arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1307.5797math.OC
收藏

图像处理中形状能量的形状计算

Shape Calculus for Shape Energies in Image Processing

Günay Doğan

摘要:

许多图像处理问题自然地表示为能量最小化或形状最优化问题,其中自由变元是形状,例如二维曲线或三维曲面。例子是图像分割,多视图立体重建,从数据点云的几何插值。为了获得这样一个问题的解决方案,人们通常采取迭代方法,梯度下降算法,它更新候选形状逐渐变形为最佳形状。计算梯度下降更新需要形状能量的一阶变分的知识,或者更确切地说是一阶形状导数的知识。除了一阶形状导数之外,还可以利用二阶形状导数并开发具有更快收敛的牛顿型方法。不幸的是,在图像处理中的形状能量的形状导数的知识是零散的。第二形状导数是已知的,只有两个能量的图像处理文献和许多结果的第一形状导数是有限的,在这个意义上,他们要么是平面上的曲线,或开发的形状的一个特定的表示或一个非常具体的功能形式的形状能量。在这项工作中,这些限制被克服,并计算的第一和第二形状导数的形状能量,是代表在图像处理中发现的能量的大类。我们得到的许多公式是新的和一些推广以前现有的结果。这些结果对于任意维数的一般曲面都是有效的。这项工作的目的是作为一个食谱的研究人员谁处理的形状能量在图像处理中的各种应用,并需要开发算法来计算这些能量最小化的形状。

英文摘要:

Many image processing problems are naturally expressed as energy minimization or shape optimization problems, in which the free variable is a shape, such as a curve in 2d or a surface in 3d. Examples are image segmentation, multiview stereo reconstruction, geometric interpolation from data point clouds. To obtain the solution of such a problem, one usually resorts to an iterative approach, a gradient descent algorithm, which updates a candidate shape gradually deforming it into the optimal shape. Computing the gradient descent updates requires the knowledge of the first variation of the shape energy, or rather the first shape derivative. In addition to the first shape derivative, one can also utilize the second shape derivative and develop a Newton-type method with faster convergence. Unfortunately, the knowledge of shape derivatives for shape energies in image processing is patchy. The second shape derivatives are known for only two of the energies in the image processing literature and many results for the first shape derivative are limiting, in the sense that they are either for curves on planes, or developed for a specific representation of the shape or for a very specific functional form in the shape energy. In this work, these limitations are overcome and the first and second shape derivatives are computed for large classes of shape energies that are representative of the energies found in image processing. Many of the formulas we obtain are new and some generalize previous existing results. These results are valid for general surfaces in any number of dimensions. This work is intended to serve as a cookbook for researchers who deal with shape energies for various applications in image processing and need to develop algorithms to compute the shapes minimizing these energies.

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。

↑