Zooming Assisted Stereo Matching

One essential problem of sensing technologies is to measure the distance of an object, and stereo vision is one of the most popular approaches. In this chapter, stereo matching with the assistance of zooming information is investigated. The proposed method adopts two zoom lens cameras to acquire several image pairs using various focal length settings. With the zooming image pairs recorded by the same camera, we are able to increase the correctness of stereo matching results. Moreover, a novel zoom rectification paradigm is developed to simplify the zoom vector computation. In the experiments, the bad pixel rate is used for the comparison of the results derived from real scene images based on conventional stereo techniques and the proposed zooming assisted stereo marching algorithm. Our technique is also tested on real images captured with optical zoom changes to demonstrate its effectiveness. The results have shown that the proposed approach is able to reduce the disparity bad pixel rate using both local and global stereo matching algorithms.