arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1411.6340cs.CV
收藏

非凸先验多标号MRF的迭代加权图割

Iteratively Reweighted Graph Cut for Multi-label MRFs with Non-convex Priors

Australian National University(澳大利亚国立大学);NICTA(澳大利亚国家信息通信技术卓越中心)

Thalaiyasingam Ajanthan, Richard Hartley, Mathieu Salzmann, Hongdong Li

摘要:

虽然在计算机视觉中被广泛认为是高度有效的,但具有非凸先验的多标签MRF难以优化。为了解决这个问题,我们引入了一个算法,迭代近似的原始能量与适当加权的替代能源,更容易最小化。我们的算法保证了原始能量在每次迭代时都在下降。特别是,我们考虑的情况下,全球最小的加权代理能量可以通过多标签图切割算法,并表明我们的算法,然后让我们处理各种各样的非凸先验。我们证明了我们的方法的好处,超过国家的最先进的MRF能量最小化技术的立体声和修复问题。

英文摘要:

While widely acknowledged as highly effective in computer vision, multi-label MRFs with non-convex priors are difficult to optimize. To tackle this, we introduce an algorithm that iteratively approximates the original energy with an appropriately weighted surrogate energy that is easier to minimize. Our algorithm guarantees that the original energy decreases at each iteration. In particular, we consider the scenario where the global minimizer of the weighted surrogate energy can be obtained by a multi-label graph cut algorithm, and show that our algorithm then lets us handle of large variety of non-convex priors. We demonstrate the benefits of our method over state-of-the-art MRF energy minimization techniques on stereo and inpainting problems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑