Fast Local Stereo Matching via DAISY Descriptor and Modified Weight Kernel
Feng Liu · Journal of Nanjing University of Posts and Telecommunications · 2012
A fast stereo matching based on DAISY feature descriptor and modified weight kernel is proposed to eliminate the ambiguity of binocular stereo problem.Firstly,the local DAISY feature descriptors of both stereo pairs are constructed fast and densely for initial matching costs being calculated from the features;two-pass aggregation with Epanechnikov weight kernel for the reliable costs are applied to resolve ambiguity of matching feature similarities;each pixel's initial disparity is obtained via Winner-Takes-All optimization from them.Secondly,in order to improve the quality of disparity map,we adopt sequentially the refining procedures with modified bilateral filtering,symmetric consistency check and multi-directional weighted disparity extrapolation.The experiments indicate that this technique with concise structure and low complexity can improve effectively the matching accuracy and obtain comparably accurate and piecewise smooth dense disparity map.