Метод рир-проекции в подсистеме визуализации тренажерно-обучающей системы

Nowadays in many industry sectors (for example, in aerospace industry) there is a shortage of qualified specialists, capable of controlling complex technical systems. There is a need of a massive increase of training centers efficiency, particularly with adaption of new training methods and developing of more effective training complexes. Using multimedia technologies in training-simulation systems development allows creating training materials data banks that include images, texts accompanied by audio, video and visual effects, interactive interface; provide an effective way of using collected information for distant and distributed training. One way of using multimedia data in training simulation systems is to insert a graphical representation of instructor into virtual environment. In order to correctly visualize an image of instructor in a 3D virtual scene a method of rear-projection has been developed. It is based on 3D keying. Its main function is to separate an object from a uniform background. This process can be described as a process of creating a mask that contains information about image translucency that separates an object from other parts of an image. Any keying method is resource-intensive, so processing of large images on CPU can lead to performance problems of visualization subsystem, while one of the requirements for visualization subsystem is real-time operation. Considering this requirement, a realization of keying algorithm uses resources of the video card to process images.