摘要
图像分割是自动目标识别的关键和首要步骤，其目的是将目标和背景分离，为计算机视觉的后续处理提供依据。笔者在对图像分割技术进行综合研究的基础上提炼出一种以最大类间方差法为基础的图像背景自适应分割算法，此算法根据目标和背景区域的灰度统计量来自动选取最优阂值。在此还通过仿真与常规的迭代法、四叉树、分水岭的效果图做了比较。
Image segmentation is a key and initial step of automatic target partition. Its purpose is to separate target and background, and it offers basis for follow - up handle of computer vision. The author comprehensively studied of image segmentation based on extracted an adaptive algorithm of image background segmentation based on the maximum variance between two classes. In this automatic threshold selection method, it took the gray level statistics of the resultant target and background region established the optimal threshold . It was testified and compared with several thresholding methods in common use such as Quadtrees, Iteration and Watershed.
出处
《西安铁路职业技术学院学报》
2010年第4期22-24,共3页
Journal of Xi’an Railway Vocational & Technical Institute
关键词
图像分割
阂值化
自适应
迭代法
分水岭
image segmentation
threshold
adaptive
Iteration
Watershed
