摘要
针对有噪声的复杂多目标图像,引入四叉树的数据结构,区域生长条件运用象素周围k×k邻域的特性进行判决.由于是两个邻域比较特性,需要用到统计学中的假设检验等方法,采用最优阈值作为区域增长的相似性检测准则.达到了以下目的:第一,通过剪切过程减少了许多计算冗余,分割速度明显快于传统的区域增长;第二,由于考虑了邻域性质,抗噪声能力和工作鲁棒性也有所增强.实验表明,把四叉树结构引入复杂多目标图像分割能取得较好的效果.
The data structure of Quadtree is introduced to segment the complicated and multi-object image in this paper.The rule of region growing uses the property of the k×k neighborhood of pixels,and it needs the method of hypothesis testing because of the comparison of neighborhood.We use the best threshold value as similiarity rules and attach the aims as follows:Firstly,the computation is simplified by the process of branch-cut and the speed is greatly improved.Secondly,the ability of anti-noise and the robustness is enhanced because of the consideration of neighborhood of pixels.The experiment results show that this method has good performance of segmentation for complicated and multi-object image.
出处
《南昌大学学报（工科版）》
CAS
2003年第4期80-82,共3页
Journal of Nanchang University(Engineering & Technology)
关键词
图像分割
分裂与合并算法
四叉树
image segmentation
split-and-merge algorithm
Quadtree
