane B2.

[0082] Specifically, Figure 3 A flow chart of triggering monitoring and initial binocular shooting in the method provided by an embodiment of the present invention is shown.

[0083] Among them, in the preferred embodiment provided by the present invention, the triggering monitoring of two target monitoring directions, determining the triggering monitoring object, selecting the early warning monitoring direction, and performing initial binocular shooting, and obtaining the initial shooting data specifically include the following steps:

[0084] Step S1011: trigger monitoring of two target monitoring directions to obtain triggered monitoring data.

[0085] Step S1012: Analyze the trigger monitoring data to determine the trigger monitoring object.

[0086] Step S1013: From the two target monitoring directions, select the target monitoring direction that does not have the trigger monitoring object, and mark it as the early warning monitoring direction.

[0087] Step S1014: perform initial binocular photography on the warning monitoring direction to obtain initial photography data.

[0088] Furthermore, the computer vision-based intelligent recognition method further comprises the following steps:

[0089] Step S102: Based on computer vision technology, the initial shooting data is identified and analyzed to determine whether there is a suspicious object to be warned.

[0090] In an embodiment of the present invention, based on computer vision technology, feature recognition is performed on the initial shooting data to determine whether it has warning appearance features, and then direction analysis is performed on the initial shooting data to determine whether it satisfies the warning direction, and distance analysis is performed on the initial shooting data to determine whether it is in the warning suspicious range. An occlusion analysis can also be performed on the intersection in the warning monitoring direction to determine whether the intersection has field of view obstruction. When it has warning appearance features, satisfies the warning suspicious direction, is in the warning suspicious range, and the intersection has field of view obstruction, it is determined that there is a warning suspicious object; when it does not have warning appearance features, does not meet the warning suspicious direction, is not in the warning suspicious range and / or the intersection does not have field of view obstruction, it is determined that there is no warning suspicious object.

[0091] It can be understood that the warning appearance features are the preset standard appearance features of pedestrians, motor vehicles and non-motor vehicles; the warning direction is opposite to the warning monitoring direction, towards the intersection; the warning suspicious range is the preset warning range value.

[0092] Specifically, Figure 4 A flow chart of initial shooting data recognition and analysis in the method provided by an embodiment of the present invention is shown.

[0093] Among them, in the preferred embodiment provided by the present invention, the identification and analysis of the initial shooting data based on computer vision technology to determine whether there is a warning suspicious object specifically includes the following steps:

[0094] Step S1021: Based on computer vision technology, feature recognition is performed on the initial shooting data to determine whether it has warning appearance features.

[0095] Step S1022: Based on computer vision technology, the initial shooting data is subjected to direction analysis to determine whether the warning direction is met.

[0096] Step S1023: Based on computer vision technology, perform distance analysis on the initial shooting data to determine whether it is within a warning suspicious range.

[0097] Step S1024: When the object has the warning appearance features, meets the warning suspicious direction and is within the warning suspicious range, it is determined to be a warning suspicious object.

[0098] Step S1025: When the object does not have the warning appearance feature, does not meet the warning suspicious direction, and / or is not within the warning suspicious range, it is determined that there is no warning suspicious object.

[0099] Furthermore, the computer vision-based intelligent recognition method further comprises the following steps:

[0100] Step S103: performing interval binocular photography on the trigger monitoring object according to a preset photography interval to obtain trigger photography data, and performing interval binocular photography on the warning suspicious object to obtain warning photography data.

[0101] In an embodiment of the present invention, an interval shooting instruction is generated according to a preset shooting interval, and then according to the interval shooting instruction, interval binocular shooting is performed on the triggered monitoring object along the trigger monitoring direction to obtain trigger shooting data, and according to the interval shooting instruction, interval binocular shooting is performed on the warning suspicious object along the warning monitoring direction to obtain warning shooting data.

[0102] Specifically, Figure 5 A flow chart of performing interval binocular shooting in the method provided by an embodiment of the present invention is shown.

[0103] Among them, in the preferred embodiment provided by the present invention, the trigger monitoring object is photographed at intervals by binocular according to a preset shooting interval to obtain trigger shooting data, and the warning suspicious object is photographed at intervals by binocular, and obtaining the warning shooting data specifically includes the following steps:

[0104] Step S1031: Generate an interval shooting instruction according to a preset shooting interval.

[0105] Step S1032: perform interval binocular photography on the triggered monitoring object according to the interval photography instruction to obtain trigger photography data.

[0106] Step S1033: perform interval binocular photography of the warning suspicious object according to the interval photography instruction to obtain warning photography data.

[0107] Furthermore, the computer vision-based intelligent recognition method further comprises the following steps:

[0108] Step S104: Analyze the initial shooting data, the trigger shooting data and the warning shooting data to calculate the trigger arrival time and the warning arrival time.

[0109] In the embodiment of the present invention, the initial distance of warning of suspicious objects is determined by performing binocular spatial analysis on the initial shooting data, and then the triggering interval distance of the monitoring object is determined by performing binocular spatial analysis on the trigger shooting data, and the warning interval distance of the suspicious objects is determined by performing binocular spatial analysis on the warning shooting data. Calculate the trigger arrival time, where T a is the trigger arrival time, T z For the shooting interval, is the preset trigger distance, is the trigger interval distance, H is the preset binocular shooting height, and then according to Calculate the warning arrival time, where T b To warn of arrival time, To warn the initial distance, The warning interval distance.

[0110] Specifically, Figure 6 A flow chart of calculating the trigger arrival time and the warning arrival time in the method provided by an embodiment of the present invention is shown.

[0111] Among them, in the preferred implementation manner provided by the present invention, the analyzing of the initial shooting data, the trigger shooting data and the warning shooting data, and calculating the trigger arrival time and the warning arrival time specifically includes the following steps:

[0112] Step S1041: Analyze the initial shooting data to determine the initial warning distance.

[0113] Step S1042: Analyze the triggered shooting data to determine the trigger interval distance.

[0114] Step S1043: Analyze the warning shooting data to determine the warning interval distance.

[0115] Step S1044: Calculate the trigger arrival time and the warning arrival time according to the warning initial distance, the trigger interval distance and the warning interval distance.

[0116] Furthermore, the computer vision-based intelligent recognition method further comprises the following steps:

[0117] Step S105: Calculate the arrival time difference according to the trigger arrival time and the warning arrival time, and issue a warning prompt for two target monitoring directions when the arrival time difference is less than a preset standard time difference.

[0118] In the embodiment of the present invention, according to T c =|T a -T b |Calculate the arrival time difference, where T c For the arrival time difference, the arrival time difference is compared with the preset standard time difference. When the arrival time difference is less than the standard time difference, an early warning signal is generated. At this time, two early warning positions are selected according to the two target monitoring directions. Then, at the two early warning positions, according to the early warning signal, early warning prompts are given for the two target monitoring directions to avoid triggering a collision between the monitored object and the suspicious object being warned.

[0119] Specifically, Figure 7 A flow chart of time difference comparison and early warning prompt in the method provided by an embodiment of the present invention is shown.

[0120] Among them, in the preferred implementation manner provided by the present invention, the calculation of the arrival time difference according to the trigger arrival time and the warning arrival time, and when the arrival time difference is less than the preset standard time difference, issuing a warning prompt for the two target monitoring directions specifically includes the following steps:

[0121] Step S1051: Calculate the arrival time difference according to the trigger arrival time and the warning arrival time.

[0122] Step S1052: compare the arrival time difference with a preset standard time difference.

[0123] Step S1053: When the arrival time difference is less than the standard time difference, a warning prompt signal is generated.

[0124] Step S1054: select two early warning prompt positions according to the two target monitoring directions.

[0125] Step S1055: At two warning prompt positions, according to the warning prompt signal, a warning prompt is given for two target monitoring directions.

[0126] Furthermore, Figure 8 The application architecture diagram of the computer vision-based intelligent recognition system provided by an embodiment of the present invention is shown.

[0127] Among them, in another preferred embodiment provided by the present invention, a computer vision-based intelligent recognition system includes:

[0128] The trigger monitoring processing module 101 is used to trigger monitoring of two target monitoring directions, determine the trigger monitoring object, select the early warning monitoring direction, perform initial binocular shooting, and obtain initial shooting data.

[0129] In an embodiment of the present invention, infrared detection points are provided at the triggering distances of the two target monitoring directions. The trigger monitoring processing module 101 performs trigger monitoring on the two target monitoring directions at the two infrared detection points to obtain trigger monitoring data. By analyzing the trigger monitoring data, it is determined whether there is a trigger signal change. When there is a trigger signal change, it is determined that a trigger monitoring object has passed by. Then, from the two target monitoring directions, the target monitoring direction with the trigger monitoring object is selected and marked as the trigger monitoring direction, and the target monitoring direction without the trigger monitoring object is selected and marked as the early warning monitoring direction. Initial shooting data is obtained by performing initial binocular shooting on the early warning monitoring direction.

[0130] The initial recognition and judgment module 102 is used to recognize and analyze the initial shooting data based on computer vision technology to determine whether there is a suspicious object to be warned.

[0131] In an embodiment of the present invention, the initial identification and judgment module 102 performs feature recognition on the initial shooting data based on computer vision technology to determine whether it has warning appearance features, and then performs direction analysis on the initial shooting data to determine whether it meets the warning direction, and performs distance analysis on the initial shooting data to determine whether it is in the warning suspicious range. It can also perform occlusion analysis on the intersection in the warning monitoring direction to determine whether the intersection has field of view occlusion. When it has warning appearance features, meets the warning suspicious direction, is in the warning suspicious range, and the intersection has field of view occlusion, it is determined that there is a warning suspicious object; when it does not have warning appearance features, does not meet the warning suspicious direction, is not in the warning suspicious range and / or the intersection does not have field of view occlusion, it is determined that there is no warning suspicious object.

[0132] Specifically, Fig. 9 The application architecture diagram of the initial identification and judgment module 102 in the system provided by the embodiment of the present invention is shown.

[0133] Among them, in the preferred embodiment provided by the present invention, the initial identification and judgment module 102 specifically includes:

[0134] The feature recognition unit 1021 is used to perform feature recognition on the initial shooting data based on computer vision technology to determine whether it has warning appearance features.

[0135] The direction analysis unit 1022 is used to perform direction analysis on the initial shooting data based on computer vision technology to determine whether the warning direction is met.

[0136] The distance analysis unit 1023 is used to perform distance analysis on the initial shooting data based on computer vision technology to determine whether it is within the warning suspicious range.

[0137] The identification and judgment unit 1024 is used to determine that there is a warning suspicious object when it has the warning appearance characteristics, meets the warning suspicious direction and is within the warning suspicious range; when it does not have the warning appearance characteristics, does not meet the warning suspicious direction and / or is not within the warning suspicious range, it is used to determine that there is no warning suspicious object.

[0138] Furthermore, the computer vision-based intelligent recognition system also includes:

[0139] The interval binocular shooting module 103 is used to perform interval binocular shooting on the trigger monitoring object according to a preset shooting interval to obtain trigger shooting data, and to perform interval binocular shooting on the warning suspicious object to obtain warning shooting data.

[0140] In an embodiment of the present invention, the interval binocular shooting module 103 generates an interval shooting instruction according to a preset shooting interval, and then performs interval binocular shooting on the triggered monitoring object along the trigger monitoring direction according to the interval shooting instruction to obtain trigger shooting data, and performs interval binocular shooting on the warning suspicious object along the warning monitoring direction according to the interval shooting instruction to obtain warning shooting data.

[0141] The arrival time calculation module 104 is used to analyze the initial shooting data, the triggered shooting data and the warning shooting data to calculate the triggered arrival time and the warning arrival time.

[0142] In the embodiment of the present invention, the arrival time calculation module 104 performs binocular spatial analysis on the initial shooting data to determine the initial warning distance of the warning suspicious object, and then performs binocular spatial analysis on the trigger shooting data to determine the trigger interval distance of the trigger monitoring object, and performs binocular spatial analysis on the warning shooting data to determine the warning interval distance of the warning suspicious object, according to Calculate the trigger arrival time, where T a is the trigger arrival time, T z For the shooting interval, is the preset trigger distance, is the trigger interval distance, H is the preset binocular shooting height, and then according to Calculate the warning arrival time, where T b To warn of arrival time, To warn the initial distance, The warning interval distance.

[0143] The early warning processing module 105 is used to calculate the arrival time difference according to the trigger arrival time and the early warning arrival time, and to issue early warning prompts to the two target monitoring directions when the arrival time difference is less than a preset standard time difference.

[0144] In the embodiment of the present invention, the early warning prompt processing module 105 is based on T c =|T a -T b |Calculate the arrival time difference, where T c For the arrival time difference, the arrival time difference is compared with the preset standard time difference. When the arrival time difference is less than the standard time difference, an early warning signal is generated. At this time, two early warning positions are selected according to the two target monitoring directions. Then, at the two early warning positions, according to the early warning signal, early warning prompts are given for the two target monitoring directions to avoid triggering a collision between the monitored object and the suspicious object being warned.

[0145] Specifically, Fig.10The application architecture diagram of the early warning prompt processing module 105 in the system provided by the embodiment of the present invention is shown.

[0146] Among them, in the preferred embodiment provided by the present invention, the early warning prompt processing module 105 specifically includes:

[0147] The difference calculation unit 1051 is used to calculate the arrival time difference according to the trigger arrival time and the warning arrival time.

[0148] The difference comparison unit 1052 is used to compare the arrival time difference with a preset standard time difference.

[0149] The signal generating unit 1053 is used to generate a warning signal when the arrival time difference is less than the standard time difference.

[0150] The position selection unit 1054 is used to select two early warning prompt positions according to two target monitoring directions.

[0151] The early warning prompt unit 1055 is used to provide early warning prompts for two target monitoring directions at two early warning prompt positions according to the early warning prompt signals.

[0152] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0153] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0154] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An intelligent recognition method based on computer vision, characterized in that: The method specifically comprises the following steps: Trigger monitoring is performed on two target monitoring directions, the trigger monitoring object is determined, the early warning monitoring direction is selected, and initial binocular shooting is performed to obtain initial shooting data; Based on computer vision technology, the initial shooting data is identified and analyzed to determine whether there is a warning suspicious object; According to the preset shooting interval, the trigger monitoring object is photographed by binocular interval to obtain trigger shooting data, and the warning suspicious object is photographed by binocular interval to obtain warning shooting data; Analyze the initial shooting data, the triggered shooting data and the early warning shooting data, and calculate the trigger arrival time and the early warning arrival time; The arrival time difference is calculated according to the trigger arrival time and the warning arrival time, and when the arrival time difference is less than a preset standard time difference, a warning prompt is issued for the two target monitoring directions.

2. The computer vision-based intelligent recognition method according to claim 1, characterized in that: The triggering monitoring of two target monitoring directions, determining the trigger monitoring object, selecting the early warning monitoring direction, and performing initial binocular shooting to obtain initial shooting data specifically include the following steps: Perform trigger monitoring on two target monitoring directions to obtain trigger monitoring data; Analyzing the trigger monitoring data to determine the trigger monitoring object; From the two target monitoring directions, select the target monitoring direction that does not have the trigger monitoring object, and mark it as the early warning monitoring direction; An initial binocular shooting is performed on the early warning monitoring direction to obtain initial shooting data.

3. The computer vision-based intelligent recognition method according to claim 1, characterized in that: The method of identifying and analyzing the initial photographed data based on computer vision technology to determine whether there is a warning suspicious object specifically includes the following steps: Based on computer vision technology, feature recognition is performed on the initial shooting data to determine whether it has warning appearance features; Based on computer vision technology, the initial shooting data is analyzed for direction to determine whether the warning direction is met; Based on computer vision technology, the distance analysis is performed on the initial shooting data to determine whether it is within the warning suspicious range; If the object has the warning appearance characteristics, meets the warning suspicious direction and is within the warning suspicious range, it is determined to be a warning suspicious object; When the object does not have the warning appearance features, does not meet the warning suspicious direction and / or is not within the warning suspicious range, it is determined that there is no warning suspicious object.

4. The computer vision-based intelligent recognition method according to claim 1, characterized in that: The method of performing interval binocular photography on the trigger monitoring object according to the preset photography interval to obtain trigger photography data, and performing interval binocular photography on the warning suspicious object to obtain warning photography data specifically includes the following steps: Generate interval shooting instructions according to the preset shooting interval; According to the interval shooting instruction, the trigger monitoring object is photographed by binocular at intervals to obtain trigger shooting data; According to the interval shooting instruction, the warning suspicious object is photographed by binocular at intervals to obtain warning shooting data.

5. The computer vision-based intelligent recognition method according to claim 1, characterized in that: The analyzing the initial shooting data, the trigger shooting data and the warning shooting data to calculate the trigger arrival time and the warning arrival time specifically comprises the following steps: Analyzing the initial shooting data to determine an initial warning distance; Analyzing the triggered shooting data to determine the trigger interval distance; Analyze the warning shooting data to determine the warning interval distance; The trigger arrival time and the warning arrival time are calculated according to the warning initial distance, the trigger interval distance and the warning interval distance.

6. The computer vision-based intelligent recognition method according to claim 5, characterized in that: The calculation formula of the trigger arrival time is: Among them, T a is the trigger arrival time, T z For the shooting interval, is the preset trigger distance, is the trigger interval distance, H is the preset binocular shooting height; The calculation formula of the warning arrival time is: Among them, T b To warn of arrival time, To warn the initial distance, The warning interval distance.

7. The computer vision-based intelligent recognition method according to claim 1, characterized in that: The calculation of the arrival time difference according to the trigger arrival time and the warning arrival time, and issuing a warning prompt for two target monitoring directions when the arrival time difference is less than a preset standard time difference specifically includes the following steps: Calculate the arrival time difference according to the trigger arrival time and the warning arrival time; Compare the arrival time difference with a preset standard time difference; When the arrival time difference is less than the standard time difference, generating an early warning signal; According to the two target monitoring directions, two early warning prompt locations are selected; At the two early warning prompting positions, early warning prompts are given to the two target monitoring directions according to the early warning prompting signals.

8. An intelligent recognition system based on computer vision, characterized in that: The system includes a trigger monitoring processing module, an initial recognition and judgment module, an interval binocular shooting module, an arrival time calculation module and an early warning prompt processing module, wherein: The trigger monitoring processing module is used to trigger monitoring of two target monitoring directions, determine the trigger monitoring object, select the early warning monitoring direction, perform initial binocular shooting, and obtain initial shooting data; An initial recognition and judgment module is used to recognize and analyze the initial shooting data based on computer vision technology to determine whether there is a warning suspicious object; An interval binocular shooting module is used to perform interval binocular shooting on the trigger monitoring object according to a preset shooting interval to obtain trigger shooting data, and to perform interval binocular shooting on the warning suspicious object to obtain warning shooting data; An arrival time calculation module is used to analyze the initial shooting data, the trigger shooting data and the warning shooting data to calculate the trigger arrival time and the warning arrival time; The early warning processing module is used to calculate the arrival time difference according to the trigger arrival time and the early warning arrival time, and to issue early warning prompts for the two target monitoring directions when the arrival time difference is less than a preset standard time difference.

9. The computer vision-based intelligent recognition system according to claim 7, characterized in that: The initial identification and judgment module specifically includes: A feature recognition unit, used to perform feature recognition on the initial shooting data based on computer vision technology to determine whether it has a warning appearance feature; A direction analysis unit, used to perform direction analysis on the initial shooting data based on computer vision technology to determine whether the warning direction is met; A distance analysis unit, used to perform distance analysis on the initial shooting data based on computer vision technology to determine whether it is within a warning suspicious range; The identification and judgment unit is used to determine that there is a warning suspicious object when it has the warning appearance characteristics, meets the warning suspicious direction and is within the warning suspicious range; when it does not have the warning appearance characteristics, does not meet the warning suspicious direction and / or is not within the warning suspicious range, it is used to determine that there is no warning suspicious object.

10. The computer vision-based intelligent recognition system according to claim 7, characterized in that: The early warning processing module specifically includes: A difference calculation unit, used for calculating the arrival time difference according to the trigger arrival time and the warning arrival time; A difference comparison unit, used for comparing the arrival time difference with a preset standard time difference; A signal generating unit, configured to generate a warning signal when the arrival time difference is less than a standard time difference; A position selection unit, used to select two early warning prompt positions according to two target monitoring directions; The early warning prompt unit is used to provide early warning prompts for two target monitoring directions at two early warning prompt positions according to the early warning prompt signals.

Images