基于图论的图像分割算法分析研究

图像分割是把图像分成各具特性的区域并提取出感兴趣目标的技术和过程,图像分割新算法一直在更新,但都有局限性。本文介绍几种基于图论的图像分割算法,并对各自算法特点进行分析,通过将Normalized Cut归一化割集准则与图像的阈值分割联合起来区分目标和背景,提出了改进权值公式的算法。通过实验比较分析,改进的归一化图像分割算法有效消除了噪声,取得良好的实验效果,更接近人眼视觉的分割效果,并减少了算法的运算量。

Research of Image Segmentation Algorithm Based on Graph Theory

Image segmentation means to divide the image into several regions with different characteristics and the technology and process which extract interested targets. The algorithms of image segmentation have always been updated, but all have limitations. This paper introduced several image segmentation algorithms based on graph theory, and analyzed the characteristics. By combining the Normalized Cut standards with image threshold segmentation to distinguish between target and background, an improving algorithm was puts forward. The experimental＇results showed that the improved normalized image segmentation algorithm can get a better experi- mental effect, closer to the human eye vision effect, and reduce the computational complexity of the algorithm.