A machine-vision-based system for PPT remote control

It is popular to give a presentation by Microsoft Power Point, using computer and digital projector. In this paper, a kind of new software for PPT remote control is proposed and realized. The hardware configurations are a PC camera and a common laser pointer. The software identifies the intention of the user by the recognition of the trails of the laser spots through machine vision techniques and then controls the PPT paging. During the implement of the software, some new methods were proposed. We generated a brightest background image for the pointer trail extraction, instead that the literatures often used means of several images. We defined sub window to improve the adaptability and decrease the time complexity of the processing. We chose the first pixel of the laser spot encountered instead of the center of the spot to avoid the complex position computation. We provided a fine neural network structure for the classification of the trails of the laser spot. The application of the system showed that it has the advantages of low hardware cost, convenient installation, good portability, and fine function scalability.