1 grid Pyramid Definition
Pyramids can be used to improve performance. They are a reduced-sample version of the original raster dataset and can contain multiple downsampling layers. Each successive layer of the pyramid is downsampling at a scale of 2:1. The following is an example of a level two pyramid created for a raster dataset:
Figure 1?1 Pyramid Example
Pyramids can speed up the display of raster data by retrieving only data that uses the specified resolution (depending on the display requirements). With pyramids, you can quickly display a lower-resolution copy of the data as you draw the entire dataset. As the zoom operation progresses, the finer resolution levels are drawn gradually, but the performance remains the same because you are drawing smaller areas in succession. The database server automatically selects the most appropriate pyramid level based on the user's display scale. If you do not use Pyramids (pyramid), you must read the entire dataset from disk and then resample it to a smaller size. This is called "Show resampling", which occurs when the ArcGIS display is refreshed.
Each raster dataset can only be built once, and then each time you view the raster dataset, the pyramids are accessed. The larger the raster dataset, the longer it takes to create the pyramid set. However, this also means that you can save more time for the future.
Although you cannot build pyramids for a raster catalog, you can build pyramids for each raster dataset in the raster catalog. A mosaic dataset is similar to a raster catalog. You can build pyramids for each raster dataset in a mosaic dataset, while building pyramids for a mosaic dataset can only be done with specific tools in the Mosaic DataSet toolset.
2 Pyramid Files
Pyramids are stored in separate files, which are usually located next to the source rasters. There are two types of pyramid files: Pyramids (. ovr) and reduced-resolution datasets (. rrd). Both types of pyramid files are readable by ArcGIS, but can only be written to. ovr files (a few exceptions are described below).
The. rrd file was created for the ERDAS IMAGINE file.
. ovr files are created and used by ArcGIS version 10 or later.
Typically, create an RDD file for a raster dataset (Rasterdataset) and create a OVR file for a mosaic dataset (Mosaicdataset).
Here are the two types of pyramid files that are detailed below.
2.1 OVR File
The pyramid (. ovr) file is the most recent format for the Pyramid (pyramid) layer that stores the raster dataset. You can create a. ovr file when the following conditions are true:
The file format is not a ERDAS IMAGINE (. img) file.
Build pyramids (Pyramid) using ArcGIS 10 or later.
One of the advantages of the. ovr file compared to the. rrd file is that you can control the compression type and the quality of the pyramid.
An enhanced feature of the. ovr file is that it can be compressed using the LZ77 or JPEG compression scheme. Remember that JPEG compression is only available for file types that can store data based on the JPEG specification. If you select the JPEG pyramid, you can also specify the compression quality of the JPEG pyramid.
The size of the Pyramid (overview) file is approximately 8% of the original uncompressed file. If the pyramid file is compressed, the. ovr file may be smaller (such as 2%). Because compression relies on a similar degree of data, it is difficult to estimate the size of the. ovr file.
The following is an example of the increase in the storage size of raster datasets under each pyramid level (depending on the compression type). Level 0 represents the original raster dataset.
Table 1 OVR File size
Note: The wavelet compressed raster file format (such as JPEG 2000, ECW, and MrSID) has an internal pyramid. This usually means that pyramids cannot be created using ArcGIS, but you can create pyramids for JPEG 2000. In addition, a geodatabase raster with wavelet compression can also build pyramids.
Note: It is generally not recommended to build pyramids for military data. For more information, see Create military data and add it to a mosaic dataset.
When you mosaic raster data in an ArcSDE geodatabase or file geodatabase, you can build pyramids for the raster datasets while the raster data is mosaicked to the raster dataset, or build the pyramids after the load is complete. ArcGIS allows a partial pyramid to be constructed, that is, only the part of the pyramid that overlaps the source data is rebuilt during the tessellation operation. This is useful when updating a mosaicked raster dataset, since adding a new raster dataset eliminates the need to reconstruct the pyramids for the entire raster dataset. However, if you are updating data at the origin of the raster dataset (the reference point of the pyramid), you need to rebuild the pyramid for the entire raster dataset.
2.2 RRD File
A decrement resolution dataset (. rrd) file is a method for storing the pyramid layer of a raster dataset. You can create an. rrd file in the following two scenarios:
The file format is ERDAS IMAGINE (. img) file.
Build pyramids using versions prior to ArcGIS 10.
2.2.1 RRD File Requirements
The necessary condition for building an. rrd file for a raster dataset is to have write access to the directory where the dataset resides. When you select Build Pyramids (Pyramid), the. rrd file is created in the same directory as the source dataset. These pyramids are used the next time the raster dataset is added. An exception to this rule is the ERDAS IMAGINE file, where the pointer to the pyramid is stored in an image file instead of being stored in a secondary (. aux) file as in other formats. In this case, you must have write access to the raster dataset and the directory in which it resides.
2.2.2 RRD File Size
For rasters that have been decompressed, the. rrd file has a minimum size of approximately 8% of the original raster dataset size. However, in some cases, the size of the. rrd file that you create can be larger than the original file (33%), depending on the compression technique used in the original raster file. If the. rrd file exceeds 2000 megabytes, a descending resolution dataset external (. rde) Raster data file is created as a supplemental file.
The following is an example of an increase in the storage size of a raster dataset when using all the pyramid levels. Level 0 represents the original raster dataset.
Table 2 Raster DataSet storage size and pyramid level
Comparison of two kinds of 2.3 ovr&rrd
Table 32 Comparison of files
2.4 Mosaic DataSet and overview
From the above, it can be concluded that at the base level, the mosaic dataset pyramid is similar to the raster DataSet Pyramid (Pyramid). They are low-resolution images that are created to improve display speed and reduce CPU usage because fewer rasters are tested to display the mosaicked image. However, they are quite different because you can control many of the parameters used to create pyramids for the mosaic dataset. You can create a mosaic dataset pyramid that covers only specific areas, or you can create a mosaic dataset pyramid with only a specific resolution. By creating the pyramid, you can view all the rasters that are contained in the entire mosaic dataset, not just a single raster. Pyramids usually start at the point where the Raster pyramid (pyramid) stops, but if you do not want to use all of the raster's pyramids (pyramid), you can specify the base pixel size of the build pyramid.
Pyramids make it possible to quickly display a lower-resolution copy of your data while viewing the entire mosaic dataset. The fine resolution level is drawn when zoomed in, while maintaining performance, because the mosaicked image is created using a reduced area. This can be helpful when you provide a mosaic dataset in the form of an image service or on a network. The most appropriate pyramid should be selected based on the display scale. If you do not use pyramids, the entire data set will be processed in a dynamic manner.
There is a property in the mosaic dataset that restricts the number of rasters to be used to generate the mosaicked image. You can modify this property, but if you reach this limit, we recommend that you build pyramids. The purpose of this restriction is to prevent the computer from performing too many pixel processing operations and to maintain a good display speed.
When a pyramid is generated, it is processed using the function applied to the source raster. It is not processed using any functions applied at the mosaic dataset level. Therefore, you do not need to regenerate the pyramids when you modify the mosaic dataset level functions.
2.4.1 Generating Pyramids
There are two types of tools used to generate pyramids. The Define Pyramids (Overview) tool can be used to customize the pyramids that will be generated. The Build Pyramids (Overview) tool defines missing pyramids, updates, and build pyramids.
You can use the default settings to build pyramids when you add raster data using the Add Rasters to Mosaic DataSet tool or the build Pyramids (Overview) tool. Which tool you use will depend on your workflow. If you are creating a simple mosaic dataset (for example, a mosaic dataset that contains all the ortho images for your city) and you have not made any modifications to the footprints, you can select the option to build the Pyramids in the tool dialog box when you add data. This will apply all the available options of the build Pyramids (Overview) tool to build the complete pyramid. If you have not built your raster dataset's pyramids (pyramid) and statistics, you might want to add a ortho image to the mosaic dataset, use the Build pyramids and Statistics tool, and then use the build Pyramids (Overview) tool.
Many raster datasets have internal pyramids (pyramid); You can also build pyramids (pyramid). Pyramids on the Raster dataset (pyramid) will reduce the number of pyramids required for the mosaic dataset.
2.4.2 Definition Overview
Use the Define Pyramids (Overview) tool if you want more control over where pyramids are generated, how they are generated, or what they build.
Figure 2?1 Definition Overview
1. Output location (optional), used to store the pyramid's folder or Geodatabase. Pyramids stored and managed in personal and file geodatabases have a default pyramid folder location, which is located in the same workspace as the existing geodatabase.
For ArcSDE Geodatabases, the pyramid location must be the ARCSDE database itself.
2. The dataset determines the range (optional), a raster dataset or Polygon feature class that defines the extent or shape of the pyramid. The extent of the raster dataset will be used when using a raster dataset. The shape of the polygon is used when the feature class is used.
3. Range (optional), four coordinates are generated to define the pyramid range. If no scope is defined, the extent of the pyramid is determined using the Mosaic dataset boundary. The Clear button will reset the range back to the extent of the raster catalog.
4. Pixel size (optional), which is used to generate the base pixel size of the pyramid. The default value is determined by the software.
5. Series (optional), the number of pyramid levels that will be generated.
If the number is greater than 0, it will be the number of pyramid levels generated. For example, 3 indicates that three pyramid levels will be generated. If the value is left blank or 1, the resulting pyramid can be up to 1500 rows (or columns).
6. Number of rows (optional), the best number of rows in the pyramid. The larger the value, the greater the file size, and the more likely it is to regenerate the file if any of the images below change. This number may affect the number of pyramid images created. If it is a larger number, fewer pyramids are generated. If it is a smaller number, more files will be generated.
7. Number of columns (optional), the best number of columns in the pyramid. The larger the value, the greater the file size, and the more likely it is to regenerate the file if any of the images below change. This number may affect the number of pyramid images created. If it is a larger number, fewer pyramids are generated. If it is a smaller number, more files will be generated.
8. Overview sampling factor (optional), used to determine the ratio of the next pyramid size. For example, if the cell size of the first level is X and the pyramid factor is 3, the next pyramid pixel size will be 3x.
9. Force Pyramid (Overview) block parameter (optional), which affects the level at which the pyramid will be generated. Unchecked-pyramids (overview) are created only on top of the primary raster pyramid (pyramid) level. This is the default setting. Checked-Pyramids (overview) will be created at all levels, although the main grid has pyramids (pyramid).
10. Resampling method (optional), the resampling algorithm used when creating pyramids (Overview). nearest-nearest Neighbor allocation method. bilinear-bilinear interpolation method. This is the default setting. cubic-three-time convolution interpolation method.
11. Compression method (optional), this parameter defines the type of data compression used when storing the pyramid image. jpeg-lossy compression. This is the default setting. None-no data compression. lzw-lossless compression.
12. Compression quality (optional), the compression quality used by the JPEG compression method. The range of compression quality varies from 1 to 100. A larger number means that the image is of higher quality, but less compressed.
2.2.3 Building Overview
Figure 2?2 Building Pyramids
The Build Pyramids (Overview) tool has several options that apply to workflows that are modified or updated on the mosaic dataset.
1. Through the query definition, you can use SQL to define the query, or use the Query Builder to build the query.
2. Define the missing overview map tile (optional) to generate pyramids when you define a pyramid that is insufficient or adds new data without defining an additional pyramid. Where: checked-automatically identifies where the pyramid is needed and defines the pyramid at that point. This is the default setting. Uncheck-The new pyramid is not defined.
3. Generate Pyramids (overview) (optional) to generate all pyramids that need to be created or recreated. This includes missing pyramids and outdated pyramids. Checked-generates pyramids for all types and all states. This is the default setting. Uncheck-builds only the pyramids that have been defined but not built.
4. Generate only the missing Pyramid (overview) image (optional), which is used when the pyramid has been defined but not yet generated. Checked-generates pyramids that have been defined but not yet generated. This is the default setting. Unchecked-pyramids that have been defined but not yet generated are not generated.
5. Regenerate only obsolete pyramid (overview) images (optional), if the underlying raster dataset changes or its properties have been modified, the pyramid is recognized as an outdated pyramid. Check-update the outdated pyramids. This is the default setting. Uncheck-Do not update the outdated pyramids.
The option to regenerate the outdated pyramid image again is highlighted below. Obsolete indicates that the source raster used to create the pyramid has been modified or missing. Use this option if you removed the raster from the Mosaic dataset or modified the footprints after the pyramid was generated. The option to define missing pyramid slices is useful when you have made a number of modifications to the Mosaic dataset, including adding or removing rasters, or changing the footprints of the rasters. This option evaluates the mosaic dataset to determine whether enough pyramids have been defined, or if new data has been added without defining another pyramid, so that you can generate a complete set of pyramids for the mosaic dataset.
2.3.4 Maintenance Overview When adding raster data
Figure 2?3 Adding raster data
Adding a raster dataset to an established mosaic dataset can also be part of a overview setting, as follows:
1. Update pyramids (Overview) (optional) to define and generate pyramids for the mosaic dataset. Checked-the pyramid will be defined and generated. Unchecked-pyramids will not be defined or generated. This is the default setting.
2. Maximum level (optional), same as above.
3. Maximum cell size (optional), same as above.
4. Minimum number of rows or columns (optional), as above.
2.3.5 Instance Description
Below we combine a specific image data to tell the relationship between the Mosaic dataset and the overview.
Figure 2?4 Overview Example
As shown, browse the mosaic dataset, and when zoomed-in images are less common, the images generated by the pyramids are generally used; when you zoom out of the view to include many images, you can see many even global images, which are typically called overview. The figure has three original images, each with a pixel size of 0.5. Pyramids are created for each original image, each creating two levels, resampling twice times, which is the default value. The overview is built on top of the pyramid, and the default resampling parameter is 3. Overview is a well-mosaicked image and is limited by the size of the data map. In this example, there are only 2 overview at the first level, even though the original data is 3 images, and eventually only 1 are overview at the top level.
can have different combinations. For example, the original image above can have no pyramids, in which case the overview will start with a size of 1.5 pixels (1:5,670) which is 3 times times and will continue until it reaches the appropriate overview size display, thus increasing the speed of the entire image. If the original image has already generated pyramids, overview will need less, resulting in less time to generate overview.
Overview and Pyramid in 2.2.6 mosaic datasets
In general, the execution of pyramids (overview) is faster than the pyramid (pyramid) that displays each raster in the mosaic dataset. When using the following imagery, you might consider building Pyramids (overview) on the Raster Pyramid (Pyramid):
Pre-processed block images, such as Ortho image quadrant
Docking (non-overlapping) imagery without changing the mosaic method to produce
Images are processed dynamically, but the parameters and mosaic methods do not change
When building more complex mosaic datasets, especially when using mosaic methods and dynamic processing, it is useful to build pyramids (pyramid) on the source rasters and to build pyramids (overview) only where they are needed, for example:
Raster datasets larger than 5000 columns
Because the image overlaps, the mosaic method is used
Dynamic processing will appear on the source rasters in any scale range
Imagery is not a static preprocessing raster
Also, if you plan to use lock raster (for example) to view individual rasters in a mosaic dataset, raster datasets with pyramids (pyramid) (and statistical values) themselves may render at different scales at a faster rate. It is also faster to preview the raster in the Raster Viewer window (accessed from the grid column in the property sheet). Pyramids can sometimes be defined but not built; the location of the storage pyramid crashes, the file is lost, or, for some reason, the pyramid is missing, but it is defined in the Mosaic dataset. You can create or recreate missing pyramids by using the option to generate missing images.
In addition, you can use other datasets as pyramids, such as layers in arcgis.com or other pyramids that are generated for other mosaic datasets. This is usually recommended because when you zoom out, many of the details in the image are lost, the colors degenerate, and the features are mixed, and large features are difficult to identify, especially when using high-resolution imagery or ortho imagery. Use the Add Rasters to Mosaic DataSet tool to add these raster datasets and to process them as you would any other raster in the mosaic dataset. The only difference between these raster datasets and the pyramids generated from the mosaic dataset is that the category field in the property sheet will not be pyramid.
3 Pyramid Settings
There are three pyramid resampling methods available: Nearest neighbor, Bilinear, and three-time convolution. The nearest neighbor method is the default, and is typically applied to any type of raster dataset. However, it is recommended that you apply the nearest neighbor method to discrete (nominal) data or raster datasets with a colormap (such as land-use data, scanned maps, and pseudo-color images). For continuous data, such as satellite imagery or aerial photography, you should use bilinear interpolation or a double three-time convolution method. Although bilinear interpolation is performed faster, the results are not as sharp as the results of the double three Convolution method. Bilinear interpolation is recommended for 1-bit TIFF or IMG.
3.2 Mosaic and Raster DataSet origin
The origin of the raster dataset is the coordinates of the upper-left corner of the raster dataset. The structure of the pyramid starts at that coordinate and then continues to the right and downward. If you are mosaicking data to the left or up from the origin of the raster dataset, ArcSDE needs to transform the origin so that the origin is still in the upper-left corner. After you transform the origin of an existing raster dataset, ArcSDE needs to rebuild the pyramid. Rebuilding the pyramid can be time-consuming, especially if a large number of raster dataset source files (or other raster datasets) have been mosaicked into a raster dataset to make the raster dataset larger.
Because rebuilding the pyramid operation is time consuming, you should identify the coordinates of the upper-left raster of the raster dataset by analyzing the source data, and then enter that coordinate when you create the raster dataset. When you create a raster dataset, you set the X-and y-coordinates of the pyramid reference point yourself (see Creating a Raster DataSet) without using the upper-left coordinate of the first raster dataset you inserted. You can therefore avoid transforming the origin of a raster dataset by setting a reference point for the pyramid when you create the raster dataset.
3.3 Storage location of pyramids
No matter when or how pyramids are generated, the first step in the process is always to define the pyramids (by adding pyramids to the Mosaic DataSet property sheet), and then to generate pyramids (that is, raster datasets). By default, pyramids are generated and stored in a folder next to the location where the mosaic dataset is located. The name of this folder is the same as the name of the Geodatabase with the. overview extension. However, if the mosaic dataset is created in an ArcSDE geodatabase, pyramids will be created in that geodatabase.
The mosaic dataset references its source rasters, and if the pyramids are not saved in the geodatabase, you might prefer to store the pyramids with the source rasters. In addition, if you plan to use pyramids created in another mosaic dataset in a mosaic dataset, you might need to manage the location of the pyramids separately. Before you generate pyramids, you can use the Define Pyramids (Overview) tool to define the output location of the pyramid.
Define where pyramids are generated in the mosaic dataset
There are several parameters that control the area of the mosaic dataset used to generate the pyramid and the resulting pyramid level. By default, pyramids are generated for areas that are contained within the boundaries of the mosaic dataset. In some cases, however, you might want to control the generated zone. You can define rectangular regions by specifying the minimum, maximum x, and y coordinates in the spatial reference system of the mosaic dataset.
In addition, you may want to control the level or resolution of the build pyramid. You can define the base pixel size to use to generate pyramids. Use this option to perform preprocessing on the raster to help improve the display speed. For example, if a raster in a mosaic dataset is being ortho-corrected, pan-sharpened, and color-balanced, the amount of computation that is used to view the mosaicked image at a resolution of all three functions is very large. You may need to create a processed raster in an area that is frequently viewed. To do this, define the range first, and then define the pixel size (the resolution of the panchromatic image) where the pan-sharpening function is applied. By default, the tool will define pyramids for this area (starting at the defined resolution until the minimum pyramid size limit is reached). You can further control this by defining the number of levels to be generated. For example, you might only need to generate a pyramid at the resolution of a panchromatic image. To do this, you can specify a pyramid level.
You can also control whether the pyramid (pyramid) of the raster is considered in the pyramid generation. You might have generated the original pyramid (pyramid) using a resampling method that you did not like, or you might not need to use any pyramids (pyramid) for some other reason. Select the Force pyramid (Overview) tile parameter option in the Define Pyramid (Overview) Tool dialog box, which ignores all raster pyramids (pyramid) and defines the pyramid from a factor that is larger than the raster base pixel size. For example, if the coefficient is 3, and the base pixel size or grid is 1 meters, the pyramid is generated at 3 meters.
In addition, the mosaic dataset may be mixed by a large number of data sources and, for some reason, need to use part of the first pyramid (pyramid) rather than all. You can define the pixel size at which the pyramid starts.
3.4 Defining the size of the pyramid
It is recommended to use software to determine the size of the pyramid, but in some cases you may need to specify a specific size. You can control the size in three ways: by Rows and columns, by coefficients, or by compression. When you limit the physical dimensions of a pyramid by defining the best number of rows and columns, the larger the value, the larger the file, and the more likely it is to regenerate the file if any of the images below change. This value also affects the number of pyramid raster datasets that are created. If it is a larger number, a smaller pyramid is generated. If it is a smaller number, more files will be generated.
You can also control the number of pyramid raster datasets created by modifying the pyramid factor. This factor is used to determine the size of subsequent pyramids. For example, if the cell size of the first level is X and the pyramid factor is 3, the next pyramid pixel size will be 3x. The larger the coefficients, the fewer pyramid raster datasets are generated, but this affects the display speed of the mosaicked image. Typically, the more pyramid levels, the faster the display, because this reduces the number of resampling.
There are several compression methods that you can use to control the space required to store the pyramids. You can store an uncompressed pyramid, but this consumes the most disk space. It is recommended to use a format for compression. If you focus on saving the pixel values in the pyramid, it is recommended to use LZ77 compression because it is a lossless compression. In addition, JPEG compression can be used, although it is a lossy compression, but can provide a high compression ratio. If you use JPEG compression, you can control the quality and thus affect the size of the file. The higher the compression quality, the higher the image quality, but the smaller the file size.
3.5 Pyramid Image Quality
There are several image parameters in the Define Pyramid (Overview) tool that you can use to control the Resampling method used to create pyramids, control pyramid compression, and compress quality. Therefore, for thematic data, you can choose the nearest neighbor method, whereas for aerial photography or satellite imagery, you can use bilinear interpolation or three convolution interpolation methods. And as mentioned earlier, you can choose quality. As a result, you can choose not to compress, lossless compress, or compress to define the overall image quality. For example, you can choose JPEG compression with a quality of 75, as this provides the best data quality and file size. Of course, before you create any large mosaic dataset or any database, it is always recommended that you first create a smaller test case to define the best parameters.
3.6 Pyramid Generation Time
Adding data to a mosaic dataset is fast, because no pixel data is moved. However, when you create large mosaic datasets, you need to consider the build time of the pyramids in your plan. Most of the options mentioned above affect the build time, so estimating the length of time can be quite difficult. The more pyramids you need, the longer it takes. This is another reason to create a test case.
To use or provide a mosaic dataset, you do not have to wait until the pyramid is generated. After you create a mosaic dataset, you can access the dataset and its underlying rasters, and after the pyramid is built, you start to see the mosaicked images that are affected or created when the mosaic dataset shrinks further.
4. Experimental results
In conjunction with the description of the pyramid described in ArcGIS above, a series of experiments are carried out to illustrate the relationship between the original image and the Ovr,rdd file using specific data.
The experiment uses 24 images in 4 different raster dataset formats; some pyramids are not; images are standard tile-type ortho images, each of which is 5000x5000,8, 3-band, and below is 3 of the 24-piece data:
The following is a comparison of the data model after the establishment of compression, pyramid aspects of the specific situation.
The original data is in the MRSID format. Convert other formats that are used in ArcGIS Desktop 10.0. The following table lists the combinations used. The 75% compression value is the default for ArcGIS, which provides a reasonable compression and preserves image quality. (Note: The compression value represents the image quality and compression%).
In addition to the above two formats using lossy compression, this means that from the original pixel values can be changed. TIFF is uncompressed, and LZW-compressed TIFF is lossless, and these two formats consume most of the space on the disk. Using a lossless compression usually does not cause much compression unless there is a large area of nodata or the same value (such as a boundary).
Note that the Mrsid file is still minimal because of its unique wavelet compression. Some people ask why their mrsid files take a long time to add to the ArcSDE geodatabase, because you want to transfer the highly compressed format to a different, less compressed format. You can check the estimated uncompressed size in the Raster DataSet Properties dialog box.
The pyramid is a TIFF file that produces some use of bilinear resampling. The top two examples of the following pyramids are generated using JPEG compression and other TIFF files that use LZ77 compression lossless compression. In this case, the additional required storage space is 2.2% to 9%, and the size of the space depends on how the thumbnail is compressed. The third Test produces a pyramid that uses the same compressed JPEG compressed TIFF file. Here you can see the additional storage percent rise to nearly 30%, but you'll also notice that each pyramid size is the same as uncompressed TIFF files produced using the same pyramid compression
Next, you set up 9 mosaic datasets and add a collection of 24 raster datasets accordingly. Although the mosaic dataset does not contain actual pixel data and is relatively small on disk, it represents a total mosaic image:
30,000 columns and 20,000 rows 30,000 columns and 20000 rows
Each mosaic dataset is created for 6-8 seconds. This is a very simple operation and should not take a long time to complete. The overview is established with the Add Raster Mosaic DataSet tool and the check option, and the following information is logged:
As you can see, it takes very little time, less than 20 seconds for a raster dataset to be added to the Mosaic dataset and to calculate the required properties, such as the range and bounds of the pixel size, and you can see that the total size of the data on the disk does not affect the time it takes to add the data to the Mosaic dataset. The remaining time depends on determining how many overview are needed and generating them. For those with internal pyramid, such as JPEG 2000 and MRSID and external overview raster datasets, less overview is required. However, more overview are generated and the total time is increased.
From these numbers you can see that it takes much less time to add a raster mosaic dataset and then build the overview. This is one of the reasons why you can also add data to a mosaic dataset at the same time that you created overview.
A detailed overview of the raster data in ArcGIS