by Heval Can Aslan ÖZEN and Ömer Faruk KARABOSTAN
Advisor: Assist. Prof. Dr. Burak KALECİ
A Report Presented in Partial Fulfillment of the Requirements for the Degree Bachelor of Science in Electrical Electronics Engineering
ESKISEHIR OSMANGAZI UNIVERSITY
JANUARY 2022
The closed-door detection problem has been a favored research area in different domains such as robotics and building information modeling. The main reason for this is that door locations separate important parts of the environment, such as rooms and corridors. Previous studies that focused on closed-door detection problems generally employed visual data. However, the success of those studies depends on the lighting conditions of the environments. In addition, the distance and angle of the door according to the camera position significantly affect the performance of the studies.
In this study, we utilized point cloud data to overcome these drawbacks of visual data and exploit the ability to describe the 3D characteristics of scenes of the point cloud data. We proposed a rule-based approach to identify closed doors and determine door positions. The rules were extracted regarding the relationship between walls and hinged doors. The experiments were conducted with the ESOGU DOORS dataset to analyze the effectiveness of the proposed method. The test results showed that the door detection rate was 95.93%.
We would like to express our gratitude to our supervisor, Asst. Prof. Dr. Burak KALECİ, for his valuable comments and ongoing support.