Content area
Abstract
Zoom tracking has become a demanded feature in digital still cameras during the zooming operation. Zoom tracking provides improved video quality by maintaining image sharpness while users zoom in or out during different modes of camera operation including movie capture and live preview. It involves the automatic adjustment of the focus motor in response to the zoom motor movements for the purpose of keeping an object of interest in focus. Zoom tracking is normally achieved by moving the focus motor according to so-called trace curves in response to changes in the zoom motor position. A trace curve denotes focus motor positions versus zoom motor positions for a specific object distance. Thus, any zoom tracking approach is characterized by the way these trace curves are estimated and followed. The major drawback with the existing zoom tracking approaches is that they trade off tracking speed for tracking accuracy or vice-versa due to the one-to-many mapping problem in following trace curves. The goal in this research effort has been the development of zoom tracking approaches to overcome this limitation of the existing approaches. More specifically, two approaches, named relational and predictive zoom tracking, have been introduced. In addition, this dissertation addresses the real-time implementation issues in zoom tracking and provides a real-time working zoom tracking implementation on an actual digital camera platform. Different indoor and outdoor scenes captured under varying lighting conditions with different object distances have been used to compare the performance of the developed approaches with the existing ones. The performance results indicate that the developed zoom tracking solutions achieve better tracking accuracy than the existing ones without adding to tracking speed. The outcome of this work is general purpose in the sense that the developed solutions can be utilized on any digital camera platform.
