An ROI Optimization Method Based on Dynamic Estimation Adjustment Model

An important research direction in the field of traffic light recognition autonomous systems is to accurately obtain region interest (ROI) image through multi-sensor assisted method. Dynamic evaluation performance (GNSS, IMU, and odometer) fusion positioning system optimum size ROI essential for further improvement accuracy. In this paper, we propose a dynamic estimation adjustment (DEA) model construction method optimize ROI. First, according residual variance integrated navigation vehicle velocity, divide innovation into an approximate Gaussian fitting (AGFR) convergence (GCR) estimate them using variational Bayesian gated recurrent unit (VBGRU) networks mixture (GMM), respectively, GNSS measurement uncertainty. Then, relationship between uncertainty aided acquisition error deduced analyzed detail. Further, build convert optimal lights online. Finally, use YOLOv4 detect recognize Based on laboratory simulation real road tests, verify DEA model. The experimental results show that proposed algorithm more suitable application vehicles complex urban scenarios than existing achievements.