遗传算法的噪声干扰数字图像分类性能评价  

Noise digital image classification evaluation based on genetic algorithm

作　　者：吴限光[1] 李素梅[1] 吴兆阳[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《信息技术》2012年第9期185-188,共4页Information Technology

摘　　要：针对现实中各种噪声干扰的数字图像识别分类的问题,提出了基于遗传算法优化的BP神经网络和支持向量机神经网络两种方案,先在无噪声干扰情况下建模,然后加入人工噪声模拟现实中的噪声干扰。结果表明,遗传算法优化后的支持向量机网络方案具备更好的抗噪声干扰能力,在噪声干扰数字图像分类中具有更高应用价值。This paper proposes two methods BP neural network and 5VM based on genetic algorithm for solving digital image recognition problems in real environment. Modeling first in the case of noise-free condition, and then add artificial noise for real-life noise simulations. The results show that by using genetic algorithm, SVM network solution has better noise immunity, and the genetic algorithm is more valuable in noise digital images classification field.

关 键 词：支持向量机 BP网络 遗传算法 图像处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]