A Combined VLSI Architecture for Nonlinear Image Processing Filters

In recent years, nonlinear methods and techniques have emerged as intensive research topics in the fields of signal, image, and video processing. This paper describes the combined design of high performance architectures for different filter algorithms and structures used in nonlinear image processing targeted at computer vision. These types of filters include weighted order statistics (WOS), stack, nonlinear mean, Teager, polynomial and rational filters. This architecture is suitable for real time and high quality video imaging tasks and VLSI implementations. These architectures can be used as coprocessor subsystems, integrated with other modules, to form an overall high performance imaging pipe in a system-on-a-chip platform