Estimation of baseline of single camera stereo vision based on an inspiration from SFF

Computation Stereo Vision is a widely researched technique in the field of computer vision for scene reconstruction. One of the main issues to be addressed in stereo vision is the trade-off that needs to be achieved between accuracy and resolution. A wide-baseline offers better resolution in depth estimated, contrarily a narrow baseline though offers good accuracy but suffers from poor depth resolution. The proposed work in this paper presents a variable baseline stereo vision system which permits a knowledge and control of the depth resolution for a range of geometries that the system can handle. This is achieved using an algorithm inspired from Shape From Focus (SFF) technique in computer vision. In the current research work a new SFF-inspired algorithm is developed which utilizes images acquired with low focal length lenses in place of a telecentric lens. Based on the sparse and coarse depth map obtained an approach for determining the baseline of a single camera based stereo vision system for any desired depth resolution is presented in this paper.