RADAR-VISION FUSION FOR VEHICLE DETECTION BY MEANS OF IMPROVEDHAAR-LIKE FEATURE AND ADABOOST APPROACH

摘要

This work describes a vehicle detection system that usesrnfusion of vision and radar data. The radar provides a firstrnestimation of the lateral position of vehicle candidates andrnthe related distance information. This information is used torndefine a region of interest (ROI) that is subject to verification.rnA video camera is used for the verification purpose. The projectionrnof the ROI onto the image plane is scanned via anrnAdaBoost object detection algorithm, and thus radar detectionrncan be verified and more specific data of the vehicle’srn3D position and width can be given.rnMoreover, the distance information provided by radar isrnused to choose optimal parameters during the visual detectionrnprocess, e.g. properties of the scan window and parametersrnfor fusing detections.rnIn addition, mutual information for haar-like feature selectionrnis used to increase detection rates.