الفهرس 
Chapter 1 IntroductionOnly 14 pages are availabe for public view
 |  | from | 125 |  |  |
| | | from | 125 | | |
Abstract
According to the world Health Organization (WHO), 490 million people were estimated to be visually impaired worldwide, of whom 45 million are totally blind. since loss of vision means loss of independence, the most critical problem the visually impaired face is the navigation problem. Specifically, it is very challenging for the visually impaired people to navigate safely without a help from someone or something .
One of the most used approaches for robust path detection is the RANdom SAmple Consensus (RANSAC), which is a global iterative approach for estimating the parameters of a certain model from input data points contaminated by a set of noisy data. Inappropriately, the standard RANSAC suffers from drawbacks regarding the processing time , accuracy of fitting data, and acquiring an optimal solution . This study proposes three different approaches in favor of working out these difficulties .
The first proposed approach, Fast RANdom Sample consensus (FRANSAC), is assessed for obstacle detection in indoor environments.
FRANSAC involves three three main steps: data preprocessing and 3D enhanced RANSAC. First, range data, obtained from 3D camera, is converted into 3D point clouds. Next, a preprocessing stage is introduced where a pass-through and voxel grid filters are applied. Last, planes are estimated using a modified 3D RANSAC. Significantly, the experimental results demonstrate that our approach can segment planes and detect obstacles about 7 times faster than the standard RANSAC without losing its discriminative power.
The second proposed approach, Accurate RANSAC (ARANSAC), is investigated for more accurate navigational path segmentation results.
ARANSAC is composed of three main stages: data input, normal estimation, and ground plane estimation using normal and 3D RANSAC.
By combining normal with 3D enhanced RANSAC, the results become more accurate. Accordingly, different evaluation metrics were enhanced however the processing speed is slightly affected.
The third proposed approach, fast and Accurate RANSAC (FRANSAC),helps the visually impaired navigate in a fast, reliable, and independent way. Using RGB-D scanners, fast and accurate RANSAC is proposed to eliminate the common RANSAC disadvantages.
FRANSAC consists of three main stages: data preprocessing, ground plan segmentation and object detection. Markedly, two sets of experiments have been performed using two dataset and real-world data, which show accuracy (99.9%) and speed (21 frames per second) of FRANSAC.
Furthermore, an innovative obstacles detection navigational aid for the visually impaired, Sensify, based on a 3D active sensor is proposed.
Experiments have shown that the obstacle detection algorithm is fast, effective and accurate. Particularly, the correct detection rate obstacles is 95% in average, which is obtained using the consecutive range data obtained from a 3 D active camera .