单参数控制的灰度图像着色

Gray-scale image colorization based on the control of single-parameter

灰度图像着色是计算机图形学和计算机图像处理等领域的研究热点,在影视制作、数字照片编辑、动漫艺术等方面具有重要的应用价值。传统的方法通过大量的交互操作实现灰度图像中不同区域的聚类或者色源图像和灰度图像之间的匹配,大大降低了着色的效率。提出一种新的基于单参数控制的灰度图像着色方法。首先,利用球体几何理论建立色彩变换模型。其次,基于线性回归的方法,对色源图像和灰度图像的直方图分别进行多项式拟合建模。用户输入拟合多项式的阶数之后,可以实现色源图像和灰度图像的自动聚类并建立它们之间的匹配关系。最后,通过色源图像和灰度图像之间的颜色映射实现灰度图像的着色。该方法无须烦琐的用户交互,使得着色过程更为方便快捷。

Gray-scale image colorization is a hot topic in both computer graphics and image processing. It can be widely applied in film and television production, digital photo editing, animation industry, etc. A great number of user interactions, which greatly reduces the efficiency of colorization, are involved in conventional methods to achieve clustering of the gray-scale image or the matching between the gray-scale image and the source image. This paper presents a novel gray-scale image colorization method based on the control of single-parameter. Firstly, colorization model based on the theory of analytic geometry is presented. Secondly, based on the linear regression method, we calculate the polynomial fitting model of the histograms of the source image and the gray-scale image respectively. After users input the order of the polynomials, the clustering and matching of the source image and the grayscale images could be achieved automatically. Finally, the colorization is realized by transferring the color of source image to its corresponding area of the gray-scale image. This method greatly eliminates the user interactions, making the process of colorization more convenient and efficient.