机构地区: 电子科技大学中山学院
出 处: 《重庆工学院学报(自然科学版)》 2008年第6期64-69,共6页
摘 要: 通过比较传统的Adaboost算法中样本权重的更新算法,提出了一种新的将类内归一化与全局归一化过程相结合的样本权重更新算法.并对两种算法进行了仿真实验,结果表明,该算法使用较少的弱分类器便可保证强分类器在保持较高检测率的同时,将误检率降低到可接受的范围内. Through comparing the weight updating in the traditional Adaboost algorithm, this paper presents a new algorithm of updating the weight, which combines the normalization process of the same type and between the different types. The simulation results show that the new algorithm, using lesser weak classifiers, can assure the high detection rate of the strong classifier and decrease the error detection rate to an acceptable range.
领 域: [自动化与计算机技术] [自动化与计算机技术]