复杂网络理论在彩色图像分割中的应用研究

Research on the Application of Complex Network Theory in Color Image Segmentation

摘要:

图像分割在许多图像处理应用中具有重要作用。为提高彩色图像分割效果,更好的表示图像信息,利用复杂网络理论对彩色图像分割进行研究,从网络社团结构模型的角度分析图像,提出一种更为清晰的彩色图像分割表述方法。根据彩色图像中各像素点之间的相似性构造图像的网络社团结构图,实现对图像数据的建模,之后利用谱聚类社团划分算法对较好的网络社团结构图进行社团检测,进而实现对图像相似像素的聚类,最后得到图像分割结果。在BSDS300图像库上随机选取不同的彩色图像进行实验,通过对图像分割结果的分析研究,结果表明提出的算法在精度方面优于传统彩色图像分割算法,可以实现更好的分割结果,同时验证了社团划分算法进行彩色图像分割的可行性和有效性。

Abstract:

Image segmentation plays an important role in many image processing applications. In order to improve the color image segmentation effect, better represent the image information, use the complex network theory to research the color image segmentation, analyze the image from the perspective of the network community structure model and propose a clearer method of color image segmentation expression. According to the similarities between pixels in color image, the image of network structure is constructed and the image data is modeled. Then the cluster classification algorithm is used to conduct community detection of the good network community structure, Clustering of similar pixels in the image and finally the result of image segmentation. Different color images are randomly selected from the BSDS300 image database for experiments. The results of the analysis of the image segmentation results show that the proposed algorithm is superior to the traditional color image segmentation algorithm in accuracy, and can achieve better segmentation results. At the same time, Community segmentation algorithm for color image segmentation feasibility and effectiveness.