Building a three-dimensional picture of space based on the video of the system of stereo pairs

摘要

Machine vision one of the most important applies of artificial intelligence. The machine vision applications are diverse. It covers a range of activities in automated systems. Currently, machine vision systems use additional sensors to build a proper three-dimensional picture of space, such as IR sensors of Kinect or Leap Motion. This approach solves a specific set of tasks, such as motion capture or gesture recognition, and is not valid in the general cases. This article aims the method of obtaining distances to custom objects from the video channels. The main goal of this article is building a three-dimensional picture of space based on the video of the system of stereo pairs.