基于改进遗传算法的微小图像边缘特征快速识别研究
|
摘要
针对微小图像边缘特征识别方法存在识别速度慢、识别精度低的问题,提出一种基于改进遗传算法的微小图像边缘特征快速识别方法。对经典边缘特征识别算子进行处理并构成初始化子群,利用新的编码方法对子群中的染色体进行编码,构建适应度函数计算出相应的适应值,并根据适应值设计遗传算子,完成遗传算法的改进;采用克隆选择理论对微小图像边缘特征进行预处理,形成微小图像的抗体库,基于改进遗传算法并利用K近邻判别理论完成待识别微小图像特征智能识别。仿真结果表明,与标准遗传算法相比,改进遗传算法平均进化代数减少3.15代,效率提高了35%,可满足微小图像边缘特征识别的实际要求。
Aiming at the slow recognition speed and low recognition precision in the current methods of edge feature recognition of micro-images,a fast recognition method of edge feature of micro-images based on improved genetic algorithm is proposed.The classical edge feature recognition operator is processed and initialized into a subgroup.The chromosomes in the subgroup are encoded by a new coding method.The fitness function is constructed to calculate the corresponding fitness value.The genetic operator is designed according to the fitness value to improve the genetic algorithm.The edge features are preprocessed to form the antibody library of the micro-image.Based on the improved genetic algorithm and K-nearest neighbor discriminant theory,the intelligent recognition of the micro-image features to be recognized is completed.The simulation results show that compared with the standard genetic algorithm,the average evolutionary algebra of the improved genetic algorithm is 3.15 less and the efficiency is increased by 35%,which can meet the practical requirements of edge feature recognition of micro-image.
Aiming at the slow recognition speed and low recognition precision in the current methods of edge feature recognition of micro-images,a fast recognition method of edge feature of micro-images based on improved genetic algorithm is proposed.The classical edge feature recognition operator is processed and initialized into a subgroup.The chromosomes in the subgroup are encoded by a new coding method.The fitness function is constructed to calculate the corresponding fitness value.The genetic operator is designed according to the fitness value to improve the genetic algorithm.The edge features are preprocessed to form the antibody library of the micro-image.Based on the improved genetic algorithm and K-nearest neighbor discriminant theory,the intelligent recognition of the micro-image features to be recognized is completed.The simulation results show that compared with the standard genetic algorithm,the average evolutionary algebra of the improved genetic algorithm is 3.15 less and the efficiency is increased by 35%,which can meet the practical requirements of edge feature recognition of micro-image.
引文
[1]张晓娟,樊东燕.倾斜车牌图像边缘模糊特征识别方法研究[J].计算机仿真,2017,34(1):372-375.
[2]张志佳,魏信,周自强,等.基于深度图像和点云边缘特征的典型零部件识别[J].信息与控制,2017,46(3):358-364.
[3]崔永锋,刘伟.远程采集图像特征的优化识别过程仿真[J].控制工程,2016,23(7):1053-1056.
[4]李龙龙.基于图像分析的叶片特征识别相关度研究[J].计算机工程与设计,2016,37(8):1-5.
[5]闵永智,殷超,党建武,等.基于图像色相值突变特征的钢轨区域快速识别方法[J].交通运输工程学报,2016,16(1):46-54.
[6]张衡,谭晓阳,金鑫.基于多视图聚类的自然图像边缘检测[J].模式识别与人工智能,2016,29(2):163-170.
[7]余伶俐,夏旭梅,周开军,等.基于仿生视觉的图像RST不变属性特征提取方法[J].仪器仪表学报,2017,38(4):985-995.
[8]姬晓飞,秦宁丽,刘洋.多特征的光学遥感图像多目标识别算法[J].智能系统学报,2016,11(5):655-662.
[9]卞桂平,秦益霖.基于Canny算法的自适应边缘检测方法[J].电子设计工程,2017,25(10):53-56.
[10]王林华,袁明辉,黄慧,等.太赫兹安检系统人体图像边缘物体识别[J].红外与激光工程,2017,46(11):139-144.
[2]张志佳,魏信,周自强,等.基于深度图像和点云边缘特征的典型零部件识别[J].信息与控制,2017,46(3):358-364.
[3]崔永锋,刘伟.远程采集图像特征的优化识别过程仿真[J].控制工程,2016,23(7):1053-1056.
[4]李龙龙.基于图像分析的叶片特征识别相关度研究[J].计算机工程与设计,2016,37(8):1-5.
[5]闵永智,殷超,党建武,等.基于图像色相值突变特征的钢轨区域快速识别方法[J].交通运输工程学报,2016,16(1):46-54.
[6]张衡,谭晓阳,金鑫.基于多视图聚类的自然图像边缘检测[J].模式识别与人工智能,2016,29(2):163-170.
[7]余伶俐,夏旭梅,周开军,等.基于仿生视觉的图像RST不变属性特征提取方法[J].仪器仪表学报,2017,38(4):985-995.
[8]姬晓飞,秦宁丽,刘洋.多特征的光学遥感图像多目标识别算法[J].智能系统学报,2016,11(5):655-662.
[9]卞桂平,秦益霖.基于Canny算法的自适应边缘检测方法[J].电子设计工程,2017,25(10):53-56.
[10]王林华,袁明辉,黄慧,等.太赫兹安检系统人体图像边缘物体识别[J].红外与激光工程,2017,46(11):139-144.