摘要
为了更好地利用骨肿瘤分形参数集对骨肿瘤进行模式判别 ,将基于连续变量的遗传算法和相应的交叉与变异算子应用于骨肿瘤的模式分类中 .针对该算法在实验中出现的振荡及不收敛问题 ,相应采用了自适应技术加以改进 .通过对比改进前后遗传算法的精度和速度 ,证明了改进后的自适应遗传算法稳健性能好 ,运算速度快 .利用该算法 ,可根据分形参数模式集对骨肿瘤进行有效的分类 ,达到了预期的目标 .
In order to classify the pathological characteristics of osteoma more accurately and correctly by using the combined fractal parameters, the genetic algorithm based on continuous variables was applied to the pattern classification of osteoma while the relevant crossover and mutation methods were employed too. Away from the shortcomings of initial algorithm, it is improved by adaptive method. By comparing the performance of two algorithms, it is proved that the adaptive genetic algorithm based on continuous variables is more robust and faster. Using this algorithm, osteomas can be effectively classified by the fractal parameters as expected.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2001年第2期166-170,共5页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金!重点资助项目 (6 96 310 2 0 )
关键词
模式分类
遗传算法
自适应
分形
骨肿瘤
Fractals
Genetic algorithms
Image processing
Medical imaging
Orthopedics
