Automatic generation of second image from an image for stereovision

Stereovision is the process of finding depth when multiple images of a scene are known from distinct positions. It is the basic phenomena in the construction of 3D from multiple images. Stereo vision is useful in many applications such as robotics, tracking object in 3D space and constructing a 3D model of a scene. Vision based remote control system exploits stereo vision to control the machine in a touch-free environment. Most of the cameras are monocular in nature. In our work, we have developed a generic model wherein, second image is automatically generated and can be used for 3D reconstruction from a single image. Relationship between depth and disparity for any focal length has been established. Exhaustive study on literature has indicated that no method exists to generate second image from the first for stereovision. Experiments are carried out to validate the proposed model where we got higher accuracy. The model has been tested with middlebury stereo dataset. With this model, the existing monocular cameras are sufficient to build 3D view of any scene using a single 2D image.