CN-121999416-A - Passenger transport vehicle door anti-pinch control method and system based on visual recognition

CN121999416ACN 121999416 ACN121999416 ACN 121999416ACN-121999416-A

Abstract

The invention provides a passenger transport vehicle door anti-pinch control method and system based on visual identification. Belongs to the technical field of intelligent traffic and vehicle safety control. The method comprises the steps of collecting a video picture of a vehicle door area through a video collecting module to generate original video data of the vehicle door area, carrying out video preprocessing according to the original video data of the vehicle door area to generate preprocessed video data, inputting the preprocessed video data into a target identifying module to identify a human body or a carried object target to generate target identifying data, carrying out target feature extraction according to the target identifying data to generate target feature data, and improving response speed of the vehicle door anti-pinch control method of the passenger vehicle based on visual identification, wherein the response speed of the vehicle door anti-pinch control method can be immediately reacted when the target just enters a dangerous area, and potential safety hazards caused by response lag are reduced.

Inventors

LAN YUNGANG
HU XIANHUI
XIAO MANCHENG
OUYANG QUAN
ZHOU JIE

Assignees

深圳市昊岳科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260210

Claims (10)

1. A passenger vehicle door pinch prevention control method based on visual recognition, the method comprising: S1, acquiring a door region video picture through a video acquisition module to generate door region original video data; S2, inputting the preprocessed video data into a target recognition module, recognizing a human body or a carried object target to generate target recognition data; S3, inputting target characteristic data into a dangerous area modeling module, judging whether a target enters a dangerous area of a vehicle door, and generating target position judging data; S4, inputting the vehicle door risk level data into a door closing safety decision module, if the vehicle door risk level data is at risk, generating a door closing prohibiting or interrupting instruction, and transmitting the door closing prohibiting or interrupting instruction to a vehicle door control linkage module to execute corresponding actions; S5, if the driver continuously presses the door closing button for a long time to reach a preset time threshold, inputting the operation data into a manual confirmation module to generate manual confirmation instruction data, transmitting the manual confirmation instruction data to a door closing safety decision module, temporarily removing the anti-clamping limit, and generating a door closing permission instruction; And S6, transmitting a door closing permission instruction to the door control linkage module to execute door closing action, and after the door is closed, recovering the system monitoring state, restarting the video acquisition module to acquire the video picture of the door area and entering the next monitoring cycle.
2. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 1, wherein S1 comprises: S11, performing frame synchronization calibration on an original video stream of a vehicle door area, and aligning time stamps and space coordinates of a plurality of frame images to generate a frame synchronization video stream; S12, screening effective pictures in the door moving range from the frame synchronous video stream to generate a door region effective frame set; s13, gray level equalization processing is carried out on the effective frame set of the vehicle door area, and gray level equalization video frame data are generated; s14, pixel-level noise filtering is carried out on the gray-scale balanced video frame data, and noise reduction video frame data are generated; S15, carrying out resolution unification and color space regularity on the noise reduction video frame data to generate preprocessing video data.
3. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 2, wherein S11 comprises: Extracting frame time sequence information and acquisition equipment parameters of an original video stream of a vehicle door area, and generating an original frame information set; Performing unified reference calibration on the time stamps of the multi-frame images based on the original frame information set to generate time synchronization frame data; Performing space coordinate offset correction on the time synchronization frame data by combining the acquisition equipment parameters to generate space-time initial calibration frame data; And performing fine adjustment and optimization on the time primary calibration frame data according to the frame alignment precision data to generate a frame synchronous video stream.
4. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 1, wherein S2 comprises: s21, carrying out overlapping region clustering de-duplication on the target candidate region set, merging the repeatedly detected same target region, and generating pure candidate region data; S22, carrying out target probability judgment on the pure candidate region data, screening out a high confidence region which accords with the characteristics of a human body or a carried object, and generating target identification data; S23, extracting edge contours of areas corresponding to the target identification data, separating the target from background pixels, and generating target contour data; and S24, extracting core information based on the target contour data, wherein the core information comprises contour curvature, length-width ratio and texture distribution, constructing a multi-dimensional feature vector and generating target feature data.
5. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 1, wherein S3 comprises: s31, extracting real-time position coordinates in target feature data, and performing space superposition comparison with dynamic dangerous area boundary data to generate target area superposition data; s32, calculating the area occupation ratio of the target intrusion dangerous area according to the target area overlapping data, and generating target intrusion degree data; S33, combining the target intrusion degree data and the target movement speed data, judging the movement trend of the target approaching or moving away from the vehicle door, and generating target movement trend data; s34, integrating the target intrusion degree data and the target movement trend data, dividing different risk levels, and generating vehicle door risk level data.
6. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 5, wherein S31 comprises: Analyzing the characteristic data of the target, extracting a three-dimensional coordinate point set and a coordinate acquisition time stamp of the target, and generating an original coordinate data set of the target; Correcting the coordinate precision of the original coordinate data set of the target, eliminating coordinate offset caused by the error of the acquisition equipment, and generating calibrated target coordinate data; extracting boundary coordinate points and region range parameters in the dynamic dangerous region boundary data, and generating a dangerous region boundary feature set; carrying out space dimension alignment on the calibrated target coordinate data and the dangerous area boundary feature set to generate space-time alignment data; and carrying out region intersection calculation on the space-time alignment data, marking the part of the target coordinate falling in the dangerous region, and generating target region overlapping data.
7. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 1, wherein S4 comprises: s41, if the initial safety decision signal is that the risk exists, generating an initial instruction for prohibiting or interrupting closing the door; S42, converting a signal format of the initial command, and matching a communication protocol of the vehicle door control linkage module to generate a standardized control command; S43, transmitting a standardized control instruction to the vehicle door control linkage module to drive the vehicle door to execute the stop closing or reverse opening actions; S44, extracting key risk parameters in the vehicle door risk level data, converting the key risk parameters into video acquisition adjusting signals, and transmitting the video acquisition adjusting signals to a video acquisition module; s45, the video acquisition module adjusts the acquisition frame rate, the exposure time and the focusing range according to the adjusting signal, and generates optimized video acquisition configuration.
8. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 1, wherein S5 comprises: S51, comparing the door closing button operation data with a preset time threshold, judging whether the pressing time is up to standard, and generating a time judging signal; S52, if the duration judging signal is up to standard, carrying out validity verification on the pressing strength data, eliminating invalid operation signals generated by false touch, and generating valid operation signals; s53, inputting an effective operation signal into a manual confirmation module, starting the verification of the validity of the manual operation, and generating manual confirmation instruction data; s54, transmitting the manual confirmation instruction data to a door closing safety decision module, triggering a temporary anti-pinch limit release program, and generating a door closing permission instruction.
9. The method for controlling door anti-pinch control of passenger vehicle based on visual recognition according to claim 1, wherein S6 comprises: S61, the door control linkage module adjusts the output torque and the rotating speed of the driving motor according to the door closing driving signal to generate door closing execution parameters; S62, driving the vehicle door to start closing according to the door closing execution parameters, and collecting the closing progress data of the vehicle door in real time; s63, transmitting a door closing completion signal to a system main control module, triggering a monitoring state recovery program, and resetting all identification parameters and decision thresholds; S64, the system main control module sends a starting signal to the video acquisition module, the acquisition of the video picture of the door area is restarted, a new door area original video stream is generated, and the next monitoring cycle is started.
10. A system for implementing a vision-based passenger vehicle door anti-pinch control method as set forth in claim 1, the system comprising: the video acquisition module is used for acquiring a video picture of the door area through the video acquisition module to generate original video data of the door area; the target recognition module is used for inputting the preprocessed video data into the target recognition module, recognizing the human body or the object carried by the human body to generate target recognition data; The dangerous area modeling module is used for inputting the target characteristic data into the dangerous area modeling module, judging whether the target enters the dangerous area of the vehicle door and generating target position judging data; The door closing safety decision module is used for inputting the door risk level data into the door closing safety decision module, generating a command for prohibiting or interrupting the door closing if the risk exists, and transmitting the command to the door control linkage module to execute corresponding actions; The manual confirmation module is used for inputting the operation data into the manual confirmation module to generate manual confirmation instruction data if the driver continuously presses the door closing button for a long time to reach a preset time threshold; And the door control linkage module transmits a door closing permission instruction to the door control linkage module to execute a door closing action, and after the door is closed, the system monitoring state is restored, the video acquisition module is restarted to acquire the video picture of the door area, and the next monitoring cycle is started.

Description

Passenger transport vehicle door anti-pinch control method and system based on visual recognition Technical Field The invention provides a passenger carrying vehicle door anti-pinch control method and system based on visual identification, and belongs to the technical field of intelligent traffic and vehicle safety control. Background The existing car door anti-pinch technology mostly adopts a mechanical rebound, current detection or infrared sensing mode, and the problems of lag response, high misjudgment rate and insufficient recognition capability for children and carried articles generally exist. With the popularization of passenger cameras on vehicles, the existing system still does not effectively use visual information for safety control of vehicle doors, and under the conditions of complex illumination, rain and snow reflection and the like, false detection can occur to a visual algorithm, so that the operation efficiency is affected. Disclosure of Invention The invention provides a passenger transport vehicle door anti-pinch control method and system based on visual identification, which are used for solving the problems mentioned in the background art: the invention provides a passenger transport vehicle door anti-pinch control method based on visual identification, which comprises the following steps: S1, acquiring a door region video picture through a video acquisition module to generate door region original video data; S2, inputting the preprocessed video data into a target recognition module, recognizing a human body or a carried object target to generate target recognition data; S3, inputting target characteristic data into a dangerous area modeling module, judging whether a target enters a dangerous area of a vehicle door, and generating target position judging data; S4, inputting the vehicle door risk level data into a door closing safety decision module, if the vehicle door risk level data is at risk, generating a door closing prohibiting or interrupting instruction, and transmitting the door closing prohibiting or interrupting instruction to a vehicle door control linkage module to execute corresponding actions; S5, if the driver continuously presses the door closing button for a long time to reach a preset time threshold, inputting the operation data into a manual confirmation module to generate manual confirmation instruction data, transmitting the manual confirmation instruction data to a door closing safety decision module, temporarily removing the anti-clamping limit, and generating a door closing permission instruction; And S6, transmitting a door closing permission instruction to the door control linkage module to execute door closing action, and after the door is closed, recovering the system monitoring state, restarting the video acquisition module to acquire the video picture of the door area and entering the next monitoring cycle. The invention provides a system for realizing the anti-pinch control method of a passenger vehicle door based on visual identification, which comprises the following steps: the video acquisition module is used for acquiring a video picture of the door area through the video acquisition module to generate original video data of the door area; the target recognition module is used for inputting the preprocessed video data into the target recognition module, recognizing the human body or the object carried by the human body to generate target recognition data; The dangerous area modeling module is used for inputting the target characteristic data into the dangerous area modeling module, judging whether the target enters the dangerous area of the vehicle door and generating target position judging data; The door closing safety decision module is used for inputting the door risk level data into the door closing safety decision module, generating a command for prohibiting or interrupting the door closing if the risk exists, and transmitting the command to the door control linkage module to execute corresponding actions; The manual confirmation module is used for inputting the operation data into the manual confirmation module to generate manual confirmation instruction data if the driver continuously presses the door closing button for a long time to reach a preset time threshold; And the door control linkage module transmits a door closing permission instruction to the door control linkage module to execute a door closing action, and after the door is closed, the system monitoring state is restored, the video acquisition module is restarted to acquire the video picture of the door area, and the next monitoring cycle is started. The anti-pinch control method for the passenger-carrying vehicle door based on visual identification has the advantages that the anti-pinch response speed of the door is improved, the door can immediately react when the target just enters a dangerous area, the potential safety hazard caused by response delay is reduced, real risks a