Vision-based virtual mouse system
This thesis describes a virtual mouse system that utilizes machine vision technology to enable a user to navigate through software window screens on a standard personal computer, access and control software, and/or peripheral devices attached to the personal computer. Utilizing the virtual mouse, the user can execute the standard control functions of the system mouse required to perform software application selection functions and navigation tasks without physical interface with the controlled computer. The machine vision algorithm utilizes a set of four simulated photoelectric sensors, with Gaussian filter sensitivities, to compute and track the mouse location. By utilizing Gaussian filters, instead of traditional image processing, which could consists of segmentation followed by morphological operations for feature extraction, the proposed virtual mouse can reduce the time it takes the PC to adjust the display of the cursor on the screen. At a minimum, the proposed system will perform at least equal to or better than similar systems with virtual mice implemented with the traditional image processing front end.
Watiti, Tom Wanjala, "Vision-based virtual mouse system" (2009). ETD Collection for University of Texas, El Paso. AAI1473902.