摘要
为解决脉冲耦合神经网络(Pulse Coupled Neural Network,PCNN)在图像分割中存在噪声适应性差、分割效率低等问题,提出一种基于SPCNN的双阈值自适应分割方法。首先通过整合拉普拉斯算子和高斯函数设计反馈输入域的连接系数矩阵,使图像在分割过程中在保护边缘细节的同时也具有抗噪性;然后利用最大类间方差法构造全新的双阈值点火判别模型,实现对目标像素的耦合点火。实验表明,该方法在实现参数自适应性的同时提高了分割效率,且具有良好的抗噪性。
To solve the problem of poor noise adaptability and low efficiency of Pulse Coupled Neural Network methods in image segmentation,a double threshold adaptive segmentation method based on SPCNN is proposed. First,the weight matrix of the feeding input field is designed by combined Laplace operator and Gauss function to protect the edge details and also have the noise resistance in the process of segmentation. Then,in order to realize the adaptability of parameter selection adopts the method of the orientation information measure. Last,put forward a new double threshold ignition discriminant model based on the method of maximum variance between classes to realize the adaptability of parameter selection. Experiments show that the proposed method improves the segmentation efficiency and achieves good noise immunity while achieving adaptive parameters.
作者
马跃辉
辛月兰
MA Yue hui;XIN Yue lan(College of Physics&Electronic Information Engineering,Qinghai Normal University,Xining 810000,China)
出处
《电子设计工程》
2019年第22期55-60,共6页
Electronic Design Engineering
基金
国家自然科学基金资助项目(61662062)
青海省自然科学基金项目(2016-ZJ-745)
