人工神经网络在含噪图像边缘检测算法中的应用  

Edge detection algorithm of noisy image based on artificial neural network

作　　者：马雪景 王文焕 刘国巍[1] MA Xuejing;WANG Wenhuan;LIU Guowei(School of Electrical and Information Engineering,Anhui University of Science & Technology,Huainan 232001,Anhui,China)

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001

出　　处：《西安工程大学学报》2021年第2期79-84,共6页Journal of Xi’an Polytechnic University

基　　金：安徽省自然科学青年科研项目(2019b0286)。

摘　　要：为提高含噪图像边缘表现度,在分析图像边缘噪声的基础上,通过人工神经网络对噪声进行反馈与识别,获取全部噪声数据,然后对噪声进行滤除。首先,根据图像本身具有的噪声系数特点,对图像边缘信息的类型进行划分;其次,将边缘检测器作为噪声特征的神经元,抗体作为人工神经网络中的神经反馈,对图像边缘信息展开融合输出;最后,实现对图像边缘信息的检测。结果表明:相比于传统算法,含噪图像边缘表现度得以提高,且在检测边缘点数量更多的情况下缩短了检测时间。In order to improve the edge performance of the noisy image,based on the analysis of the image edge noise,the artificial neural network was used to respond and identify the noise,and then all the noise data were obtained,and the noise was filtered.Firstly,according to the noise coefficient characteristics of the image,the image edge information was divided into various types.Secondly,the edge detector was used as the neuron of noise feature,and the antibody was used as neural feedback in the artificial neural network,image edge information was fused and output.Finally,the detection of image edge information is realized.The results show that compared with the traditional algorithm,the edge performance of noisy image is improved,and the detection time is shortened in the case of more edge points.

关 键 词：人工神经网络 含噪图像 边缘检测 噪声滤波器 噪声系数 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]