Light field cameras are cameras that capture both the angular and spatial information of rays in a scene in a single shot. This means that the light field can be used to calculate depth information of a scene.
Current state-of-the-art light field depth calculation techniques are slow and computationally expensive and so are not suitable for most robotics applications.
Machine learning approaches have been used in many areas of computer vision to increase the speed and robustness of various algorithms with much success, however their use on light fields is still very limited.
The activities to be undertaken during this project will include, but are not limited to:
- creating data to be used in training and testing machine learning algorithms for light fields
- designing a machine learning algorithm for generating depth fields from light fields
The outcomes of this project should be a machine learning algorithm that has been trained to generate depth fields from light fields.
The algorithm should be compared to current state-of-the-art light field depth calculation techniques in terms of speed, accuracy and robustness.
Skills and experience
The student should have the following skills and experience:
- programming with a background in MATLAB and Python
- machine learning
Contact the supervisor for more information.