Massive Space database implementation policy-raster data 3

I. Raster Data Model

In Geodatabase, there are four ways to organize raster data. Includes raster dataset, raster catalog, mosaic dataset, and feature raster fields. The feature raster field stores a raster object as an attribute in a table, so it cannot be used as a real storage method for raster data. In addition to file-based storage, the preceding three organization modes in Geodatabase are generally available. Mosaic dataset is a new raster data model of ArcGIS 10.

· Raster Dataset
And raster catalog

Raster dataset can be considered as single-band or multi-band raster data. For example, if a single image file is imported directly through arccatalog, it becomes a raster dataset in Geodatabase, or multiple images are imported and spliced with mosaic to form a raster dataset.

Figure 1 importing a single grid to raster Dataset

Raster catalog can be considered as a set of raster dataset. For example, in a raster catalog, we can import several raster data:

Figure 2 importing multiple raster catalog

In File Geodatabase, a raster dataset is actually composed of several files, including storing all the grid pixel values in the gdbtable file and several index files.

Figure 3 File: files storing raster dataset in Geodatabase

When creating a raster catalog in file Geodatabase, there is an important option: management type. There are two optional management types: managed and unmanaged.

Figure 4 hosting and non-hosting

Managed means that all data will be physically imported to Geodatabase, while unmanaged stores the data in the file, in Geodatabase, only the pointer pointing to the real data is stored ). If we select the hosting method, the file Geodatabase contains several sets of files to store the Raster Data in the raster catalog. However, obviously, the pixel data of all grids with the largest data volume is stored in one gdbtable.

Figure 5 file the file storing raster catalog in Geodatabase

If you select an unmanaged mode, the files stored in file Geodatabase do not have files that store real unit data, but some ing files store locations of actual grid data. From the perspective of data management, the unmanaged mode imports data very quickly, because it does not need to actually store the data to Geodatabase. The hosting Mode means the traditional "warehouse receiving ", it takes a considerable amount of time.

The above are the basic concepts of raster dataset and raster catalog and their organization in file Geodatabase. Here we will take a short break and summarize the features of the two. In essence, raster dataset and raster catalog are very similar, that is, to organize one or more raster data in Geodatabase; if it is based on file Geodatabase, raster catalog can also organize the "raster" into "catalog" without having to store the original raster data in an unmanaged manner. If it is managed, raster dataset is similar to raster catalog with only one "raster". This can be seen more clearly in the ArcSDE discussed below.

In ArcSDE, the situation becomes more complex. Although there is no unmanaged mode in ArcSDE, all data must be imported physically, however, there are different storage types in the database. Taking ORACLE as an example, the raster data can be stored in long raw, blob, st_raster, or sdo_georaster. Next, let's take oracle as an example to see how raster dataset and raster catalog stored in the default blob type are organized in the database.

Raster dataset/CATALOG stored in blob are stored in several associated tables in the database. tables that actually store raster data include: auxiliary table, block table, band table, and raster Attribute Table. If you import a grid, its real data will be organized in the following tables:

Figure 6 related tables of raster dataset/CATALOG stored in blob in ArcSDE

Sde_aux _ [1]
It is an auxiliary table that stores color ing, image statistics, and other information. sde_ras _ is a raster Attribute Table that stores the "raster" attributes contained in raster dataset/catalog, the table should contain one (raster dataset) or multiple (raster Catalog) records; sde_bnd _ is the band table, which stores all the band information of "raster; sde_blk _ is a block table, which truly stores the grid pixel values. All the raster data (including the pyramid) in the raster dataset/catalog are sliced (128x128 pixels by default) is stored in each record, which is generally the most data table.

[1]
It is the rastercolumn_id recorded in the raster_columns table.