CN-121989988-A - Vehicle optimization control method and system based on vehicle-mounted camera identification information

CN121989988ACN 121989988 ACN121989988 ACN 121989988ACN-121989988-A

Abstract

The invention relates to the technical field of vehicle control optimization, in particular to a vehicle optimization control method and system based on vehicle-mounted camera identification information. The method comprises the following steps of collecting road scene images and vehicle running state information, extracting identification information such as lane lines, barriers, traffic signs and the like through a vehicle-mounted identification unit, fusing the identification information with the vehicle running state information to generate comprehensive state data, carrying out multi-frame constraint and correction based on identification confidence and motion continuity to obtain corrected comprehensive state data, predicting running trend, and outputting control instructions such as acceleration, deceleration, steering or braking and the like to realize optimal control. According to the vehicle-mounted camera identification information and the vehicle running state data are fused, a multi-frame confidence constraint and dynamic correction mechanism is constructed, and accurate prediction and intelligent control of the vehicle running state are realized, so that the identification stability, the running safety and the control smoothness are remarkably improved.

Inventors

SONG ZHILIANG
XU XIAOYU
Qiu Zhizhuo

Assignees

江西应用技术职业学院

Dates

Publication Date: 20260508
Application Date: 20260202

Claims (10)

1. The vehicle optimization control method based on the vehicle-mounted camera identification information is characterized by comprising the following steps of: S1, collecting road scene images and vehicle running state information; S2, inputting the road scene image into a vehicle-mounted recognition unit to extract the recognition information of the lane lines, the obstacles and the traffic signs, and performing fusion operation on the recognition information and the vehicle running state information through a vehicle-mounted control unit to form comprehensive state data; S3, performing multi-frame confidence constraint on the comprehensive state data based on the recognition result confidence of the recognition information and the vehicle motion continuity of the vehicle running state, and correcting the comprehensive state data according to the constraint result to obtain corrected comprehensive state data; And S4, predicting the running trend of the vehicle according to the corrected comprehensive state data, and generating corresponding control instructions, wherein the corresponding control instructions comprise at least one of acceleration, deceleration, steering or braking, and the corresponding control instructions are executed by a vehicle executing mechanism so as to realize optimal control.
2. The vehicle optimization control method based on the in-vehicle camera identification information according to claim 1, characterized in that step S2 includes the steps of: step S21, performing illumination equalization and distortion correction based on the acquired road scene image to generate road image data after vision correction; s22, extracting the characteristics of lane lines, barriers and traffic sign areas in the road image after vision correction, and obtaining target identification characteristic data; Step S23, carrying out category discrimination and position positioning on the target identification feature data through a vehicle-mounted identification unit to generate identification element data; and step S24, carrying out trend fitting and state consistency check processing on the vehicle motion trail by fusing the feature matrix data to generate comprehensive state data.
3. The vehicle optimization control method based on the in-vehicle camera identification information according to claim 2, wherein performing fusion calculation based on the identification element data and the vehicle running state information includes: performing time stamp alignment and synchronous interpolation on the vehicle running state information based on the identification element data to generate time sequence matching data; Performing projection mapping on the spatial distribution of the identification element data by using the time sequence matching data to generate spatial corresponding data; performing difference quantization on response deviation between the identification element data and the vehicle running state information through the space corresponding data to generate fusion deviation index data; Dynamic association weight distribution is carried out based on the fusion deviation index data, and weighted fusion coefficient data are generated; and carrying out feature superposition and matrix normalization on the identification element data and the vehicle running state information by using the weighted fusion coefficient data to generate fusion feature matrix data.
4. The vehicle optimization control method based on the vehicle-mounted camera identification information according to claim 2, wherein the vehicle motion trajectory acquisition method in step S24 includes: Acquiring a vehicle history track; extracting space position coordinates of the vehicle at continuous moments based on the fusion feature matrix data, and generating preliminary track point sequence data; Performing direction consistency calculation on displacement vectors of adjacent track points by using the preliminary track point sequence data to generate track direction sequence data; And judging the shape continuity of the historical track of the vehicle according to the track direction sequence data so as to obtain the motion track of the vehicle.
5. The vehicle optimization control method based on the in-vehicle camera identification information according to claim 3, wherein projecting the spatial distribution of the identification element data using the time-series matching data includes: extracting the space coordinate attribute and the time stamp attribute of each identification object based on the identification element data to generate identification element space-time characteristic data; performing time alignment on the time-space characteristic data of the identification elements according to the time sequence matching data to generate time-aligned identification element sequence data; performing space positioning check on the time-aligned identification element sequence data, and removing identification points with position drift or time abnormality to obtain an effective identification point set; Performing space-time correlation projection mapping based on the effective identification point set, and performing weighted projection on the spatial positions of the identification elements in the continuous frames according to the time sequence weight to generate mapping plane data; And carrying out region aggregation and connectivity correction on the spatial distribution of the identification element data by using the mapping plane data to obtain spatial corresponding data.
6. The vehicle optimization control method based on the in-vehicle camera identification information according to claim 1, wherein step S3 includes the steps of: s31, extracting the confidence coefficient of the recognition result of each frame of road scene image based on the recognition information, and generating frame-level confidence coefficient sequence data; S32, calculating the speed change rate and the steering angle change rate of road scene images between continuous frames according to the vehicle running state information to generate vehicle movement continuity data; Step S33, performing time synchronization and sequence matching on the frame-level confidence sequence data and the vehicle motion continuity data to generate multi-frame corresponding matching data; And step S34, executing dynamic confidence balance constraint based on the multi-frame corresponding matching data, and correcting the comprehensive state data according to the constraint result to obtain corrected comprehensive state data.
7. The vehicle optimization control method based on the in-vehicle camera identification information according to claim 6, characterized in that step S34 includes the steps of: Step S341, counting the change condition of the recognition confidence coefficient of each frame of road scene image in the multi-frame corresponding matching data to obtain inter-frame confidence fluctuation data; Step S342, comparing the variation trend of the interframe confidence fluctuation data and the variation trend of the vehicle movement continuity data, judging whether the confidence mutation is synchronous with the movement state variation, and obtaining a judging result; step S343, when the judging result is in the first change condition, carrying out adjacent frame mean value compensation on the frame-level confidence coefficient sequence data to obtain a first constraint result; and step S344, correcting the comprehensive state data according to the first constraint result and the second constraint result, and recalculating the state change difference value of the continuous frames to obtain corrected comprehensive state data.
8. The vehicle optimization control method based on the vehicle-mounted camera identification information according to claim 7, wherein the first change condition and the second change condition in step S343 specifically include: When the confidence average reduction amplitude of the recognition results of the continuous three-frame road scene images exceeds 20% of the average value and the change rate of the movement speed of the vehicle or the target is less than 0.5m/s, judging that the first change condition is achieved; And when the change rate of the movement speed of the two continuous frames of road scene images is larger than 1m/s and the confidence degree reduction amplitude is smaller than 10%, judging the second change condition.
9. The vehicle optimization control method based on the vehicle-mounted camera identification information according to claim 1, further comprising, after generating the corresponding control instruction in step S4: extracting running speed, steering angle and acceleration information of the vehicle at the current moment based on the corrected comprehensive state data; judging whether the motion state of the vehicle at the next moment deviates from a preset safety threshold according to the result of the operation trend prediction; When the judgment result is deviation, corresponding control target parameters are determined according to the deviation type, and control instructions matched with the control target parameters are generated, wherein the control instructions are used for adjusting a driving system, a braking system or a steering system of the vehicle.
10. A vehicle optimization control system based on in-vehicle camera identification information for executing the vehicle optimization control method based on in-vehicle camera identification information according to claim 1, the vehicle optimization control system based on in-vehicle camera identification information comprising: the acquisition module is used for acquiring road scene images and vehicle running state information; The vehicle-mounted control unit is used for carrying out fusion operation on the identification information and the vehicle running state information to form comprehensive state data; the control constraint module is used for carrying out multi-frame confidence constraint on the comprehensive state data based on the recognition result confidence coefficient of the recognition information and the vehicle motion continuity of the vehicle running state, and correcting the comprehensive state data according to the constraint result to obtain corrected comprehensive state data; and the control execution module is used for predicting the running trend of the vehicle according to the corrected comprehensive state data and generating corresponding control instructions, wherein the corresponding control instructions comprise at least one of acceleration, deceleration, steering or braking, and the control instructions are executed by the vehicle execution mechanism so as to realize optimal control.

Description

Vehicle optimization control method and system based on vehicle-mounted camera identification information Technical Field The invention relates to the technical field of vehicle control optimization, in particular to a vehicle optimization control method and system based on vehicle-mounted camera identification information. Background Traditional vehicle control methods rely primarily on vehicle sensors, such as speed sensors, accelerometers, and radars, to make control decisions by collecting vehicle motion state information. However, such methods have problems of limited perception range, poor environmental adaptability, and low prediction accuracy in complex road scenes. In recent years, vehicle-mounted cameras are widely used to acquire road scene images due to their high resolution perceptibility, and assist in vehicle control by recognizing lane lines, obstacles, and traffic signs. Some prior art attempts to fuse the camera identification information with the vehicle running state information so as to realize more accurate vehicle motion prediction and control, but in practical application, the following specific defects still exist, namely under the conditions of illumination change, rainy and foggy weather or shielding, the confidence coefficient of the camera identification result may be suddenly changed, but the existing method does not generally dynamically restrict the fluctuation of the confidence coefficient, so that instantaneous misjudgment is easily fed back to a control system directly, and the vehicle acceleration and deceleration or steering instructions generate misoperation. Disclosure of Invention Based on this, it is necessary to provide a vehicle optimization control method and system based on the identification information of the vehicle-mounted camera, so as to solve at least one of the above technical problems. In order to achieve the above purpose, a vehicle optimization control method based on vehicle-mounted camera identification information comprises the following steps: S1, collecting road scene images and vehicle running state information; S2, inputting the road scene image into a vehicle-mounted recognition unit to extract the recognition information of the lane lines, the obstacles and the traffic signs, and performing fusion operation on the recognition information and the vehicle running state information through a vehicle-mounted control unit to form comprehensive state data; S3, performing multi-frame confidence constraint on the comprehensive state data based on the recognition result confidence of the recognition information and the vehicle motion continuity of the vehicle running state, and correcting the comprehensive state data according to the constraint result to obtain corrected comprehensive state data; And S4, predicting the running trend of the vehicle according to the corrected comprehensive state data, and generating corresponding control instructions, wherein the corresponding control instructions comprise at least one of acceleration, deceleration, steering or braking, and the corresponding control instructions are executed by a vehicle executing mechanism so as to realize optimal control. In the present specification, there is provided a vehicle optimization control system based on vehicle-mounted camera identification information for executing the above-described vehicle optimization control method based on vehicle-mounted camera identification information, the vehicle optimization control system based on vehicle-mounted camera identification information including: the acquisition module is used for acquiring road scene images and vehicle running state information; The vehicle-mounted control unit is used for carrying out fusion operation on the identification information and the vehicle running state information to form comprehensive state data; the control constraint module is used for carrying out multi-frame confidence constraint on the comprehensive state data based on the recognition result confidence coefficient of the recognition information and the vehicle motion continuity of the vehicle running state, and correcting the comprehensive state data according to the constraint result to obtain corrected comprehensive state data; and the control execution module is used for predicting the running trend of the vehicle according to the corrected comprehensive state data and generating corresponding control instructions, wherein the corresponding control instructions comprise at least one of acceleration, deceleration, steering or braking, and the control instructions are executed by the vehicle execution mechanism so as to realize optimal control. The intelligent correction and control optimization method has the beneficial effects that the intelligent correction and control optimization of the running state of the vehicle is realized through multi-level data processing and dynamic confidence constraint. The method comprises the steps of firstly en