一种基于矢量阈值的自适应图像分割方法

A Self-adaptive Method of Image Segmentation Based on Vector Threshold

针对灰度图像,提出一种基于空间特征矢量的图像分割方法,分割前首先对边缘进行增强处理,并构建以像素的灰度、梯度、像素邻域均值为特征的三维特征空间.将图像像素点引入对应于空间特征点.通过计算像素特征矢量与特征矢量阈值的差矢量,求出矢量差与特征矢量阈值的夹角,比较夹角与动态分割参数的关系,以判定像素所在区域(目标或背景).实验表明,该方法能较快的实现图像分割,分割的效果也较好.

As to the gray image, a new method of image segmentation based on space feature vector is presented. Its procedures are mainly divided into three steps. Firstly, Edge is strengthened before image segmentation, and a feature space with three dimensions based on gray, gradient, pixels neighborhood is constructed in advance. Ever pixels in image is related to a feature point in the space. Edge is strengthened before segmentation. Secondly, computing the vector differences between feature vector and feature threshold vector, and the angle between difference vector and feature vector threshold can be obtained. Finally, comparing its angle with segmentation parameter one, the region （ target or background） of pixels can be determined. According to experimental results, it is an effective and accurate method in image segmentation .