Influence Of Vehicle Dynamics On Vehicle Vision System And Its Countermeasures

As one of the "New Four Modernizations" in automobile industry,intellectualization is widely considered as the inevitable trend of automobile system development.Camera is one of the most important sensing components in intelligent vehicle environment sensing system.The stability of the video collected by camera has a great impact on the subsequent image processing process.When the vehicle is driving on some extreme road conditions,such as continuous speed bumps and large road grooves,the irregular vibration of the vehicle body leads to the jitter of the camera rigidly connected with the vehicle body,which further causes the irregular jitter of the image sequence collected by the camera,that is,the blur of the video.This paper explains this phenomenon from the perspective of vehicle dynamics for the first time,and establishes a model based on vehicle dynamics platform to analyze the above problems on the basis of determining relevant evaluation indexes.At the solution level,inspired by the existing handheld device anti shake countermeasures,this paper studies from two aspects: mechanical active image stabilization and electronic image stabilization1.Study the influence of vehicle dynamics on vehicle vision system.Based on the carmaker platform,a vehicle dynamics simulation model including vehicle,road surface and traffic scene is built,which can collect video of vehicle camera based on this model.Before the simulation analysis,the above-mentioned impact evaluation index is determined as peak signal to noise ratio.Then,the influence of common vehicle dynamics factors,including bump amplitude,suspension system parameters,pitch motion and vehicle speed,on the visual system is studied.In the process of analyzing the above factors,we pay attention to the change of the peak signal-to-noise ratio to measure the effect of the subsequent solutions.According to the analysis results in this paper,if the relevant solutions can improve the peak signal-to-noise ratio by 1-2d B,the performance of the countermeasure is good.2.The active image stabilization of machinery based on preview information is studied.Considering the consolidation relationship between the vehicle body and the camera,the active suspension is regarded as the first stage image stabilizing device of the camera,and the active control of the suspension is carried out based on the model predictive control method.Under the simulation condition,the comprehensive vibration reduction performance of vehicle suspension is 23.63% higher than that of passive suspension when the sampling number and sampling period of the controller are 20 and 0.01 s respectively.On this basis,this paper focuses on the selection of controller parameters,and analyzes the performance of the suspension when there are controller errors in vehicle model parameters,actuator performance,preview information and state perception in practical engineering application.Finally,the virtual experiment based on carmaker platform is carried out to verify the suspension control effect and the image stabilization effect of the control suspension.According to the simulation results,peak signal to noise ratio of the image can be increased by 1.31 d B after the mechanical active image stabilization is carried out.3.The electronic image stabilization algorithm which meets the needs of vehicle is studied.This paper focuses on the application of electronic image stabilization to solve the inter frame fuzzy problem in video blur.In this paper,we choose the image stabilization technology based on feature point detection and matching,and based on ORB algorithm to meet the real-time requirements.Furthermore,in order to improve the matching accuracy of feature points,the traditional random sample consensus algorithm is improved.In order to adapt to the various jitter conditions in vehicle scene,adaptive Kalman filter is used to solve the problem of classical Kalman filter sensitive to initial value.In order to verify the performance of electronic image stabilization,a gasoline model vehicle with remarkable vibration characteristics is modified to carry out the experiment.After the electronic image stabilization processing proposed in this paper,peak signal to noise ratio of video is increased by 1.26 d B on average,which proves the positive performance of the system.