异常识别与分离的自适应曲率结构分选滤波器

来源期刊：中南大学学报(自然科学版)1993年第3期

论文作者：文百红

文章页码：289 - 294

关键词：位场结构; 识别; 分离; 样条函数/曲率结构; 分选滤波器

Key words：potential field structure; recognition; separation; spline function/curvature structure; sorting filter

摘    要：在考察划分位场结构线性和非线性滤波器的基础上,设计出一种自适应曲率结构分选滤波器.提出和应用自适应曲率结构特征分析方法,借用图象结构识别技术进行异常识别和特征分析,而后对识别的异常应用三次样条函数进行分离.理论模型和野外实例表明,该滤波器具有异常划分精度高、适用性强、可分层次提取异常结构的特点.特别是通过自适应曲率结构分选,减少了滤波参数选择的主观性,便于实现计算机自动综合处理.

Abstract: On the basis of studying the current linear and nonlinear filters for identifying potential field structure,an adaptive curvature sorting filter is designed.A method of analyzing the characteristics of the field curvature is proposed and utilized in the paper.By means of techniques in image pattern recognition,anomalies are recognized and their features are delineated.By using cubic spline function to fit non-anomalous data,the background field is determined,and consequently the recognized anomalies are separated by substracting from the original field.Theoretical models and field examples show that the proposed filter has high accuracy and good applicability in anomaly identification and can be used to strip multi-scale field structure.With adaptive curvature sorting,the subjectiveness in specification of filter's parameters is reduced and is incorporated efficiently in automatic computer processing.