Quality control and uncertainty analysis of image metadata

Metadata is an important document of image about outcome, which reflects history and status of the results. One of the factors affecting the quality of image data is quality of metadata, and quality of metadata is a decisive factor whether the user adopts dataset. Therefore, quality of metadata must be controlled; and metadata will introduce a random uncertainty in the process of entrying, editing, modifing, which must be carried out uncertainty analysis in order to improve the accuracy of check of metadata. First, analyzing the qualitative requirements of image metadata, and exploring how to control the quality of image metadata from acquired methods for manual entry and automatically generated by software. Second, analyzing the results of metadata and checking metadata, by checking image metadata to find and correct data items and attribute values to control the quality of metadata. There are three methods of check, one is template check for there being standard data template; the other is classification check for there no being standard template; the last one is the check of coordinate metadata to check the coordinated cutting scope. At last, I use the certain theory to make uncertainty analysis for metadata attributes.