Christopher M. Brislawn

FBI compression standard for digitized fingerprint images

The FBI has formulated national standards for digitization and compression of gray-scale fingerprint images. The compression algorithm for the digitized images is based on adaptive uniform scalar quantization of a discrete wavelet transform subband decomposition, a technique referred to as the wavelet/scalar quantization method. The algorithm produces archival-quality images at compression ratios of around 15 to 1 and will allow the current database of paper fingerprint cards to be replaced by digital imagery. A compliance testing program is also being implemented to ensure high standards of image quality and interchangeability of data between different implementations. We will review the current status of the FBI standard, including the compliance testing process and the details of the first-generation encoder.

FBI wavelet/scalar quantization standard for gray-scale fingerprint image compression

The FBI has recently adopted a standard for the compression of digitized 8-bit gray-scale fingerprint images. The standard is based on scalar quantization of a 64-subband discrete wavelet transform decomposition of the images, followed by Huffman coding. Novel features of the algorithm include the use of symmetric boundary conditions for transforming finite- length signals and a subband decomposition tailored for fingerprint images scanned at 500 dpi. The standard is intended for use in conjunction with ANSI/NBS-CLS 1-1993, American National Standard Data Format for the Interchange of Fingerprint Information, and the FBIs Integrated Automated Fingerprint Identification System.

Preservation of subband symmetry in multirate signal coding

The article derives necessary and sufficient conditions for a multirate linear phase FIR filter bank, acting on a symmetrically extended finite-length input signal, to produce either symmetric or antisymmetric downsampled subbands. Such techniques have gained popularity for image coding applications, and we provide the solution to a key technical problem in terms of the input and filter symmetries and the downsampling ratio.

The wavelet/scalar quantization compression standard for digital fingerprint images

A new digital image compression standard has been adopted by the US Federal Bureau of Investigation for use on digitized gray-scale fingerprint images. The algorithm is based on adaptive uniform scalar quantization of a discrete wavelet transform image decomposition and is referred to as the wavelet/scalar quantization standard. The standard produces archival quality images at compression ratios of around 20:1 and will allow the FBI to replace their current database of paper fingerprint cards with digital imagery.<<ETX>>

Revisiting wavelet compression for large-scale climate data using JPEG 2000 and ensuring data precision

We revisit wavelet compression by using a standards-based method to reduce large-scale data sizes for production scientific computing. Many of the bottlenecks in visualization and analysis come from limited bandwidth in data movement, from storage to networks. The majority of the processing time for visualization and analysis is spent reading or writing large-scale data or moving data from a remote site in a distance scenario. Using wavelet compression in JPEG 2000, we provide a mechanism to vary data transfer time versus data quality, so that a domain expert can improve data transfer time while quantifying compression effects on their data. By using a standards-based method, we are able to provide scientists with the state-of-the-art wavelet compression from the signal processing and data compression community, suitable for use in a production computing environment. To quantify compression effects, we focus on measuring bit rate versus maximum error as a quality metric to provide precision guarantees for scientific analysis on remotely compressed POP (Parallel Ocean Program) data.

Feature extraction from hyperspectral images compressed using the JPEG-2000 standard

We present results quantifying the exploitability of compressed remote sensing imagery. The performance of various feature extraction and classification tasks is measured on hyperspectral images coded using the JPEG-2000 Standard. Spectral decorrelation is performed using the Karhunen-Loeve transform and the 9-7 wavelet transform as part of the JPEG-2000 process. The quantitative performance of supervised, unsupervised, and hybrid classification tasks is reported as a function of the compressed bit rate for each spectral decorrelation scheme. The tasks examined are shown to perform with 99% accuracy at rates as low as 0.125 bits/pixel/band. This suggests that one need not limit remote sensing systems to lossless compression only, since many common classification tools perform reliably on images compressed to very low bit rates.

The polyphase-with-advance representation and linear phase lifting factorizations

A matrix theory is developed for the noncausal polyphase representation that underlies the theory of lifted filter banks and wavelet transforms. The theory presented here develops an extensive matrix algebra framework for analyzing and implementing linear phase two-channel filter banks via lifting cascade schemes. Whole-sample symmetric and half-sample symmetric linear phase filter banks are characterized completely in terms of the polyphase-with-advance representation, and new proofs are given of linear phase lifting factorization theorems for these two principal classes of linear phase filter banks. The theory benefits significantly from a number of group-theoretic structures arising in the polyphase-with-advance representation and in the lifting factorization of linear phase filter banks. These results form the foundations of the lifting methodology employed in Part 2 of the ISO/IEC JPEG 2000 still image coding standard.

Reflected boundary conditions for multirate filter banks

Several methods for applying perfect reconstruction quadrature mirror filter (PR QMF) banks to finite-length signals are described and compared. Although simple periodization produces a transform that does not increase the size of the transformed signal, it has the disadvantage of introducing a jump discontinuity at the signals boundary. Various methods of transforming smoother extensions are considered and analyzed in terms of their ability to conserve data storage costs and reproduce the signal in a numerically efficient manner. A complete classification of two-channel schemes based on periodizing symmetric (reflected) signal extensions and using linear phase filters is described, for both even- and odd-length signals. More general techniques based on transforming linear signal extrapolations and truncating the resulting subbands to conserve data size are also presented. An example using reflected boundary extension is discussed.<<ETX>>

Group Lifting Structures for Multirate Filter Banks I: Uniqueness of Lifting Factorizations

Group lifting structures are introduced to provide an algebraic framework for studying lifting factorizations of two-channel perfect reconstruction finite-impulse-response (FIR) filter banks. The lifting factorizations generated by a group lifting structure are characterized by Abelian groups of lower and upper triangular lifting matrices, an Abelian group of unimodular gain scaling matrices, and a set of base filter banks. Examples of group lifting structures are given for linear phase lifting factorizations of the two nontrivial classes of two-channel linear phase FIR filter banks, the whole- and half-sample symmetric classes, including both the reversible and irreversible cases. This covers the lifting specifications for whole-sample symmetric filter banks in Parts 1 and 2 of the ISO/IEC JPEG 2000 still image coding standard. The theory is used to address the uniqueness of lifting factorizations. With no constraints on the lifting process, it is shown that lifting factorizations are highly nonunique. When certain hypotheses developed in the paper are satisfied, however, lifting factorizations generated by a group lifting structure are shown to be unique. A companion paper applies the uniqueness results proven in this paper to the linear phase group lifting structures for whole- and half-sample symmetric filter banks.

Symmetric extension for lifted filter banks and obstructions to reversible implementation

Symmetric pre-extension is a standard approach to boundary handling for finite-length input vectors with linear phase filter banks. It works with both conventional linear implementations and the so-called reversible, or integer-to-integer, implementations of odd-length linear phase (whole-sample symmetric) filter banks. In comparison, significant difficulties arise when using symmetric pre-extension on reversible filter banks with even-length (half-sample symmetric) linear phase filters. An alternative approach is presented using lifting step extension, in which boundary extensions are performed in each step of a lifting factorization, avoiding some of these difficulties while preserving reversibility and retaining the nonexpansive property of symmetric pre-extension. Another alternative that is capable of preserving both reversibility and subband symmetry for half-sample symmetric filter banks is developed based on ideas from the theory of lattice vector quantization. The practical ramifications of this work are illustrated by describing its influence on the specification of filter bank algorithms in Part 2 of the ISO/IEC JPEG 2000 image coding standard.