Super-Resolution Reconstruction And Its Application In Gaze Tracking System

Image resolution is not only a measure of the ability of an imaging system to distinguish the details of an image but also shows the fineness of details of the image. High resolution images can offer more details that may be critical in various applications. Therefore, with the development of multimedia communication and information processing, there is high demand for high resolution. However, due to the limitations of the imaging system, images obtained in reality are usually degraded by blurring, noise and down-sample, thus with limited resolution. Improving the performance of the hardware is clearly one way to increase the resolution of the acquired images. But this method may not be feasible for the high cost. Therefore, the way by means of digital signal processing software technique to obtain high resolution images is becoming more significant. One of these techniques is Super-Resolution (SR) reconstruction. At present, SR technique has been widely used in military remote sensing, astronomy, public safety, medical imaging, computer vision and other fields.SR reconstruction technique is the process of combing multi-frame low resolution images of the same scene to reconstruct a high resolution image. It makes full use of the complementarity between the multi-frame image information, and make up the deficiency of the process of image acquisition and transmission which overcomes the constraint of hardware imaging system. Super-Resolution reconstruction and its application in eye tracking system are discussed in this dissertation. In general, the main ideas are included as follows:First, the imaging process of the low resolution images is introduced and the corresponding mathematical model is also built. On this basis, the dissertation summarizes the classical algorithms of SR reconstruction technique, and focuses the attention on their basic principles, theory evidences, implementation procedures, the superiorities and deficiencies. Secondly, a super-resolution method based on gradient-based adaptive interpolation is proposed. We utilize the frequency domain registration algorithm to estimate the motions of the low resolution images and map the low resolution images to the uniform high resolution grid, and then the gradient-based adaptive interpolation is used to form a high resolution image. Finally, wiener filter is applied to reduce the effects of blurring and noise. This method not only considers the distance between the interpolated pixel and the neighboring valid pixel, but also takes into account the local gradient of the original image, which is robust to the registration error and has low computational complexity.Thirdly, a novel infrared gaze tracking system based on non-uniform interpolation is proposed. The application of Super-Resolution idea in gaze tracking system is an innovative point in this dissertation. Select some different line points on the screen as calibration points and record their actual coordinates on the screen. The integral projection algorithm and canny edge detection are applied to extract the pupil boundary points from the captured eye images, and then the pupil center is computed using an efficient and accurate ellipse fitting algorithm. Finally, in order to estimate where the user looks, a novel mapping method based on the non-uniform interpolation algorithm is used. In this mapping method, the complicated geometric eyeball model and the nonlinear mapping between the pupil center coordinates and computer monitor screen coordinates do not need to be taken into account. Moreover, our system is the remote eye gaze tracking system. The users do not need to wear any device, which make the users feel more comfortable.Finally, the problems to be solved related to SR reconstruction and future research topics about application of SR reconstruction in eye tracking system are summarized. Furthermore, the prospect of the development and application tendency is analyzed as well.