算法概述
这篇文章是基于SGM的扩展参数模型,考虑了不同路径方向的单独惩罚,不同路径聚合代价的加权,以及与灰度梯度相关的惩罚。对于8个方向的SGM来说,扩展模型有20个参数,不宜进行手动调参,所以作者还提出利用covriance matrix adaptation evolution strategy (CMA-ES)进化算法来自动调参,调参可以在Shark3 open-source machine learning library [1]进行实现。然后基于GPU进行SGM算法的加速,结果可以达到11.7fps,视差范围为64像素。经过测试,扩展参数的SGM的精度更高。
下面给出扩展参数的SGM变体的参数组合:
实验结果
算法在middleburry 数据集和synthetic sequence数据集[2]上进行参数训练和验证。得到的结果如下:
参考文献
[1] C. Igel, T. Glasmachers, and V. Heidrich-Meisner, “Shark,” Journal of Machine Learning Research, vol. 9, pp. 993–996, 2008.
[2] T. Vaudrey, C. Rabe, R. Klette, and J. Milburn, “Differences between stereo and motion behaviour on synthetic and real-world stereo sequences,” in Proceedings of the Conference on Image and Vision Computing New Zealand. IEEE Press, 2008, pp. 1–6.