Patent US4550437A successfully refreshed from Google Patents.
US4550437A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Apparatus for parallel processing of local image data
Assignee: Hitachi Ltd
•
Filing: 1982-06-17
•
Granted: 1985-10-29
•
Expires: 2002-06-17
Abstract
Abstract An apparatus for parallelly processing image data from a local image data source, in which m rows x n columns of image data are processed by simultaneously processing n image data for each of the m rows (m and n being integers). The apparatus includes a member for serially receiving image data, a memory section having processing parameter data stored therein, a plurality of juxtaposed processor stages coupled to the memory section and arranged to be simultaneously supplied with sequentially shifted image data to thereby simultaneously produce processing results from the supplied image data and the processing parameter data, and an output section for receiving the outputs of the processor stages and delivering a processed output. The processing apparatus is suitable to be implemented in an IC form.
Inventors
- 👤 Yoshiki Kobayashi
- 👤 Tadashi Fukushima
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-23 13:22
CPC Classifications
G; G06; G06T; G06T5/00; G06T5/20
Claims Analysis
Total Claims: 19
Claim 1
Independent Claim
207 words
An apparatus for parallel processing of m rows by n columns (m, n: integers) of local image data from an image data source, comprising: means for receving said columns of local image data sequentially as series transmitted data; a plurality of data memories each having m weight data stored therein and each serving to store the m local image data from a respective one of said n columns serially introduced via said receiving means; a plurality of processor stages each including a multiplier and being provided one for each of said data memories for producing the products of each of said m image data with all of said m weight data in successive operations, said data memories and the corresponding processor stages constituting respective processor units connected in cascade; means for sequentially shifting said serially introduced image data between adjacent ones of said processor units to provide said sequential image data to be processed successively to said multipliers so that said products of said m image data with different ones of said m weight data are simultaneously produced from all of said processor units each time a shift of said image data is effected; and means for summing the outputs of said multipliers in said processor stages.
Claim 2
Independent Claim
207 words
An apparatus for parallel processing of m rows by n columns (m, n: integers) of local image data from an image data source, comprising: means for receving said columns of local image data sequentially as series transmitted data; a plurality of data memories each having m weight data stored therein and each serving to store the m local image data from a respective one of said n columns serially introduced via said receiving means; a plurality of processor stages each including a multiplier and being provided one for each of said data memories for producing the products of each of said m image data with all of said m weight data in successive operations, said data memories and the corresponding processor stages constituting respective processor units connected in cascade; means for sequentially shifting said serially introduced image data between adjacent ones of said processor units to provide said sequential image data to be processed successively to said multipliers so that said products of said m image data with different ones of said m weight data are simultaneously produced from all of said processor units each time a shift of said image data is effected; and means for summing the outputs of said multipliers in said processor stages.
Claim 3
Dependent Claim
55 words
An apparatus according to claim 1, in which each of said data memories includes a two-port RAM, one port for delivering renewed image data and weight data to be fed to an associated processor stage, the other port for delivering image data to be shifted to the data memory of the next adjacent processor unit.
Claim 4
Dependent Claim
78 words
An apparatus according to claim 1, in which each of said data memories includes a shift register and a one-port RAM, the arrangement being such that the renewed image data is supplied, via said shift register, to said RAM for storage and to the multiplier in the corresponding processor stage for production of said product and that image data stored in said RAM is shifted, via said shift register, to the data memory in the succeeding processor units.
Claim 5
Dependent Claim
99 words
An apparatus according to claim 1, in which each of said data memories includes a shift register for storing predetermined number of shifted image data and a one-port RAM having said weight data stored therein, the arrangement being such that the renewed image data is supplied from the shift register in the preceding processor unit to the first-mentioned shift register for storage and to the multiplier in the corresponding processor stage for production of said product and that image data stored in the first-mentioned shift register is shifted therefrom to the data memory in the next adjacent processor unit.
Claim 6
Independent Claim
220 words
An apparatus for parallel processing of local image data from an image data source, said local image data including m×n image data, where m and n represent the numbers of rows and columns covering a two-dimensional area of predetermined size, the apparatus comprising: means for receiving said local image data sequentially as serial data; n processor stages each including a data memory having m weight stored therein and each serving to store the m image data from a respective one of said n columns sequentially supplied via said receiving means, each processor stage including a multiplier coupled to said data memory for producing products of said image data and said weight data, said data memories in said processor stages further having m memory area respectively and being coupled to one another in cascade; means for sequentially shifting said sequentially supplied image data between the data memories of adjacent ones of said processor stages for each individual row of data so that each of said data memory areas has always one image data stored therein, thereby simultaneously effecting said production of products of said m image data with different ones of said m weight data in said multipliers each time a shift of said image data is effected; and means for summing the outputs of said multipliers in said processor stages.
Claim 7
Independent Claim
220 words
An apparatus for parallel processing of local image data from an image data source, said local image data including m×n image data, where m and n represent the numbers of rows and columns covering a two-dimensional area of predetermined size, the apparatus comprising: means for receiving said local image data sequentially as serial data; n processor stages each including a data memory having m weight stored therein and each serving to store the m image data from a respective one of said n columns sequentially supplied via said receiving means, each processor stage including a multiplier coupled to said data memory for producing products of said image data and said weight data, said data memories in said processor stages further having m memory area respectively and being coupled to one another in cascade; means for sequentially shifting said sequentially supplied image data between the data memories of adjacent ones of said processor stages for each individual row of data so that each of said data memory areas has always one image data stored therein, thereby simultaneously effecting said production of products of said m image data with different ones of said m weight data in said multipliers each time a shift of said image data is effected; and means for summing the outputs of said multipliers in said processor stages.
Claim 8
Independent Claim
149 words
An apparatus for parallel processing of m rows and n columns (m, n: integers) of local image data representing a kernel portion of a data frame from an image data source, comprising: means for receiving said kernel portion of local image data in sequence from said image data source; memory means having n weight data stored therein; n processor stages connected in cascade and each including a multiplier, the multipliers in said processor stages being connected to said receiving means so as to be simultaneously supplied with said image data representing said kernel portion via said receiving means and being coupled to said memory means for simultaneously producing during successive operations the products of each of said supplied local image data with a plurality of said weight data received from said memory means; and means for summing the outputs of said processor stages during each of said successive operations.
Claim 9
Independent Claim
149 words
An apparatus for parallel processing of m rows and n columns (m, n: integers) of local image data representing a kernel portion of a data frame from an image data source, comprising: means for receiving said kernel portion of local image data in sequence from said image data source; memory means having n weight data stored therein; n processor stages connected in cascade and each including a multiplier, the multipliers in said processor stages being connected to said receiving means so as to be simultaneously supplied with said image data representing said kernel portion via said receiving means and being coupled to said memory means for simultaneously producing during successive operations the products of each of said supplied local image data with a plurality of said weight data received from said memory means; and means for summing the outputs of said processor stages during each of said successive operations.
Claim 10
Independent Claim
175 words
An apparatus for parallel processing of local image data from an image data source, said local image data including m×n image data where m and n represent the numbers of rows and columns covering a two-dimensional area of predetermined size forming a kernel portion of a data frame, the apparatus comprising m image data processing modules in cascade connection each responsible for the image data of a different one of said rows, each of said m image data processing modules including: means for receiving one of the m rows of said local image data sequentially; memory means having n weight data stored therein; n processor stages juxtaposed and each having a multiplier coupled to said memory means, said multipliers in said juxtaposed processor stages being connected to said receiving means so as to be simultaneously supplied with said one row of said local image data to simultaneously produce products of said supplied image data and the weight data from said memory means; means for summing the outputs of said multipliers in said processor stages; and
Claim 11
Independent Claim
175 words
An apparatus for parallel processing of local image data from an image data source, said local image data including m×n image data where m and n represent the numbers of rows and columns covering a two-dimensional area of predetermined size forming a kernel portion of a data frame, the apparatus comprising m image data processing modules in cascade connection each responsible for the image data of a different one of said rows, each of said m image data processing modules including: means for receiving one of the m rows of said local image data sequentially; memory means having n weight data stored therein; n processor stages juxtaposed and each having a multiplier coupled to said memory means, said multipliers in said juxtaposed processor stages being connected to said receiving means so as to be simultaneously supplied with said one row of said local image data to simultaneously produce products of said supplied image data and the weight data from said memory means; means for summing the outputs of said multipliers in said processor stages; and
Claim 12
Dependent Claim
28 words
means for cumulatively adding the output of said summing means and the output of the summing means of the preceding image data processing module in said cascade connection.
Claim 13
Dependent Claim
28 words
means for cumulatively adding the output of said summing means and the output of the summing means of the preceding image data processing module in said cascade connection.
Claim 14
Dependent Claim
90 words
An apparatus according to claim 7 in which said receiving means in each of said image data processing modules includes n shift registers provided one for each of said processor stages, the shift register for the first processor stage being connected to sequentially receive said local image data from said local image data source, each of the other shift registers interconnecting two adjacent ones of said processor stages so that said sequentially introduced local image data is simultaneously supplied from said shift registers to said multipliers in said processor stages.
Claim 15
Dependent Claim
42 words
An apparatus according to claim 8, in which said memory means in each of said image data processing modules has such a structure that different one of said n weight data is supplied to each of said multipliers in said processor stages.
Claim 16
Dependent Claim
44 words
An apparatus according to claim 8, in which said memory means in each of said image data processing modules has such a structure that said n weight data are supplied in common to each of said multipliers in said processor stages one by one.
Claim 17
Dependent Claim
42 words
An apparatus according to claim 7, in which said receiving means in each of said image data processing modules includes a direct connection between each of said processor stages so that image data is sequentially supplied to said processor stages in common.
Claim 18
Dependent Claim
90 words
An apparatus according to claim 9, 10 or 11, in which in each of said image data processing modules: a pair of first pipeline registers are provided for each of said processor stages so that each processor stage receives input image data via one of said pair of first pipeline registers and delivers processor image data via the other of said pair of first pipeline registers; and a second pipeline register is provided for said summing means so that said summing means delivers its output via said second pipeline register.
Claim 19
Dependent Claim
104 words
An apparatus according to claim 12, in which a third pipeline register is provided for said adding means in each of said image data processing modules so that said adding means receives the output of the adding means of the immediately preceding image data processing module via said third pipeline register, and each of the image data processing modules except the most preceding image data processing module is provided with a delay register for delaying supply of the output of the summing means for ensuring cumulative addition of the outputs of the adding means of said m image data processing modules in cascade connection.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1988-10-28 | 1989-04-29 | 1989-10-29 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1992-10-28 | 1993-04-29 | 1993-10-29 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 1996-10-28 | 1997-04-29 | 1997-10-29 | $7700.00 | Window Closed / Lapsed |