摘要
为了提升勘探工作效率,准确提取复杂地质分形特征,提出一种支持向量机的复杂地质分形特征自动提取方法。在两种线性可分情况下,将最佳的分类面简化为最佳分类直线,获得最佳分类平面;考虑噪声及其它干扰因素造成线性不可分的状况,得到广义的最佳分类面;基于两点可以得出支持向量机内积;通过抽取异常重磁分维数的频率域角度获得分维值,反映出不同频率能量关系,计算地质图像内的方差、均值、相关性、对比度以及角二阶矩,即可完成地质分形特征自动提取。实验结果表明,所提方法能够很好区分各类泥岩样本,特征提取效果准确。
In order to improve the efficiency of exploration and accurately extract fractal features of complex geological structure,an automatic extraction method for fractal features of complex geological structure based on support vector machine was proposed.In the case of linear separability,the best classification surface was reducible to the best classification straight line to obtain the best classification plane.However,the linearly non-separable problems might be caused by noise and other interference factors,so the optimal separating plane must be obtained in a broad sense.On this basis,the inner product of support vector machine was obtained.The fractal dimension value was calculated by extracting the angle of frequency domain of fractal dimension of gravity and magnetic anomaly.This value reflected the energy relationship between different frequencies.Moreover,the variance,mean,correlation,contrast and angular second moment in the geological image were calculated to complete the automatic extraction of geological fractal feature.Simulation results show that the proposed method can accurately distinguish mudstone samples with better feature extraction.
作者
张钰帆
郑寒
王艳霞
ZHANG Yu-fan;ZHENG Han;WANG Yan-xia(Southwest Forestry University,Kunming Yunnan 650224,China)
出处
《计算机仿真》
北大核心
2021年第1期261-264,335,共5页
Computer Simulation
基金
国家自然科学基金(41561004)。
关键词
支持向量机
复杂地质特征自动提取
异常重磁分维数
频率能量
Support vector machine(SVM)
Automatic extraction of complex geological feature
Fractal dimension of gravity and magnetic anomaly
Frequency energy
