Stereoscopic Vision System

Given two images taken from different camera orientations, it is possible to determine the depths of objects within the scene. This is the same process the human brain uses to provide depth perception from one's eyes.

First, a disparity map is generated, which is essentially an injective mapping between the pixels of the two images. Given information about the orientation of the cameras, a depth map can be generated, which is the prerequisite for creating a 3D mesh representation of the scene.

The Stereo Vision System is capable of automating this process. Additionally, an intuitive GUI is provided to enable the common user to process stereo image pairs at home using off-the-shelf hardware (e.g. webcams).

Tags

api framework library robotics vision