harmonization of hdf with national imagery transfer format (nitf) 2.1

20
Harmonizing HDF5 and NITF Alan M. Goldberg HDF Workshop X November 2006 [email protected] The opinions expressed in this presentation are those of the author. They do not represent any organization or agency. opyright ©2006. All rights reserved.

Upload: the-hdf-eos-tools-and-information-center

Post on 26-May-2015

101 views

Category:

Technology


0 download

DESCRIPTION

HDF5 has already demonstrated its ability to adapt to diverse applications, and to integrate with other standards, e.g., netCDF. The National Imagery Transfer Format (NITF) is another format which might benefit from HDF5 as it evolves. NITF is the mandated standard for formatting digital imagery and imagery-related products, and exchanging them among the DoD and a number of US government agencies. Although NITF has been improved over the years, and although designed to be extensible, there are technical and conceptual limits to its original paradigm: mono- and polychromatic images, symbols, text and associated data. NITF has been a mandated standard for many years, and enterprise architectures have been built around it. There are important reasons why it should be retained. New sensors and algorithms are at the verge of stressing the standard with multispectral, hyperspectral, extended response, variable scale, time series, radar, video, and multisensor fusion products. Metadata are becoming more complex as the need for annotation and supporting data grows. Some imagery-like products are already originated in HDF, and others would benefit from a flexible format such as HDF. Portions of the NITF user community are exploring ways to move beyond its limits, to incorporate enhanced metadata, and to set procedures for suitable product profiles. This presentation develops a mapping between HDF5 and NITF structures and features. It concludes with some ideas on how NITF could be extended by harmonization with HDF5, while affording minimal disruption to operational uses.

TRANSCRIPT

Page 1: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Harmonizing HDF5 and NITF

Alan M. GoldbergHDF Workshop XNovember 2006

[email protected]

The opinions expressed in this presentation are those of the author. They do not represent any organization or agency.

Copyright ©2006. All rights reserved.

Page 2: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Outline

Background on standards and NITF How HDF might help Issues & Conclusions

Page 3: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Purpose of data standardization Standards typically help in three ways:

supported data formats and methods of organization

annotation of the data profiles which limit the way in which data is used

Standards may give fixed specifications, permit variable specifications, or provide preferred implementations

Different standards provide different levels of detail, and provide different options to the user.

Page 4: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

New standards need to find a niche among existing standards

Other Standards

Specific Instance

Profile

My Standard

Foundation Standards

Other Standards

Other Standards

Other Standards

Any standard often has a complicated relationship with other standards and implementations

Meaning

Identification

Structure

Elements

Physical Representation

Page 5: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

National Imagery Transfer Format 10USC467. Definitions: The term “imagery” means … a

likeness or presentation of any natural or manmade feature or related object or activity and the positional data acquired at the same time the likeness or representation was acquired, including [air- or space-borne systems] or other similar means.

MIL-STD-2500C. NITF Ver. 2.1 for the NITF Standard “The NITF provides a common basis for storage and interchange of images and associated data among existing and future systems. The NITF can be used to support interoperability by providing a data format for shared access applications, while also serving as a standard file format for dissemination of images, graphics, text, and associated data.”

Also used in SDTS & known as ISO/IEC 12087-5:BIIF and NATO Secondary Imagery Format (STANAG 4545)

Page 6: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Essence of NITF NITF is based on the concept of an image with

annotation. It is strongly tied to the concept of a regular gridded 2-D data array, and often explicitly to a map projection. Defines metadata which is required to geolocate the

image and to understand its provenance and other, similar metadata attributes

Difficult to associate gridded ancillary data with the gridded image.

Requires much predefinition and can be inflexible. Recently extended by incorporating XML, but this

requires two levels of interpretations Provides other features, eg. compression, color space

Page 7: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

A NITF image

Page 8: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Complex displays are a NITF strength

Implementation of a multi-level NITF display in common display coordinates

Source this page and previous: ITT, ENVI Certified NITF/NSIF Module Guide, October 2003, http://www.ittvis.com/envi/pdfs/NITFguide.pdf

Page 9: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

NITF file structure

Segments establish the basic file contents Additional major data elements can be incorporated in DES Additional descriptive information can be incorporated in Tagged

Record Extensions (TREs) A TRE is a deterministic structure of text elements Recent TREs permit XML

DES and TRE content is defined in reference documents; not self-describing

Multiple Image Segments are overlays or complex data sets, as appropriate

Page 10: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Not our father’s ‘Imagery’ anymore

Imagery

ImageryData File

ImageryDisplay

can be can be

processes into

Topical data arraysMetadataSupporting data

RGB 2D regular arrayStill or motionAnnotations

built of built of

When imagery was 10b 1- or 3-color, data set and display were the same.New imagery is arrays over 2D space, time, & color; more bits; RF; mosaics…

Derived dataCollections

Page 11: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Some current constraints NITF is a mandated standard NITF has large installed operational

base: users, libraries, industry NITF has conceptual limitations

Fixed format components Limited data model Display paradigm Size limits (10GB IS, 1GB DES)

Original data arriving in other formats

Page 12: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Potential georeferenced standards to supplement NITF XML is more flexible, but only incorporates ASCII data;

requires strong profile. Advanced Authorizing Format is based on the concept of

multiple time series, primarily video or audio. geoTIFF seems to have functionality similar to NITF HDF-EOS is based on geolocated 1D-2D arrays and other

ad hoc structures. netCDF is based on n-D arrays, with strong metadata

requirements and even stronger conventions. FGDC has developed metadata content standards,

related to this issue. HDF is extremely flexible, enabling all of the above.

However, it also requires a strong profile to avoid becoming uncontrolled.

Page 13: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Approaches to harmonization There are several ways to go:

One way is to use the best portions of NITF and other standards to establish a strong profile for HDF. E.g., NITF tags can be directly mapped to NITF attributes.

Another is to use XML to establish a framework within which data arrays can be created in HDF or some other data-oriented standard.

Example of NITF to HDF mapping File header mostly attributes Image Seg subhead mostly attributes, some data space Image Seg TREs mostly attributes Image Seg data array(s) Graphics Seg array wraping a shapefile Text Seg array of CHAR, or atttributes Ext Seg n/a Reserved Seg n/a

Page 14: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Some options for HDF in a NITF environment

HDFFile

FileHdr.

NITF 2.1 File

DESSubhd

HDF DES

HDFFile

HDFRef.

FileHdr.

NITF 2.1 File

DESSubhd

HDF DES

HDFFile

NITFAttrib.

ImageFileHdr.

NITF 2.1 File

ISSubhd

Image Seg.

HDFFile

File Hdr.

NITF 2.1 File

User Block

1. NITF wraps HDF

2. NITF points to HDF

3. NITF coexists with HDF

4. NITF in the HDF user block

Pervasive problem is difference between NITF fixed structure, and HDF API paradigms.

* ** Possible XML aid to HDF

Page 15: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Conclusions HDF5 was designed with sufficient flexibility

to handle the data engineering needs of NITF users.

An appropriate profile would permit NITF data content to be stored in HDF5.

HDF5 would permit emerging & related extensions which are difficult for NITF.

NITF is a mandated, interoperable standard. Converting operational users will not be easy.

Are there creative solutions out there?

Page 16: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Backup

Page 17: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

New standards need to find a niche among existing standards

Other Standards

Specific Instance

Profile

My Standard

Foundation Standards

Other Standards

Other Standards

Other Standards

Any standard often has a complicated relationship with other standards and implementations

Meaning

Identification

Structure

Elements

Physical Representation

HDF NITF

Page 18: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

AbstractThe National Imagery Transfer Format (NITF) is the mandated standard for formatting digital

imagery and imagery-related products, and exchanging them among the DoD and a number of US government agencies.

Although NITF has been improved over the years, and although designed to be extensible, there are technical and conceptual limits to its original paradigm: mono- and polychromatic images, symbols, text and associated data. NITF has been a mandated standard for many years, and enterprise architectures have been built around it. There are important reasons why it should be retained.

New sensors and algortihms are at the verge of stressing the standard with multispectral, hyperspectral, extended response, variable scale, time series, radar, video, and multisensor fusion products. Metadata are becoming more complex as the need for annotation and supporting data grows. Some products are arriving from sensors as HDF files, and others would benefit from a flexible format such as HDF.

Portions of the NITF user community are exploring ways to move beyond its limits, to incorporate enhanced metadata, and to set procedures for suitable product profiles. This presentation develops a mapping between HDF5 and NITF structures and features. It concludes with some ideas on how NITF could be extended by harmonization with HDF5, while affording minimal disruption to operational uses.

Page 19: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

New standards need to find a niche among existing standards

My Standard

Basic Standards

Peer Standards

Applied Standards

Profiles

Hardware

Application

Page 20: Harmonization of HDF with National Imagery Transfer Format (NITF) 2.1

Problems

NITF has a large embedded base supported by industry. Difficult to reorient

NITF has a fixed data layout and structure. HDF is defined by its API.

There are conceptual differences between the standards, not just data engineering differences.