中文摘要：

常规化探异常信息识别通常都是通过对比观测值与某一异常阈值的高低来判定某样品是否为异常样品,很多方法或者建立在经典统计学基础之上,要求数据符合一定的分布形式,或者面向整个研究区计算异常阈值,而无法顾及实际的地质环境。根据常规方法以数值大小计算异常阈值的原则,并且关注化探数据分布特征信息的分析和挖掘,提出了晕状特征提取方法,该方法能够有效识别局部异常及低缓异常。将此方法用于克拉玛依地区对金矿预测具有指示意义的化探数据的异常信息识别工作,结果表明：该方法能够有效识别化探异常信息,这些异常信息与研究区内已知金矿具有很好的对应关系。晕状特征提取方法在新疆东部的应用案例也显示出较好的结果。该方法可以作为一种有效的化探异常信息识别方法应用于成矿预测实际工作中。

英文摘要：

Geochemical anomaly identification using conventional methods are always based on the comparison of the observed value and a threshold. The conventional methods may be not suitable for geochemical data analysis because they are presented according to the classical theory of statistics and designed for normally distributed data,however,geochemical data is not follow normal distribution. Furthermore,most of the conventional methods will produced a threshold for the whole study area and pay no attention to the situation that the geological environment may be different everywhere. In this paper,a new method（named Halo feature extraction）is proposed to explore the distribution pattern of trace elements of gold deposits in the study area. The result shows that the Halo feature extraction method will identify gold deposits-related geochemical anomalies effectively when it was used to the geochemical data from Karamay and from east Xinjiang. The Halo feature extraction method is recommended to be used for geochemical anomaly identification and mineral exploration.