图像处理工作流内部的数据存储

Data Storage in Workflow of Image Processing

图像处理由一系列的步骤组成,针对在处理序列中某个活动的处理参数变化后,从头开始重新则执行计算量大的问题,文中提出了增加"存储活动",从而图像处理时可以使用工作流技术进行处理链的动态自动优化,同时可以很容易地利用已有的图像处理程序。由于遥感图像巨大的数据量,不可能在每个步骤保存中间结果,文中提出的方法可以选择关键的活动以及保存适当的数据内容和格式,从而可以在工作流中某些活动发生变化的情况下快速重建图像。通过试验证明所提出的评估算法有效。

Image processing mostly consists of a sequence of steps, dense computation is needed when the process starts from the stretch after some parameters having been changed. This paper brings forward a method of adding ＇ cache activity＇ to a workflow, which makes it easy to optimize image processing automatically. It can also reuse the existent image processing application easily. It is impossible to save every halfway image since vast image data are used in image processing. The method can determine which key cache activity is worthy of preservation and which structure the data are kept in, thus making it quicker to produce real - time result by avoiding unnecessary computation. Tests proved that the method of evaluating the importance of a activity is effective.