Abstract: It is difficult for?the segmentation of different objects and backgrounds in color image. To solve the problem, a MRF segmentation algorithm based on RGB color statistical distribution is proposed. The key parameters of the MRF model are set up, and the related formulas are deduced. The model improves the energy function, using the color distribution model of RGB. Comparing with the traditional one, the proposed algorithm greatly reduces the computational difficulty, and expresses more information of pixel value using RGB color space. Experimental results prove that the algorithm is more accurate and fairly effective.

Key words: color image, Markov Random Field（MRF）, RGB color space, energy function

摘要： 为了解决彩色图像中实现不同目标、不同背景的分割问题，提出了一种基于RGB色彩统计分布的MRF分割算法。在MRF模型的基础上，设定了MRF模型的关键参数并对相关公式进行了推导。对RGB颜色分布模型进行了介绍，重写了能量函数，大大降低了算法的计算难度。由于该算法在RGB彩色空间下充分表达了像素值的信息，使得该算法在分割准确性、适应性及快速性上均有提高。通过与其他算法在同等环境下进行分割实验定量对比，验证了该算法的有效性。

关键词: 彩色图像, 马尔可夫随机场（MRF）, RGB色彩空间, 能量函数