Filtering operation method for very high-speed image processing system

• US 5,315,699 A
• Filed: 03/11/1992
• Issued: 05/24/1994
• Est. Priority Date: 03/20/1991
• Status: Expired due to Term

• Current Assignee: Research Development Corporation Of Japan (Japan Science And Technology Agency)
• Original Assignee: Research Development Corporation Of Japan (Japan Science And Technology Agency)
• Inventors: Honsawa, Kunio, Imai, Masaharu, Tomita, Jota
• Primary Examiner(s): Shaw, Dale M.
• Assistant Examiner(s): Tung, Kee M.
• Application Number:
  US07/851,024
• Time in Patent Office: 804 Days
• Field of Search: 395/162-166, 395/275, 340/798-799, 358/22, 345/189, 345/190, 345/200, 345/203, 364/724.05, 364/724.01
• US Class Current: 345/502
• CPC Class Codes: G06T 5/20 using local operators

Abstract

A filtering system for very high-speed image processing performs such image processing in parallel for each unit sequentially in each line. The filtering system includes an input unit having input elements for taking up all or a part of image data of each line in the order of raster scan, a processing unit having processing elements for performing image processing computation in parallel for each pixel as image data from each input element is transferred simultaneously, an output unit having output elements where processing data from each processing element is simultaneously transferred, and a controller for controlling the input unit, the processing unit, and the output unit. Each processing element includes an endless memory having memory capacity for storing as much data as required of a row in charge of processing and has a terminal address connected to a top address, a base pointer for writing the top address next to the terminal address such that each content is changed for each line by a command from the controller, and a reference pointer for writing the content of the base pointer for each line and for specifying an address of the memory by changing each content of the data read from the endless memory. By this arrangement, a region of the endless memory corresponding to a mask size of the filter is sequentially accessed by the reference pointer.

Claims

1. A filtering system for very high-speed image processing performing image processing in parallel for each unit sequentially in each line, comprising:

an input unit having a plurality of input elements for taking up all or a part of image data of each line in the order of raster scan;

a processing unit having a plurality of processing elements for performing image processing computation in parallel for each pixel as image data from each input element is transferred simultaneously;

an output unit having a plurality of output elements where processing data from each processing element is simultaneously transferred; and

a controller for controlling the input unit, the processing unit, and the output unit;

each processing element comprising an endless memory having memory capacity for storing as much data as required of a row in charge of processing and having a terminal address connected to a top address, a base pointer for writing the top address next to the terminal address such that each content is changed for each line by a command from said controller, and a reference pointer for writing the content of the base pointer for each line and for specifying an address of said memory by changing each content of the data read from the endless memory, whereby a region of the endless memory corresponding to a mask size of the filter is sequentially accessed by the reference pointer.

2. A filtering system for very high-speed image processing performing image processing by a weighted filter having left-to-right symmetry locally and in parallel for each pixel sequentially in each line, comprising:

an input unit having a plurality of input elements for taking up all or part of image data of each line in the order of raster scan;

a processing unit having a plurality of processing elements for performing image processing computation in parallel for each pixel whereby image data from each input element is transferred simultaneously from each input element;

an output unit having a plurality of output elements where processing data from each processing element are simultaneously transferred;

a controller for controlling the input unit, the processing unit, and the output unit;

each of said plurality of processing elements including means for multiplying each pixel data of the row, for which it is in charge of processing, by a weight of each row at a center of said weighted filter, means for adding together and maintaining the results of multiplication of each pixel data of the row by the weight of each row at the center of said weighted filter for multiplying each pixel data of the row for which it is in charge of processing by a weight of other rows on one side of a central row of said weighted filter, and for adding together and maintaining the results of the multiplication by the weight of other rows on one side of the central row of said weighted filter, and wherein each of said plurality of processing elements executes a filtering operation by receiving the transferred data maintained by the processing element of the other rows.