ABSTRACT
The field of two-dimensional digital signal processing has been growing rapidly in recent years. Images such as satellite photographs, radar and sonar maps, medical X-ray pictures, radiographs, electron micrographs, and data from seismic, gravitational, and magnetic records are typical examples of 2-D signals that might need to be processed. Two-dimensional digital filters, like their 1-D counterparts, are discrete systems that can be linear or nonlinear, shift-invariant or shift-dependent, causal or noncausal, and stable or unstable. Two-dimensional digital filters can be classified as recursive or nonrecursive, depending on whether the output of the filter depends on previous values of the output; alternatively, they can be classified as infinite-impulse response ox finite-impulse response filters, depending on whether their impulse response is of infinite or finite duration. When a 2-D digital filter is implemented in terms of either software on a general-purpose computer or dedicated hardware, numbers representing transfer-function coefficients and signals must be stored and manipulated in registers of finite length.
