Lista przedmiotów z materiałami udostępnionymi dla studentów

Dla_studentów
  • Increase font size
  • Default font size
  • Decrease font size

Bożena Karbowiak


Head pose estimation from static images


Opiekun pracy dyplomowej: prof. dr hab. inż. Paweł Strumiłło
Dodatkowy opiekun pracy dyplomowej: dr inż. Aleksandra Królak

Praca dyplomowa ife obroniona 2006-07-25
Streszczenie pracy dyplomowej:
The main goal of the thesis is to work out the computer algorithm for automatic finding of the head rotation angle. The developed algorithm works on static images of the mannequin’s head. Its task is to perform image processing and give possibly good results in a so called “blue box”. In order to achieve the goal, the basic knowledge of image processing methods and head detection techniques is necessary. The algorithm can be used e.g. in a system for car driver monitoring. The research in the area of automatic head pose estimation can be divided into two main categories: systems employing single camera images, and stereovision-based systems which use the images from two cameras. In this project the images coming from single digital camera are employed for the purpose of the estimation of the pose of the mannequin’s head. The algorithm uses two images: the first one that is treated as the base (with no head rotation, '), and the second one that represents some head rotation. The system works under the assumption that the rotation is performed around single axis only and there is no head translation. The human head is assumed to be a rigid body. There are several points on the face that are independent on the head pose and the mimics, i.e. eye corners and nose tip. The developed algorithm uses only the nose position coordinates for head rotation estimation. The distances between the nose position in the base image and its position in the rotated image are found. The changing distances are directly dependent on the rotation angles, so finding the mathematical functions