基于分形理论和Kohonen神经网络的纹理图像分割方法

摘要

分形理论作为描述自然现象的一种模型，受到人们越来越多的重视。该文提出采用分形维数和多重分形广义维数谱q-D(q)作为纹理特征，采用自组织神经网络Kohonen网络作为分类器的图象分割方法。通过对纹理图象的分割实验，结果令人满意，证实该方法的有效性。%Fractal theory is attracting more and more attention as a model for natural phenomena. This paper presents a texture image segmentation method,which uses fractal dimension and multifractal function q-D(q) as texture features and uses Kohonen neural network as classifier. This method has been used in the experiment of texture images segmentation and the result is satisfactory,which verifies the effect of this method.