一种边界梯度组合的图像识别技术与分割方法  

A Image Recognition and Segmentation's Method of Boundary and Gradient Composition

作　　者：游应德[1] 李成大[1] 

机构地区：[1]四川宜宾学院计算机科学系,四川宜宾644000

出　　处：《湘潭大学自然科学学报》2014年第2期99-103,共5页Natural Science Journal of Xiangtan University

基　　金：国家自然科学基金项目(60773031)

摘　　要：提出了一种基于边界和梯度特征的图像识别与分割方法.图像识别就是在相差仿射变化的过程中,将图像边界和梯度特征进行组合,提高图像识别效率.图像分割就是利用高斯平滑对图像进行预处理,然后通过Krisch算法得出图像梯度信息,随后利用二维最大类空间方差法进行图像分割,最后采取依据目标和背景的空间关系来消除个别误提边缘.严格的数学推导和实验都表明了算法的有效性和算法的应用性.This paper presents the image recognition and segmentation method based on the boundary and gradient features.The image recognition process is away from the affine change of the image boundary and gradient characterized in combination,to improve the efficiency of image recognition.Image segmentation is the use of Gaussian smoothing the image pre-processing,then by Krisch algorithm derived image gradient information,and then use the the 2D maximum space variance method for image segmentation,and finally taken based on the spatial relationship of the object and the background to eliminate the individual mistakenly put the edge.Rigorous mathematical derivation and experiments show the effectiveness of the algorithm and the application of the algorithm.

关 键 词：边缘 梯度特征 图像识别 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]