• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Journal of nannoplankton research: A publication of the International Nannoplankton Association >Automated pattern recognition and biometry of calcareous nannofossils: 20 years of SYRACO development
【24h】

Automated pattern recognition and biometry of calcareous nannofossils: 20 years of SYRACO development

机译：钙质Nannofossils的自动模式识别和生物测定：20年的锡拉科开发

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Performing calcareous nannofossil identification and morphometry studies can be tedious work, especially in sediments with low occurrences. Twenty years ago at CEREGE, we began to develop an automatization process for these two tasks, which led to the development of the SYRACO tool. Originally, it was composed of an artificial neural network (ANN), developed by Dollfus and trained with a back-propagation algorithm (Dollfus & Beaufort, 1999). We implemented actions in the ANN and a suite of several ANN, working in pyramids (Beaufort & Dollfus, 2004). If the level of recognition were highly acceptable (>96%), the number of false positives was also high (up to 50% for some morphological groups). Human corrections on the result-files were necessary (e.g., Beaufort et al., 2001). Automated morphometry was used to diminish the number of false positives (Beaufort et al., 2008). Morphometry was performed, not only on size measurements, but also on thickness and mass (Beaufort, 2005; Beaufort et al., 2014). Morphometry was then incorporated into., SYRACO with statistical pattern recognition tools (SPRT) such as Adaboost or SVM. The parallel use of ANN on im4ges and SPRT on morphometry allowed a large generalization for recognition (>90%) with a low level of false positives (<5%) (Barbarin, 2014). The recent technological developments on microscopes, cameras, and computers also allowed an important increase in the level of coccolithophore automated recognition. SYRACO is now able to detect all calcareous nannofossils from the last 40 Myrs. We illustrate the use of SYRACO by showing intraspecific morphometric variability in the genera Emiliania and Gephyrocapsa in natural (Beaufort et al., 2011) and artificial environments. However, the high plasticity of these placoliths in term of size and thickness should be considered before using these parameters for taxonomic purposes.
机译：表演钙质Nannofossil识别和形态学研究可以是繁琐的工作，特别是在低发生的沉积物中。二十多年前，我们开始为这两项任务开发自动化过程，从而导致了Syraco工具的开发。最初，它由人工神经网络（ANN）组成，由Dollfus开发并用背传播算法培训（Dollfus＆Beaufort，1999）。我们在金字塔（Beaufort＆Dollfus，2004）中实施了Ann和几个ANN的套件的行动。如果识别水平是高度可接受的（> 96％），则误报的数量也高（某些形态组的高达50％）。结果文件的人为矫正是必要的（例如，Beaufort等，2001）。自动形态学用于缩小假阳性的数量（Beaufort等，2008）。不仅对尺寸测量而进行了形态学，还进行了厚度和质量（Beaufort，2005; Beaufort等，2014）。然后将形态学掺入。，Syraco具有统计模式识别工具（SPRT），例如Adaboost或SVM。 ANN上的SNAN在IM4上的平行用途允许识别（> 90％）的大概括（> 90％），具有低水平的误阳性（<5％）（Barbarin，2014）。最近显微镜，摄像机和计算机的技术发展也允许CocColheophore自动识别水平的重要增加。 Syraco现在能够从过去40多人中检测所有钙质Nannofossils。我们通过在自然（Beaufort等，2011）和人工环境中，通过显示Genera Emiliania和Gephyrocapsa的内脏形态变异性来说明使用Syraco。然而，在使用这些参数以进行分类学目的之前，应考虑这些涂层的尺寸和厚度期间的高可塑性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号