图像浮雕效果细胞神经网络模板的参数设计  

Parameter Design of Cellular Neural Network Template with Relief Effect

作　　者：李硕[1] LI Shuo(School of Statistics and Data Science,Xinjiang University of Finance and Economics,Urumqi 830012,China)

机构地区：[1]新疆财经大学统计与数据科学学院,新疆乌鲁木齐830012

出　　处：《厦门理工学院学报》2021年第3期56-61,共6页Journal of Xiamen University of Technology

摘　　要：基于细胞神经网络(cellular neural network,CNN)的最优边缘检测(optimal edge detector,OED)的研究,设计出一个3阶CNN算法模板的参数规则。规定CNN的反馈模板为0矩阵,阈值为0,控制模板为中心反对称矩阵,并且它的中心元素为0,其余8个元素中的一组(4个)元素绝对值的和不为0。仿真结果表明,使用5阶CNN算法、3阶CNN算法和拉普拉斯算法对同一幅图像进行处理时,平均耗时分别为0.0716,0.0385,0.0297 s,3阶CNN浮雕处理算法的耗时几乎与拉普拉斯算法的相等,但视觉效果显著优于5阶CNN算法和拉普拉斯算法。表明3阶CNN浮雕处理算法可以快速、有效地实现对图像的浮雕效果处理,所设计的3阶CNN算法模板的参数规则是合理有效的。Based on the research of optimal edge detector(OED)of cellular neural network(CNN),a parameter rule of 3-order CNN algorithm template is designed,which stipulates that the feedback template of CNN is a 0 matrix,the threshold is 0,the control template is a central antisymmetric matrix and its central element is 0,and the sum of absolute values of a group(4 elements)of the other 8 elements is not 0.The simulation results show that the average processing time of the 5-order CNN algorithm,3-order CNN algorithm and Laplace algorithm is 0.0716,0.0385 and 0.0297 s,respectively.The time of the 3-order CNN relief processing algorithm is almost the same as that of the Laplace algorithm,but the visual effect is significantly better than that of the 5-order CNN algorithm and Laplace algorithm,which shows that the 3-order CNN relief processing algorithm can be fast and accurate in operation,and the parameter rule of its template designed in this paper is proper and effective.

关 键 词：浮雕效果 细胞神经网络 模板参数规则 最优边缘检测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]