An Application-Specific Compiler for High-Speed Binary Image Morphology

Abstract

This paper discusses a two-level compilation scheme used for generating high-speed binary image morphology pipelines from a textual description of the algorithm. The first-level compiler generates a generic morphology machine which is customized for the specified set of instructions by the second-level compiler. Because the generic machine is reused, we are able to avoid long synthesis times and achieve compile times similar to software compile times, while still achieving a 10X speed-up over the software implementation.

Reconfigurable Computing Architectures

“…As a result, the approach was able to achieve realtime matching on hundreds of 16 Â 16 pixel target images. The large compile time necessary to generate unique bitstreams for a set of image operations was avoided by using a generic architecture and specializing only the LUT ROMs to the particular operation [233].…”

Efficient implementation of greyscale morphological filters

“…Binary morphologic filters, because of their usefulness and relative simplicity, were some of the first image processing operations to be implemented on FPGAs [2,3]. Their regular structure makes a streamed pipeline implementation attractive, and most FPGA based filter implementations use this structure.…”

Reconfigurable architectures for parallel execution of image processing tasks

“…A number of projects on video systems were designed using software/hardware tools to deal with different aspects of the hardware/software relationships, the high-level design [26,27] hardware compilers [28], and the design of image analysis algorithms [29,30]. Current publications on the implementation of video systems and algorithms for image processing and analysis deal with new developments in the above directions of research: -problems of the implementation on reconfigurable systems for high resolution imaging [31,32], -the implementation of motion detection algorithms [33], -segmentation implementation [34], -neuron network implementations [35], -comprehensive imaging algorithms [36].…”