欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
传统的或改进型的中值滤波器，很难在图像噪声滤除和细节保留两方面兼顾与平衡．本文基于粗神经元构建了一种粗集神经网络，该粗集神经网络对5×5中值滤波器和多级FIR中值混合滤波器MFMHF（Multilevel FIR-Median Hybrid Filter）的处理结果进行融合．由于粗神经元的不可微性，BP算法不再适用，因此本文采用遗传算法GA来进行网络权值的学习，同时融入具有局部搜索能力的爬山法改善了进化后期的计算效率．仿真试验表明。粗集神经网络在图像融合滤波方面的性能优于BP网络和一般的中值滤波器．
英文摘要：
Generally, it is difficult for conventional or improved median filter to attain a trade-off between noise attenuation and detail preservation. A RNN （Rough Neural Network） based on rough neuron was constructed. The RNN fuses the filtering results of the 5×5 median filter and MFMHF（ Multilevel FIR-Median Hybrid Filter）. For the indifferentiability of rough neuron, BP algorithm can＇t be used any more. So, GA（Genetic Algorithm） was applied to tune the weights of the RNN. At the same time, the mountain climbing that has the local search ability was integrated to improve the computation efficiency of GA in the latter period of evolution. The results of simulation indicated that RNN outperformed BP network and general median filter in image fusion for filtering.
同期刊论文项目
同项目期刊论文
期刊信息
位置：