砂岩铀矿岩心颜色的成像光谱自动识别与定名方法研究

Study on Automatic Color Identification and Nomination of Rock Core with Imaging Spectra for Sandstone Uranium Deposit

现阶段砂岩铀矿岩心编录中颜色的辨识和描述依赖编录人员的感知与经验知识来完成,急需引入自动化手段,实现颜色识别的定量化和描述的规范化。文章分析了砂岩铀矿颜色和成像光谱数据的特点,参考石油行业岩石颜色参数命名规范,设计了可见—近红外全谱段法和真彩色三通道法的岩心颜色识别流程、结果优化与定名方法,开发了相应的软件代码。在庆阳地区ZKH3孔岩心颜色自动识别与定名实验中,针对岩心块之间空隙和岩心箱残留等干扰会造成颜色异常的问题,提出了通过判别待识别单元及前后色段颜色来去除异常色的方法,提高了颜色和定名的准确率。实验结果表明,三通道法识别的颜色偏暗且分段位置不准确,全谱段法识别结果与彩色相机照片一致,可以作为岩心颜色自动识别的有效方法。

This paper first introduced the color theory and expression method and the color nomination specification of rocks in petroleum industry,and then analyzed the color characteristics of sandstone uranium deposit host rock and color related feature of hyperspectral image.Two automatic methods of whole spectrum and true RGB color bands of VNIR image were established for the core color identification,result optimization and nomination,and the function software were programmed.Finally,a core color automatic recognition and naming experiment was finished in bore ZKH3 hole in Qingyang area with the HySpex VNIR hyperspectral data.The experimental results showed that the whole spectrum method represented the rock color very close to the camera photo,which tested that the method can be used as the effective method for automatic core color recognition,while true RGB color bands method gave out the color with low brightness and some wrong location.To eliminate the abnormal color caused by the interference of gap between the core sections and the residual image of the core box,a method comparing the color parameters of the identifying color units to the successive color segments was proposed to optimize the color recognition results and names.However,this method is infective at the end of the core column,and therefore the optimization scheme needs to be improved in the feature.