一种基于马尔科夫随机场模型的彩色纹理图像分割
Gabor-MRF Model Based on Color Texture Image Segmentation
-
摘要: 提出了一种基于Gabor滤波和马尔科夫随机场的彩色纹理特征图像的分割算法。首先对色彩和纹理特征进行了分析,将RGB色彩空间非线性变换到CIE-LUV空间,构造颜色的特征向量;然后对原始彩色图像进行Gabor滤波和高斯平滑处理,得到恰当表示原图像的灰度纹理图像;再对原图像建立MRF分割模型,结合色彩和纹理信息,运用贝叶斯理论和迭代优化算法估计最大后验概率(MAP)。实验表明,本文方法可以有效地实现图像分割。Abstract: We propose a Gabor filter and Markov random fields(MRF)-based method for color texture image segmentation.First,we analyze color and texture feature,transforme RGB space to LUV space to get color feature vector,and then do Gabor filtering and Gaussian smoothing processing on original color image and MRF model is used to represent the regional relationship.Finally,we combine color and texture information and use Bayesian method to estimate maximum a Posteriori(MAP).The experimental results show that this algorithm is efficiently doing color texture image segmentation.
下载:
