基于高斯混合及相对熵测度的图像分割

针对图像分割问题,结合高斯混合模型与信息论中的相对熵测度概念,提出一种新的图像阈值化方法。在提出方法中把图像阈值化问题看成是两个概率向量之间的匹配问题,因此首先用高斯混合模型去拟合图像直方图的灰度级分布,然后用相对熵测度去度量拟合分布与图像原灰度级分布之间的差异,并把该度量作为图像阈值化的准则函数。在对图像实施分割时,通过在图像灰度级范围中求取所定义的准则函数的最小值获得最佳阈值。在NDT、SAR及红外图像上的分割实验中用提出方法与传统及最新的图像阈值化方法进行比较,结果表明提出方法获得的结果要优于相比较方法获得的分割结果,因此提出方法是一种有效的图像分割方法。

Image segmentation based on Gaussian hybrid model and relative entropy measure

Aiming at the problem of image segmentation,a new image thresholding method based on Gaussian hybrid model and relative entropy measure is presented.In the proposed method,the problem of image segmentation is seen as a matching problem between two probability vectors.The gray level distribution of image histogram is fitted by Gaussian hybrid model firstly.Then the difference between fitting distribution and gray level distribution of original image is measured by relative entropy measure,and the measurement is used as the criterion function for image thresholding.When the image is segmented,the optimal threshold is obtained by minimizing the criterion function in the range of gray levels.For demonstrating the effectiveness of the proposed method,the experiments on NDT,SAR and infrared images are carried out and the performances of the proposed method is compared with those of three famous tradition methods and two state-of-the-art methods.The experiment results indicate that the proposed method can obtain better segmentation results,and this method is an effective approach for image segmentation.