结合同异性度量的超像素分割方法

A Superpixel Segmentation Method Combined with Value of Sameness

针对以往分割方法在分割影像时出现的分割效果较差以及精度较低等问题,提出一种结合超像素同异性度量的高分辨率遥感影像分割方法,对分割效果进行改善。首先利用高斯滤波对影像进行平滑处理;再通过简单线性迭代聚类算法对影像进行像素级分割,生成超像素;之后根据灰度离散程度将离散超像素归并到邻近权重最大的超像素,解决独立像素点问题;然后计算超像素同异性度量值,根据阈值将超像素合并,达到分割影像的目的。此方法在传统分割算法基础上增强了超像素对梯度和颜色的敏感性,提高了影像分割的精度。实验表明,该方法可有效减小噪声点的影响,改善以往算法存在的过分割以及分割线偏移等缺陷。

Aiming at the problems of poor segmentation effect and low accuracy of the previous segmentation methods when segmenting images,this paper proposes a high resolution remote sensing image segmentation method based on the value of sameness.Firstly,use Gaussian filtering to smooth the image.Secondly,a simple linear iterative clustering algorithm is used to perform pixel-level segmentation on the image to generate superpixels.Next,the discrete superpixels are merged into superpixels with the largest weight according to the grayscale discrete threshold to solve the problem of independent pixels.Finally,calculate the value of sameness,and merge the superpixels according to the threshold to achieve the purpose of segmenting the image.This method enhances the sensitivity of superpixels to gradients and colors based on traditional segmentation algorithms,and improves the accuracy of image segmentation.Experimental results show that it can effectively reduce the influence of noise points,and improve the defects of the previous algorithm such as over-segmentation and segmentation line offset.