【作者】 徐朝阳；

【导师】 陈建勋； 陈黎；

【作者基本信息】 武汉科技大学 ， 计算机应用技术， 2010， 硕士

【摘要】 水平集(Level Set)方法最早由Osher和Sethian提出的,该方法的基本原理是将演化的曲线或者曲面作为零水平集嵌入到高一维的水平集函数中,通过演化高维中的函数,达到演化零水平集的目的。在高维中不仅易于拓扑变换,而且无需重新参数化,计算更加精确,并且水平集方法可以非常容易的向更高维推广。水平集方法自提出以来,其良好的特性在图像处理和计算机视觉领域都得到了广泛的应用。针对水平集在演化过程中要不断重复计算修正符号距离函数的问题,Li提出的无须重新初始化水平集轮廓分割模型(Li模型)。该模型很好地保持水平集函数为符号距离函数,无需重新初始化,极大的提高了计算的速度和准确度,但几乎对于所有的水平集方法,在处理那些对象与背景颜色相近的图像时,任意零水平集进行演化都无法准确的分割出目标。水平集在处理图像时,对象与背景有明显的区别,即在对象边缘表现为灰度值的突变,否则就会出现过分割。本文在Li模型的基础上,提出了基于边界约束的水平集分割方法。该方法利用对象边界临近像素值近似相等的特性,在水平集进行演化过程中添加约束项,使其更加快速精确的分割出目标对象。针对自动白平衡算法计算色温存在较大误差的问题,本文提出基于水平集的色温估计方法。先将RGB彩色图像转换为YCbCr颜色空间的图像,通过亮度和颜色的差值与阈值的比较,计算出符号距离函数,再利用水平集方法进行演化,得到处理区域,那些孤立的像素点以及正常的近似白像素点将随着水平集的演化而被排除在外,使色温估计更加准确。更多还原

【Abstract】 The Level Set method is proposed by American mathematicians Stanley Osher and James Sethian, the basic principle of this method is the curve evolution, it is embedded in a higher dimensional space as zero horizontal level set function. The advantage of the Level Set method is that one can perform numerical computations involving curves and surfaces on a fixed Cartesian grid without having to parameterize these objects. Also, the Level Set method makes it very easy to follow shapes that change topology. And the calculation method can be used in higher dimension. Because of its good properties, it has become popular in many disciplines, such as image processing, computer graphics, computational geometry, optimization, and computational fluid dynamics.Li proposed a model: level set evolution without re-initialization a new variational formulation. The level set function can be initiatized with general functions that are more efficient to construct and easier to use in practice than the widely used signed distance function. The level set evolution in our formulation can be easily implemented by simple finite difference scheme and is computationally more efficient. However almost all of the level set method can deal with those objects color which is obvious difference with the background , otherwise it will produce over-segmentation. In this paper, a new model based on boundary constraint level set is proposed. This method utilizes the property that the object boundary has approximately equal value. This restrictions will make the segment step more quickly and accurately.Estimating color-temperature about automatic white balance algorithm is not accurate. This paper proposed a method estimating color temperature based on the level set. First of all, change RGB color images to YCbCr color space, then calculate the sign distance function, through the difference about brightness, chromatism and the threshold, the evolution of the level set will remove those isolated pixels and the normal approximation white pixels, make more accurately to estimate color temperature.更多还原