Eye Tracking Guided Real Time 3D Presentation Of Stitched Light Field Images On 3D TV

Seeing images in 3D has more visual appeal to the eyes than seeing it in 20,since it mimic our way of seeing the world around us.An important factor of visual perception is sense of depth.An approach to light field image stitching and real time 3D display of images on 3D TV is proposed in this thesis.The stitched light field images(sub-aperture images)will be rendered based on the eye movement of an observer in real time.This is a low cost solution for physically challenged person to interact with a computer.This system can be further used for visual perception related experiments.The system we propose is easy to setup and use.Moreover it is very convenient to interact with a computer without much effort.Light field image stitching is done is sub-aperture image level rather than directly working on 4D light field data.This method substantially reduces the stitching time compared to other methods.We followed conventional feature based stitching method.Speeded Up Robust Features(SURF)is used for feature detection,due to its speed and robustness.Lytro camera is being used to capture light field images.The wide viewing range of a scene is captured by taking multiple photos of a scene.Conventional image stitching on sub-aperture level image is much faster than light field image stitching techniques.Display of three dimensional image is achieved by projecting two different images of a scene side by side on to a 3D TV,which require 3D glasses for viewing.The 3D visual perception is more appealing than the 3D effect achieved by single light field image.The eye movement based mechanism(EyeTribe eye tracking device is used)to navigate the scene is experimented.The idea is to change scene perspective based on user eye movement,and the images are displayed in real time using GPU shaders.