CASIA OpenIR  > 智能感知与计算
FastLCD: Fast Label Coordinate Descent for the Efficient Optimization of 2D Label MRFs
Liu KW(刘康伟); Zhang JG(张俊格); Yang PP(杨沛沛); Huang KQ(黄凯奇); Huang KQ(黄凯奇)
Conference NameInternational Joint Conference on Artificial Intelligence
Source PublicationInternational Joint Conference on Artificial Intelligence
Conference Date2016
Conference Place美国
Recently, MRFs with two-dimensional (2D) labels have proved useful to many applications, such as image matching and optical flow estimation. Due to the huge 2D label set in these problems, existing optimization algorithms tend to be slow for the
inference of 2D label MRFs, and this greatly limits the practical use of 2D label MRFs. To solve the problem, this paper presents an efficient algorithm, named FastLCD. Unlike previous popular movemaking algorithms (e.g., α-expansion) that visit all the labels exhaustively in each step, FastLCD optimizes the 2D label MRFs by performing label
coordinate descents alternately in horizontal, vertical and diagonal directions, and by this way, it does not need to visit all the labels exhaustively. FastLCD greatly reduces the search space of the label set and benefits from a lower time complexity. Experimental results show that FastLCD is much faster, while it still yields high quality results.
Other Abstract
 随机场结构模型是计算机视觉任务中一个非常重要的工具。随机场结构模型在很多物体识别任务中都有广泛的应用。作为一个特例,带有二维标号的随机场(2D label MRF)模型在很多视觉任务中也有不可替代的应用,如变形物体匹配任务,光流估计,图像分类,物体检测等任务。在这些任务中,结构模型的推理速度往往对算法的广泛应用是至关重要的,但是由于二维标号随机场模型中的标号空间非常大,传统的优化算法在推理这类随机场模型时推理速度往往非常慢。在本文中,我们主要解决这类二维标号随机场模型的快速推理问题,并提出一个基于标号坐标梯度下降(Label coordinate descents)的快速推理算法,我们称之为 FastLCD 算法。FastLCD 算法通过在标号空间的水平、垂直和对角方向上执行标号坐标梯度下降来对二维标号随机场模型进行快速推理优化。  与之前基于标号移动的算法(如 $\alpha$-expansion)需要详尽地遍历标号空间所有标号不同,FastLCD 算法利用了标号集合的二维空间结构信息,并且限制随机场中的节点只能在标号空间沿一个特定方向进行移动。通过这种方式,FastLCD 算法不需要遍历二维标号空间中的所有标号,假定二维标号集合的大小为$m\times n$,FastLCD 算法在每次迭代优化过程中得到一个更小的 $O(m+n)$ 时间复杂度,而相对地,$\alpha$-expansion 算法的时间复杂度为 $O(m*n)$。我们同时在模拟数据和真实的视觉任务上对 FastLCD 算法进行评价。实验结果表明 FastLCD 算法明显地提高了传统算法的推理速度,并且没有太大的质量损失,从而得到有竞争力的推理效果。
Keyword马尔科夫随机场 标号坐标梯度下降
Indexed ByEI
Document Type会议论文
Corresponding AuthorHuang KQ(黄凯奇)
First Author AffilicationInstitute of Automation, Chinese Academy of Sciences
Recommended Citation
GB/T 7714
Liu KW,Zhang JG,Yang PP,et al. FastLCD: Fast Label Coordinate Descent for the Efficient Optimization of 2D Label MRFs[C],2016.
Files in This Item: Download All
File Name/Size DocType Version Access License
ijcai16.pdf(448KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Liu KW(刘康伟)]'s Articles
[Zhang JG(张俊格)]'s Articles
[Yang PP(杨沛沛)]'s Articles
Baidu academic
Similar articles in Baidu academic
[Liu KW(刘康伟)]'s Articles
[Zhang JG(张俊格)]'s Articles
[Yang PP(杨沛沛)]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Liu KW(刘康伟)]'s Articles
[Zhang JG(张俊格)]'s Articles
[Yang PP(杨沛沛)]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: ijcai16.pdf
Format: Adobe PDF
This file does not support browsing at this time
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.