Research On Dynamic Object Detection In Garage Environment Based On 77GHz MMW Radar

The personal and property safety issues in the garage environment have always been the focus of attention.Garage safety monitoring system is mainly based on video surveillance,but it can no longer be insufficient with images data alone.For this reason,data fusion has become a current study hotspot,but none of the existing studies is aimed at the garage security environment.Therefore,this thesis conducts the following studies on dynamic object detection in the garage environment based on 77GHz MMW Radar(millimeter wave radar)and vision camera:First,this thesis designs a denoising and target recognition algorithm for the data collected by 77GHz MMW Radar in the garage environment.The algorithm improves the radar’s recognition rate of dynamic objects significantly;Second,this thesis designs a data fusion method for radar and image data.The method involves the improvement of Zhang Zhengyou’s calibration method.The fusion method realizes the multi-sensor cooperation of radar and image data;Third,this thesis first proposes an image fusion dataset,XMU Cardata,with radar information in a garage environment,and uses deep learning networks Faster R-CNN and YOLO v2 to study the dataset.This dataset can be used for deep learning,data fusion,vehicle detection and so on;Fourth,this thesis first proposes a 77GHz MMW Radar data preprocessing system.The system provides radar data conversion functions,radar data reproduction functions and fusion data display functions,which greatly facilitates subsequent researchers.