Robust Dense Disparity Estimation Using Hierarchical Adaptive Curve Matching
VISION MODELING, AND VISUALIZATION 2002, PROCEEDINGS(2002)
摘要
In this paper, a novel area-based dense stereo correspondence estimation algorithm is proposed. The new algorithm which we call Hierarchical Adaptive Curve Matching or HACM in short, is suitable for stereo imaging of both static and dynamic scenes. The core to this method is the derivation of a unique dense adaptive curve representation for each pixel, in the pair of 2D images, and the curve is considered to characterize the changes in shape, texture direction, and the luminance properties in its neighboring area of the pixel. The dense matching is then carried out in a hierarchical manner that incorporates several ideas to improve its efficiency and robustness. Extensive experiments, conducted on both synthetic, real stereo pairs, and real video streams, reveal the favorable performance of this new algorithm.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络