A Kalman filter approach to direct depth estimation incorporating surface structure

被引：42
作者：
Hung, YS [1 ]
Ho, HT [1 ]
机构：
[1] Univ Hong Kong, Dept Elect & Elect Engn, Pokfulam Rd, Hong Kong, Peoples R China
来源：
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 1999年 / 21卷 / 06期
关键词：
depth-from-motion; Kalman filter; gradient method; surface structure; image sequence;
D O I：
10.1109/34.771330
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
The problem of depth-from-motion using a monocular image sequence is considered. A pixel-based model is developed for direct depth estimation within a Kalman filtering framework. A method is proposed for incorporating local surface structure into the Kalman filter. Experimental results are provided to illustrate the effect of structural information on depth estimation.
引用
收藏
页码：570 / 575
页数：6
相关论文
共 18 条
12