基于SLIC超像素分割的图分割算法

A Graph Partitioning Algorithm Based on SLIC Superpixels

图像分割是对图像进行分析和理解的关键步骤,是计算机视觉的基本技术之一.计算复杂度是评判一个图像分割算法好坏的重要标准,因此降低算法的计算复杂度是当前图像分割领域的主要任务之一.本文提出了一种基于SLIC超像素分割的图像分割方法.该方法利用SLIC算法生成超像素,通过构造相应的相似性矩阵,有效降低了Ncut分割算法的计算复杂度,大幅度缩短了Ncut算法的运行时间.由于SLIC超像素分割算法的准确性与高效性,在进行三类自然图像分割实验时,本文提出的方法无论在分割效果,还是在运行时间上,都要明显优于Ncut分割方法及它的改良算法.

Image segmentation is a key step in the analysis and understanding of the image,and it is also one of the basic techniques in computer vision field. Computational complexity is an important criterion to judge the quality of an image segmentation algorithm, therefore,it is one of the main tasks to reduce the computational complexity of algorithm in the field of image segmentation. An image segmentation method based on SLIC superpixels is proposed in this paper. This new algorithm generates the super-pixels using SLIC algorithm, and reduces effectively the computational complexity of Ncut algorithm via constructing a corresponding similarity matrix. Furthermore, the new algorithm can reduce greatly the running time of Ncut algorithm. Because of the accuracy of SLIC algorithm, the experiments of three natural images demonstrate that our algorithm is better than Ncut algorithm and its improved algorithm no matter on segmentation results or running time.