Image Dense Stereo Matching by Technique of Region Growing
Tang Li · Chinese Journal of Computers · 2004
A new dense stereo matching algorithm is presented based on technique of region growing, which can be applied to a wide range of image pairs including those with large disparity or without rectification even if part of the images are less textured. Firstly, a number of feature points in the image pair are matched using a new two-level matching method. Then these reliably matched points are taken as the seeds from which corresponding relations propagate towards other regions of the images under two strategies. Weighted SSD (Sum of Squared Difference) is developed as the target function. The size of the template window is adaptive with textures within it and the size of the search window is varied in inverse proportion to the confidence coefficient. The algorithm has been tested with real stereo images and the results demonstrate its accuracy and efficiency.