摘要
针对现有梯度算子在图像边缘检测中存在的对噪声比较敏感的问题,提出了一种改进的高斯—拉普拉斯算子的边缘检测方法。噪声图像中的边缘检测是一项关键任务,然而目前常用的几种梯度算子,包括已经提出的高斯—拉普拉斯算子都没能取得理想效果。提出的方法对传统的拉普拉斯边缘检测算子作了改进,并与高斯滤波器相结合,应用高斯滤波器平滑图像,抑制噪声,再基于拉普拉斯梯度边缘检测器进行边缘检测。最后,在imagenet数据集中选取了10幅图像进行实验,将提出的高斯梯度边缘检测器与传统的边缘检测器进行比较。评估结果显示,提出的方法所获得的峰值信噪比(PSNR)高于对比算法,而均方误差(MSE)更小。实验结果表明,提出的方法在实际应用中能够有效提高噪声图像边缘检测的质量。
Aiming at the problem that the existing gradient operator is sensitive to noise in image edge detection,this paper proposed an improved Gaussian-Laplacian edge detection method.Edge detection in noisy images was a key task.However, several commonly used gradient operators, including the proposed Gaussian-Laplace operator,had failed to achieve the desired results.The proposed method improved the traditional Laplacian edge detection operator and combined it with a Gaussian filter.First,it used a Gaussian filter to smooth the image and suppress noise.Then it performed edge detection based on a Laplacian gradient edge detector.Finally,it selected 10 images in the imagenet data set.It compared the Gaussian gradient edge detector proposed with the traditional edge detector.The evaluation results show that the PSNR obtained by the proposed method is higher than that of the comparison algorithm,and the mean square error(MSE) is smaller.Experimental results show that the proposed method can effectively improve the quality of noise image edge detection in practical applications.
作者
代文征
杨勇
Dai Wenzheng;Yang Yong(School of Information Engineering,Huanghe S & T University,Zhengzhou 450063,China;School of Computer Science & Technology,Huazhong University of Science & Technology,Wuhan 430074,China)
出处
《计算机应用研究》
CSCD
北大核心
2019年第8期2544-2547,2555,共5页
Application Research of Computers
基金
国家青年科学基金资助项目(61502432)
河南省科技厅科技攻关计划资助项目(152102210001)
河南省人力资源与社会保障厅博士后项目(2014022)
关键词
边缘检测
高斯-拉普拉斯
高斯滤波器
噪声图像
峰值信噪比
均方误差
edge detection
Gauss-Laplace
Gauss filter
noise image
peak signalto noise ratio(PSNR)
mean square error(MSE)
