Parallel pipeline processing system

Image analysis – Image enhancement or restoration – Image filter

Reexamination Certificate

Rate now

  [ 0.00 ] – not rated yet Voters 0   Comments 0

Details

C382S261000, C708S308000, C708S316000

Reexamination Certificate

active

06647150

ABSTRACT:

FIELD OF THE INVENTION
This invention relates to a method and apparatus of digitally processing image data and, particularly, a method and apparatus for performing image processing using pipelined image processing operations combined together on a single integrated circuit.
BACKGROUND OF THE INVENTION
Pipelined image processing operations involve the connection of discrete storage and discrete pipelined image processing components which perform image operations. These image processing operations include, for example, image convolution, image warping, nonlinear image processing operations, and other specialized processes such as connected component analysis of binary images. These systems do not integrate the processing and storage circuitry into single devices. Typically, these components are connected together with external crosspoint switches or dedicated data flow routing. The use of discrete image storage and processing components results in large circuits that are difficult to fit on small printed circuit boards. In addition, the use of discrete components and longer circuit paths reduces the execution speeds of the processes.
Even discrete processing components, such as pipelined image convolution units, require additional external circuitry during implementation. For example, an image convolver chip typically has external routing to and from the chip and external image line delay elements and pixel delay elements. The line and pixel delay elements delay image data so that simultaneous operations can be performed on a pixel neighborhood. The routing and line delay circuitry increases the amount of circuit board area and reduces execution speeds.
Dedicated pipelined image processing components can provide very high performance for image processing operations at a reduced cost to performance ratio over general-purpose processors such as von Neumann and Harvard architectures. General-purpose processors typically cannot provide fundamental image operations with the same performance as pipelined architectures because general-purpose processors are limited to a data read-process-store operation for each pixel operation. Although performance of the general-purpose processors has improved with technological advances in caching and other memory management concepts, these processors are typically not suited for image operations due to the large amount of data that is processed.
Massively parallel processors and computers can provide very fast performance in comparison to general purpose processors that use von Neumann architectures, and can match the processing speeds of pipelined hardware in some instances. These devices, however, are typically very complex to program and expensive to implement. In addition, the process of providing image data to each of the processors in a timely manner before parallel execution begins is generally slow. The parallel processors first load all of the information into the processors, then execute the image processing operations, and then read the processed data out of the processors to external devices that will perform further operations on the image data. Although the parallel processors will execute the process step at rates that are typically much faster than their von Neumann counterparts, the parallel processors are typically limited by the read-process-store cycle.
Further, certain image processing operations are difficult to implement with parallel architectures. While image convolution and other filtering operations that use relatively small pixel neighborhoods can be implemented efficiently in parallel systems, operations such as image warping are still quite inefficient.
Image processing applications are typically separated into two components: fundamental, i.e.—front-end, image processing operations, such as filtering, feature extraction, image alignment, and arithmetic operations, and higher-level processes that operate on the processed image data from the front-end processing to fulfill the requirements of a particular application. In efficient implementations, the front-end processing is typically performed with dedicated hardware that can provide processing at a very high performance to cost ratio. The higher-level processing of the image data is typically performed with general-purpose processors because of their flexibility. Front-end processes typically incorporate the combination of many discrete steps, however, which makes the implementation of these processes in dedicated hardware slow and complicated.
SUMMARY OF THE INVENTION
The present invention is embodied in a pipelined parallel processor (PPP) integrated circuit. The PPP includes a filtering unit, internal routing circuitry such as a crosspoint switch, an internal frame store controller, and one or more function circuits. The function circuits may be, for example, arithmetic units, lookup tables, timing compensators, adders/subtractors, statistics modules, image shifting circuitry, and other useful processing devices. The components of the PPP are interconnected by the crosspoint switch which routes data between the frame store controller, the filtering unit, function circuits, external input channels, and external output channels. The internal frame store controller includes a plurality of programmable bi-directional buffer memories that are coupled to an external field or frame store memory. Each buffer memory is assigned to provide data to, or receive data from one of the PPP components by a controller.


REFERENCES:
patent: 4601055 (1986-07-01), Kent
patent: 4692806 (1987-09-01), Andersen et al.
patent: 5103306 (1992-04-01), Weiman et al.
patent: 5134480 (1992-07-01), Wang et al.
patent: 5208872 (1993-05-01), Fisher
patent: 5325449 (1994-06-01), Burt et al.
patent: 5359674 (1994-10-01), Van der Wal
patent: 5550937 (1996-08-01), Bell et al.
patent: 5561617 (1996-10-01), Van der Wal
patent: 5657402 (1997-08-01), Bender et al.
patent: 5841722 (1998-11-01), Willenz
patent: 0 626 661 (1994-11-01), None
EPO Search Report, Aug. 29, 2000.
EPO Search Report, Aug. 13, 2002.
SMI ASSP IP90C61, Enhanced Frame Memory Controller (EFMC, Description Form, Sumitomo Industries, Ltd., (Avail. 4th Qtr. 1995) (1pp).
SMI ASSP IP90C20, Median/Max/Min. Filter (RKFil), Description Form, Sumitomo Metals, (2pp).

LandOfFree

Say what you really think

Search LandOfFree.com for the USA inventors and patents. Rate them and share your experience with other people.

Rating

Parallel pipeline processing system does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Parallel pipeline processing system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Parallel pipeline processing system will most certainly appreciate the feedback.

Rate now

     

Profile ID: LFUS-PAI-O-3115615

  Search
All data on this website is collected from public sources. Our data reflects the most accurate information available at the time of publication.