基于图像区域特征的自适应边缘检测  

Adaptive edge detection based on image region characteristics

作　　者：刘付民[1] 张治斌[1] 王娜[1] 沈记全[1] 

机构地区：[1]河南理工大学计算机科学与技术学院,河南焦作454000

出　　处：《计算机应用研究》2012年第6期2382-2384,共3页Application Research of Computers

基　　金：河南省软科学研究计划资助项目(102400450034);河南省科技计划基金资助项目(092300410216)

摘　　要：边缘检测技术的关键在于检测前的图像去噪问题。Madenda方法考虑了尖锐、模糊、噪声的影响,但在全图像范围内采用的是统一的噪声抑制参数和模糊控制参数,从而缺乏对不同区域内噪声抑制的针对性。为此,提出基于图像区域特征的自适应边缘检测方法,通过熵和对比度的计算,将原始图像逐级分解,直到达到预先设定的阈值条件。然后根据各个区域的特点,自适应地计算出相应的噪声抑制参数和模糊参数,从而达到对该区域噪声最合理的抑制,以达到最佳的边缘检测效果。实验结果表明,该方法可以有效地抑制噪声的影响,获得高质量的边缘检测图像。The key problem is noise filtering before detection.Sharp,blur and noise are considered in Madenda method.But exclusive parameter to control noise and blur is adopted in the whole image.This parameter is hard to have efficiency to all kinds of noise in different region.This paper proposed adaptive edge detection based on image region characteristics,and segmented original image by calculating entropy and contrast until the threshold was satisfied.Then it computed the parameter to control noise and blur according to specialty in all regions.These parameters was best of all noise in different regions.Experiments results show that this proposed method proposed is effective to control noise and good edge detection image will be obtained.

关 键 词：边缘检测 区域分割 自适应 模糊参数 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]