As camera sensor technology has advanced, image sizes have grown and the methods employed to process imagery have become more advanced and complex. For infrared (IR) imaging systems, more processing is typically required than in electro-optical (EO) systems in order to adjust the data so that it is visibly more meaningful or to preserve details that may be lost in compression. JPEG and JPEG2000 typically achieve compression through a loss of higher frequencies, which may contain important information for a variety of applications. A common algorithm employed on IR imagery before compression is an unsharp mask, which is a type of sharpening that preserves and boosts higher frequencies in the IR imagery. For many wide-area surveillance applications, this is essential in order to preserve important content, which may consist of a few pixels. With very large images and a pipelined processing chain, software systems quickly saturate memory bandwidth and processing throughput. A solution to this problem is offloading algorithms to specialized processing cards, such as those with FPGAs or GPUs. When implementing an algorithm in hardware, floating point calculations are resource-hungry and time consuming to complete, so the calculations are typically done in fixed-point notation while scaling the data and filter coefficients up in order to preserve precision. This paper presents an integer-based unsharp mask technique and a hardware implementation that gives equivalent results to its software floating-point counterpart and allows for an embedded implementation to achieve computational speedup.
An integer-based unsharp mask algorithm for infrared imagery
2010-07-01
540104 byte
Conference paper
Electronic Resource
English
Rational unsharp masking technique
British Library Online Contents | 1998
|Nonlinear unsharp masking methods for image contrast enhancement
British Library Online Contents | 1996
|Unsharp Masking by the Rank-Order Filters with Spatially Adaptive Neighborhoods
British Library Online Contents | 2002
|Infrared testing and mask alignment.
NTRS | 1967
|Algorithm For Integer Cosine Transforms
NTRS | 1994
|