Visual odometry: feature based tracking and velocity estimation based on ground looking camera
Permanent address of the item is
The computer vision field has close relation with autonomous robotics which is in use for several purposes which include mobile robot vision tasks, robot navigation, motion trajectory, object detection and tracking and security surveillance tasks. There are many techniques available which are being used for completing these tasks but many of them lead to problems like low resolution, limited applicability and high capital investment. This thesis examines the way in which we could achieve results within some tolerance level of accuracy and low cost. Visual odometery is one of the main objectives of this thesis which is achieved with simple and practical method. The algorithm is developed to estimate the velocity using a corner detection technique based on ground looking camera. The thesis is divided into three main parts. In the first part of the thesis, literature review and previous work done in the relevant field is explained. The theoretical background of the topic is also described in the first part of the thesis. Second part of the thesis demonstrates the development of the algorithm, pre and post processing and implementation of the algorithm. Last part of the thesis describes the different test environments where the developed algorithm is implemented. The test environments are further classified into two main categories. Conclusions, results, problems faced during the whole process and future tasks are also included in the last part of the thesis. The study indicates that, selection of the right pre-processing parameters can enhance the results quality. At the same time by providing the appropriate illumination for the camera system can also increase the efficiency of the outcome. This research and developed algorithm has the potential to be used for further implementation at commercial level by changing some necessary parameters in the algorithm and implementation. This research could be more useful by implementing addressed future tasks in Section 5.3.2, in order to achieve higher efficiency in the results. Implementation of all necessary parameters explained in this thesis and by considering future tasks will make this research more effective and beneficial for the business.