Tvorba 3D modelu řidiče na základě hloubkových map

Abstract

The diploma thesis deals with reconstruction of model of driver from single depth image using RGB-D sensor Microsoft Kinect second generation. The aim of this work is to recognise parts of a human body and interpret those as a model of driver. The model is computed based on analysis of input data and parts of a human body are marked based on this analysis. The analysis includes a detection of a torso or more precisely abdoman area from which paths to limbs and head are searched. These are claimed as a model of driver. The designed analyzer doesn't use pre-trained classifier of poses and it runs over single images.