Obstacle detection and classification system for visually impaired people using smart glasses and smartphones | |
| Author | Perera, Thammitage Jude Delani Rushanka |
| Call Number | AIT Thesis no.ICT-17-02 |
| Subject(s) | Mobile apps--Development Imaging systems Detectors |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Information and Communication Technology, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. ICT-17-02 |
| Abstract | Vision is the most important sense of human beings as four fifths of all the impressions on the senses come from the eyes. Deterioration of the vision leads to dramatic reduction of the mobility by limiting the person to his sense of touch and hearing. According to the estimations of World Health Organization (WHO) the visually impaired population is around 285 million all around the world. They face countless difficulties mainly when walking in an unfamiliar indoor environment due to lack of prior information. This research aim is to improve the indoor mobility of visually impaired people. This system is based on an Android smart phone due to rapid increase of smart-phone usage among visually impaired people and to reduce the requirement of additional hardware. The main objective of this research is to provide useful information in real time to make appropriate and fast decision making for a safety navigation in indoor space by studying about the static and dynamic object behavior and propose a suitable alert method. Secondly, to propose a reliable application for visually impaired people. In this work, there are two main modules, first module is static object recognition which is based on feature detection, extraction and use Support Vector Machine (SVM) classification algorithm to classify and recognize the object. Second module is to detect the moving objects and alert when obstacle is getting closer to the user. The proposed system works in real-time video with audio guidance and this system occupies only a smart glass and an Android based smart phone. The proposed system is achieved 98% precision in obstacle detection with a imbalance data-set The dynamic obstacle detection is done by using a cascade classifier where the system could achieve only 72% precision and the main reason is occlusion due to the movement of the camera and the object. |
| Year | 2017 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. ICT-17-02 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Information and Communication Technology (ICT) |
| Chairperson(s) | Attaphongse Taparugssanagorn; |
| Examination Committee(s) | Dailey, Matthew N.;Mongkol Ekpanyapong; |
| Scholarship Donor(s) | Asian Institute of Technology Fellowship; |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2017 |