Circuit for motion estimation in digitized video sequence encoders

• US 6,122,320 A
• Filed: 11/13/1998
• Issued: 09/19/2000
• Est. Priority Date: 03/14/1997
• Status: Expired due to Term

• Current Assignee: Cselt - Centro Studi E Laboratori Telecommunicazioni SPA
• Original Assignee: Cselt - Centro Studi E Laboratori Telecommunicazioni SPA
• Inventors: Marchisio, Mauro, Torielli, Alessandro, Nicoulaz, Didier, Dogimont, Stephanie, Gumm, Martin, Bellifemine, Fabio, Mattavelli, Marco, Bollano, Gianmario, Mombers, Frederich, Finotello, Andrea, Gandini, Marco, Garino, Pierangelo
• Primary Examiner(s): Kelley, Chris S.
• Assistant Examiner(s): An, Shawn S.
• Application Number:
  US09/180,781
• Time in Patent Office: 676 Days
• Field of Search: 375/240, 348/416, 348/402, 348/420, 348/407, 348/699, 382/236
• US Class Current: 375/240
• CPC Class Codes: G01R 31/318533 using scanning techniques, e.g. LSSD, Boundary Scan, JTAG
• CPC Class Codes: H04N 19/43 Hardware specially adapted for motion estimation or compensation
• CPC Class Codes: H04N 19/51 Motion estimation or motion compensation
• CPC Class Codes: H04N 19/56 Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search

Abstract

The circuit for motion estimation in digitised video sequence encoders comprises at least an integrated circuit component (IM, IM1 . . . IMn) which is arranged to perform either the function of determining motion vectors and associated costs for different prediction modes, or the function of vector refinement, possibly in addition to prediction mode selection. The circuit (IM) is based on the use of two operating units (M1, M2) which are arranged to concurrently process in different ways different pixel groups according to a MIMD technique. Preferably, when the circuit performs motion vector determination, the operating units (M1, M2) are programmed to execute a genetic algorithm exploiting an initial vector population taking into account the temporal and spatial correlations in the picture.

Claims

1. Circuit for motion estimation in a video encoder intended to encode a sequence of digitised pictures, which circuit receives aggregates of pixels of a current picture and, for each of said aggregate of pixels (hereinafter referred to as `current aggregate`), generates one or more vectors representing the displacement of said aggregate with respect to a corresponding aggregate of pixels (hereinafter referred to as `reference aggregate`) in a search window belonging to a reference picture, characterised in that it comprises at least one integrated circuit component (IM;

IM1 . . . IMn) including;

at least a pair of operating units (M1, M2) which are independently programmable and are arranged to concurrently process different data of the current picture and of the search window according to a multiple instruction, multiple data processing technique, to perform one of the following functions;

determination of a set of candidate motion vectors for each current aggregate and of an estimation error associated with each candidate vector, each vector in the set being the vector associated with the lowest estimation error for a given prediction mode chosen within a pre-defined set of prediction modes identifying a reference aggregate that best matches the current aggregate;

vector refinement, for improving the identification of the best-matching reference aggregate by effecting a further search in a limited portion of the window itself and by weighting each candidate vector by a weight associated with the amount of bits required for transmitting the vector itself;

a first internal bus (15), to which said operating units (M1, M2) are connected and which is intended for transferring to the component (IM;

IM1 . . . IMn) programming and initialisation information or processing parameters coming from an external controller (CP), for conveying monitoring information from the component (IM;

IM1 . . . IMn) to said external controller and for allowing information exchange between different units in the circuit;

a second internal bus (16), to which said operating units (M1, M2) are connected and which is intended for conveying video data, processing parameters and vectors associated with the relevant errors obtained as processing result, which are to be written into an external memory (SD) associated with the component (IM, IM1 . . . IMn) or which are read from said external memory (SD);

first interface means (IC) for connecting said first internal bus (15) to a line (12) for information exchange between the component (IM;

IM1 . . . IMn) and said external controller (CP);

second interface means (B1 . . . B3, IB) for connecting said second internal bus (16) to input/output lines (6 . . .

     10) conveying to the component (IM;

IM1 . . . IMn) or outputting from the component (IM;

IM1 . . . IMn) video data, processing parameters and processing results;

third interface means (IS) for connecting said second internal bus (16) to a line (11) conveying video data, processing information and vector/error pairs to and from said external memory (SD), and for controlling the access to the external memory (SD);

an internal controller (GC), which is connected to said first and second internal buses (15,

     16) and which has to;

identify and enable circuit units concerned by a data transfer over said second bus (16);

allot different portions of a processing cycle to different units;

generate and distribute the general synchronisation to the circuit units;

control the global flow of operations in the circuit;

synchronise the internal circuit activity with an external synchronisation; and

initiate the operations of the operating units (M1,M2).

2. Circuit according to claim 1, characterised in that it further comprises an internal memory (MP) for storing weights to be used in vector refinement, supplied by the external controller (CP), said internal memory being accessible in mutually exclusive way by said operating units (M1, M2) via said first bus (15).

3. Circuit according to claim 1, characterised in that it further comprises a testing unit (TE) arranged to perform a self-testing of said operating units (M1, M2), of memory means included in said second interface means (B1, B2, B3;

IB) and in said internal controller (GC), and--if present--of said internal memory (MP), the testing unit (TE) performing such self-testing in response to a specific activating instruction and comprising;

test control registers (TCR), storing a plurality of bits each enabling a particular test, which can be either a memory built-in self test or a software controlled built-in self test;

test result registers (TRR) storing a plurality of bits each indicating the result of particular test;

test status registers (TSR) storing a plurality of bits each indicating the running/not running status of a particular test;

a test access control register (TAC), providing the access to the test control, test result and test status registers (TCR, TRR, TSR) to control the tests and to evaluate the test results, providing the serial-to-parallel conversion of serial test data received by the unit and the parallel-to-serial conversion of test data to be outputted, and providing access to said first and second buses (15,

     16) to enable reading from and writing into units connected to said buses, for testing purposes.

4. Circuit according to claim 3, characterised in that said memory means and said internal memory (MP) are equipped with built-in self-test modules, and in that the test control, test result and test status registers (TCR, TRR, TSR) are directly connected to said built-in self-test modules and to said operating units (M1, M2) through a memory test bus (18) for controlling the tests and receiving test results and status information.

5. Circuit according to claim 1, characterised in that said operating units (M1, M2) are programmable to perform a selection of the prediction mode, to determine the motion vector corresponding to the prediction mode which results in the better trade-off between encoding quality and amount of information necessary to represent the encoded picture.

6. Circuit according to claim 5, characterised in that it comprises different integrated circuit components for performing candidate vector determination, vector refinement, selection of the prediction mode.

7. Circuit according to claim 5, characterised in that the selection of the prediction mode is carried out by the component(s) performing vector refinement or by the component(s) performing candidate vector determination, in a circuit in which vector refinement is not carried out.

8. Circuit according to claim 1, characterised in that it comprises a first plurality of identical integrated circuit components (IM1 . . . IMn) which are connected in cascade and are all programmed to perform the same function of vector determination or vector refinement, possibly jointly with the prediction selection function, and which receive the same data and share the same external controller (CP).

9. Circuit according to claim 1, characterised in that it comprises a first plurality of identical integrated circuit components (IM1 . . . IMn) which are connected in cascade and are all programmed to perform the same function of vector determination, and, in series to the first plurality, a second plurality of identical integrated circuit components (IM1 . . . IMn) which are connected in cascade and are all programmed to perform the same function of vector refinement, possibly jointly with the prediction selection function.

10. Circuit according to claim 8, characterised in that all components in said first and/or second plurality share the same external memory (SD) and operate on a same search window.

11. Circuit according to claim 8, characterised in that the components in the first and/or second plurality are associated each with an own external memory (SD) and are arranged to operate on different search windows.

12. Circuit according to claim 8, characterised in that a component (IM1 . . . IMn) upstream in the, or a, plurality directly feeds a downstream component with vector/error pairs, obtained as a result of the processing carried out in said upstream component, through a dedicated line (100) or through the same line (9, 10) which conveys processing parameters, the last circuit in the plurality emitting the final result of the processing on the output line (10) for the processing parameters.

13. Circuit according to claim 11, characterised in that a component (IM1 . . . IMn) upstream in the or a plurality feeds a downstream component with vector/error pairs, obtained as a result of the processing carried out in said upstream component, through said external memory (SD).

14. Circuit according to claim 1, characterised in that, for the determination of the candidate vectors, said operating units (M1, M2) are programmable to execute a genetic algorithm which uses a vector population including a first group of vectors that take into account the temporal and spatial correlations and that, at the end of a processing cycle, are transferred from the operating units (M1 . . . M2) to the external memory (SD) whence they are read by the operating units (M1 . . . M2) at the beginning of a subsequent cycle and used to build an initial vector population for said subsequent cycle.

15. Circuit according to claim 14, characterised in that said operating units (M1, M2) are arranged to write into said external memory (SD), at the end of the processing of a current aggregate of pixels, the vector associated with the lowest error for a predetermined prediction mode, said vector being read by said operating units (M1, M2) when an initial vector population is to be created.

16. Circuit according to claim 1, characterised in that each operating unit (M1 . . . M2) comprises:

a vector generation unit (GV) which is to generate motion vectors representing displacement of a current aggregate of pixels with respect to a plurality of reference aggregates which are tested in the search window, by executing instructions loaded into the unit (GV) during an initialisation phase and relevant to a pre-determined algorithm for candidate vector determination or refinement or yet prediction selection;

first and second internal memory units (MM, MF) connected to said second bus (16) and respectively intended to store said current aggregate of pixels and the pixels of the respective search window, supplied by said external memory;

an address generation unit (GIN), which is to generate the read and write addresses for said first and second internal memory units (MM, MF) starting from vectors supplied by the vector generation unit (GV);

a pixel processor (CE, CCE) which receives from said memory units (MM, MF) pixels of the current aggregate and of a reference aggregate and from said vector generation unit (GV) the vectors and information about the kind of operation the pixel processor (CE) is to perform, the pixel processor performing on said pixels arithmetical operations required either for determining said estimation error, by comparing groups of pixels in the current aggregate with homologous groups of pixels in a number of reference aggregates identified in the search window by vectors generated by said vector generation unit (GV), or for selecting the proper prediction mode, the pixel processor calculating said estimation error either with respect to an entire aggregate or to different portions of the aggregates;

a sorting unit (SU), which is connected to said pixel processor (CE;

CCE) from which it receives the motion vectors and the relevant errors together with information on the operating mode, and comprises means (CR, CL) for separating the vectors according to the prediction mode, recognising the vector associated with the lowest error for each prediction mode and storing such vector as the candidate vector for that prediction mode together with the respective error, said separating, recognising and storing means (CR) being accessible by said external memory (SD) and by other units in the circuit for writing vector/error pairs computed during processing of previous aggregates and for reading vector/error pairs retained at the end of the processing relevant to the current aggregate, and by said vector generation unit (GV) for reading vector/error pairs written by the external memory (SD);

a third internal bus (20) to which said address generation unit (GIN), said pixel processor (CE, CCE) and said sorting unit (SU) are connected, said third bus (20) being associated with means (IBU) for establishing connection between said third bus (20) and said first internal bus (15), or between said third bus (20) and said vector generating unit (GV), or between said first internal bus (15) and said vector generating unit (GV).

17. Circuit according to claim 14, characterised in that each operating unit (M1, M2) further comprises a generator (GVC) of random vectors, connected to said third internal bus (20) to supply the vector generation unit (GV) with random vectors to be used in building a vector population required for performing said genetic algorithm for the determination of the candidate vectors.

18. Circuit according to claim 17, characterised in that said random vector generator (GVC) is a generator of random numbers with a pseudo-Gaussian distribution and is arranged to generate a configurable distribution for both a horizontal and a vertical component of each vector.

19. Circuit according to claim 16, characterised in that said sorting unit (SU) also comprises means (SR, SL) for sorting the vectors provided by the pixel processor (CE, CCE) also according to the value of one among the estimation errors computed by said processor and for storing, at the end of the processing concerning a aggregate of pixels, a number of vectors exhibiting the lowest errors together with the respective errors, said sorting and storing means (SR, SL) being accessible by said external memory (SD) and by other units in the circuit for writing vector/error pairs computed during processing of previous aggregates and for reading vector/error pairs retained at the end of the processing relevant to the current aggregate, and by said vector generation unit (GV) for reading of vector/error pairs written by the external memory (SD) in an initialisation phase of the building of the vector population.

20. Circuit according to claim 16, characterised in that said vector generation unit (GV) comprises a pipelined RISC processor including a first bank of registers (GPR) comprising two sets of registers of which one (`vector registers`) is arranged to store vectors and the other (`error registers`) is arranged to store errors, each vector register being in turn divided into two parts respectively storing a horizontal component and a vertical component of the respective vector;

a same logical address being allotted to a vector register and an error register so as to establish and keep association between each vector and the relevant error, selection of data to be processed out of a vector register or an error register associated with a same address being carried out by means of an operation code contained in instructions stored in the unit.

21. Circuit according to claim 16, characterised in that said pixel processor (CE, CCE) comprises a pipelined arithmetic unit (CE) arranged to receive from said first and second memory units (MM, MF), for said error calculation, corresponding groups of pixels belonging to the current aggregate and to a reference aggregate, and to perform on said pixels the following operations:

a1) letting through unaltered the pixels in a group belonging to a reference aggregate;

or

a2) effecting an interpolation between adjacent pixels in a group or between pixels in subsequent groups, execution of operations a1) and a2) or of different interpolations in operation a2) being mutually exclusive and being determined by control signals provided by a control unit (CCE) in the processor;

b) comparing homologous pixels in said corresponding groups and calculating an error relevant to a group;

c) accumulating the errors relevant to all groups in a aggregate or in a aggregate portion to obtain an accumulated error;

d) averaging the accumulated error.

22. Circuit according to claim 21, characterised in that said arithmetic unit (CE) is arranged to calculate, before the error calculating operation in step b), a mean value of the pixel aggregate or aggregate portion and to subtract said mean value from the error calculated.

23. Circuit according to claim 21, characterised in that said arithmetic unit (CE) is arranged to calculate, in a circuit performing prediction selection, a cost associated with an intra-picture encoding, said cost calculation requiring comparing pixels in a current aggregate with a mean value thereof, accumulating the comparison results and averaging the accumulated value.

24. Circuit according to claim 16, characterised in that each operating unit (M1, M2) further comprises registers (RL), also connected to said third internal bus (20), for temporary storage of information which is provided by said internal controller (GC) and which is to be read by said vector generation unit (GV).

25. Circuit according to claim 16, characterised in that said second memory unit (MF) is made of a group of independently accessible memory elements, providing in the whole a storage capacity as required for storing the pixels of a search window comprising, at least for the generation of candidate vectors, a significant number of said reference aggregates;

and in that the addressing means (GIN) for said second memory (MF) are arranged to address said elements so that each writing operation concerns individual memory elements and causes updating of at least one entire element, and reading concerns a plurality of memory elements and randomly chosen aggregates of pixels, and causes simultaneous transfer of all pixels in a group to said pixel processor (CE).

26. Circuit according to claim 16 if referred to claims 3 and 4, characterised in that said operating units (M1, M2) further comprise a memory testing unit (UTM), connected on the one side to said testing unit (TE) for receiving control information and sending test results and status information, and connected on the other side to individual elements in said first and second memory units (MM, MF) for addressing a specific memory element concerned by the test, sending test data to the element and receiving test results therefrom.

27. Circuit according to claim 26, characterised in that said memory testing unit (UTM) is connected to said third internal bus (20) for access to and testing of an instruction memory of said vector generation unit (GV).

28. Circuit according to claim 1, characterised in that said second interface means (B1 . . . B3, IB) comprise first connection means (B1 . . . B3) for connection of said second internal bus (16) to input/output lines (6, 7, 8) conveying video data, said first connection means (B1 . . . B3) being arranged to connect said second bus (16) to two input lines (6, 7), one of which supplies the component (IM, IM1 . . . IMn) with pixels of a current picture, whatever the function performed by the component (IM;

IM1 . . . IMn), and the second (7) is active when the component (IM;

IM1 . . . IMn) performs vector refinement and supplies it with the pixels of a reference picture.

29. Circuit according to claim 28, characterised in that said second interface means (B1 . . . B3, IB) comprises second connection means (IB) for connection of said second internal bus (16) to input/output lines (9, 10;

100) conveying processing parameters and/or processing results, said second connection means (IB) being also connected to said first internal bus (15), for allowing the component (IM) to receive processing parameters from the external controller (CP).

30. Circuit according to claim 29, characterised in that said second connection means (IB) comprise:

first input and output interfaces (IIF1, UIF1) for data parallelism conversion between the formats used on said input/output lines (9, 10;

     100) conveying processing parameters and/or processing results and on said second bus (16) and for data synchronisation;

buffer means (IBB) for temporary storage of the processing parameters or vector/error pairs obtained as processing result in the component or coming from upstream components (IM, IM1 . . . IMn);

a control unit (IBC) which is to;

address the buffer means (IBB) according to an addressing law depending on the function performed by the component and the kind of information;

extract from a line (9) conveying processing parameters to the component and store information specific for a picture, to be made available to all other units in the component;

organise information to be emitted onto a line (10) outputting processing parameters and vector/error pairs;

control read and write operations involving said first internal bus (15).

31. Circuit according to claim 30, characterised in that said buffer means (IBB) comprise a plurality of buffer units (B4 . . . B7) where:

for candidate vector determination, said buffer units (B4 . . . B7) are organised into two distinct groups of which one is always dedicated to the information to be organised and emitted by the control unit (IBC) and the other manages vectors and errors;

for vector refinement, one unit stores incoming processing parameters and the other units store vector/cost pairs to be sent to other units in the component, the units (B4 . . . B7) being cyclically entrusted with the storage of the processing parameters in such a manner that reading of the parameters stored in a unit takes place after all units have been involved in a writing cycle.

32. Circuit according to claim 8, characterised in that the second connection means (IB) further comprise:

a comparison logic (CM), which is enabled in the components of a circuit comprising at least one of said first and second pluralities of cascaded components and which receives vector/error pairs locally calculated and vector/error pairs calculated by an upstream component; and

which compares the errors of homologous vectors and propagates to a downstream component the vectors associated with the lower costs;

second input and output interfaces (IIF2, UIF2) interfacing said comparison logic (CM) with a dedicated line (100) intended for transferring the vectors and the costs between subsequent components in the plurality.

33. Circuit according to claim 1, characterised in that said internal controller (GC) comprises:

an instruction sequencer (SEQ), storing a group of instructions, loaded into said sequencer (SEQ) during an initialisation phase of the circuit and concerning transfer of pixels and/or vectors over said second internal bus (16) or yet transfer of processing parameters into or between registers of the internal controller itself;

a program counter (CPR) which controls sequential execution of said instructions;

a control logic (UCC), activated upon receipt of a start signal indicating that the internal controller (GC) is ready to operate, and comprising decoding means, which decode the instructions generated by the sequencer (SEQ), and a sequential logic which, for each instruction, is to enable other units in the component to perform an operation specified in said instruction and to drive said program counter (CPR), at the end of a time allotted to execution of said instruction, so that it makes the instruction sequence to proceed, evolution of the sequential logic being controlled by synchronism signals (SYN) generated in the internal controller (GC);

a second bank of registers (RG1), including configuration registers, intended to store permanent information supplied by said external controller, as well as registers intended to store variable information which are specific for a pixel aggregate or a picture, one register in the bank providing the control logic (UCC) with said start signal;

a group of counters (CNT), which is to;

control, in a manner programmable by the configuration registers in the second bank of registers (RG1), the start of processing and data output operations;

generate said synchronism signals (SYN) and to activate different units in the component at instants which also are programmable by said configuration registers, said counters (CNT) receiving, for the execution of such tasks, synchronism signals associated with incoming data and external clock signals (CK);

a third bank of registers (RG2), storing said operating parameters concerned by parameter transfer instructions generated by said sequencer (SEQ), at least some of the registers in the third bank being accessible by other units in the component for reading operating parameters of interest for said other units or for writing operating parameters to be used in building output information governing pixel transfer over said second internal bus (16);

an output logic network (LUS) which comprises a multiplexing unit arranged to build said output information starting from values supplied by said counters (CNT) and from information stored at least in said second bank of registers (RG1).

34. Circuit according to claim 33, characterised in that said output logic network (LUS) further comprises selection means responsive to information about the type of operation provided by said control logic (UCC) and by said second and third bank of registers (RG1, RG2) to control building of said output information either from default values provided by said counters (CNT) and said second bank of registers (RG1), or from programmable values, provided by said third bank of registers (RG2).

35. Circuit according to claim 33, characterised in that said output information comprises:

the location of a picture in the external memory (SD);

the position of the current aggregate of pixels within a picture being processed;

the position of a picture region concerned by a pixel transfer operation; and

the size of the region.

36. Circuit according to claim 1, characterised in that said third interface means (IS) comprise:

an address generating network (SAG) which receives from the component (IM) information concerning the type of operation to be performed in the memory and a picture portion involved in the operation, and processes such information to convert it into;

addressing signals (ADD_-- R, ADD_-- C) which are specific for one of the following operations;

i) writing/reading pixels belonging to one or more pictures;

ii) writing/reading processing results;

iii) memory refresh; and

which allow access to the memory to write/read pixels related to picture portions whose sizes are communicated each time by the component (IM) and to write/read processing results concerning one or more pixel aggregates,

and into information (N_-- BURST, BURST_-- L) about the amount of data involved in an access to the memory for a write/read operation, the amount of data to be written/read at each access being variable in the course of the operation according to said information on the amount of data (N_-- BURST, BURST_-- L);

a command generating network (SSG) which receives the information and the addressing signals(N_-- BURST, BURST_-- L, ADD_-- R, ADD_-- C) generated by the address generating network (SAG) and processes them generating command signals which can be directly used by the memory.

37. Circuit as claimed in claim 36, characterised in that said address generating network (SAG) comprises:

a first combinatory logic circuit (OCD), which receives from the internal controller (GC) a code indicating the type of operation to be performed and an enabling signal enabling it to decode said operation code, thereby to emit signals indicating that the operation is a read, a write or a refresh operation together with configuration signals used by other units in the address generating network (SAG) to predispose themselves for a particular read/write operation;

a second combinatory logic circuit (FCA) which performs a first processing of the information concerning the picture portion involved in a reading or writing and which;

receives;

from the internal controller (GC), information about the position and the size of such picture portion;

from the first combinatory circuit (OCD), configuration signals concerning the characteristics of the picture and the operating modes;

from a control circuit (ISC), also belonging to the address generating network (SAG), a signal that indicates whether a pixel read/write operation concerns luminance or chrominance pixels; and

computes and emits;

information about the position in the picture of an initial pixel aggregate and of the first pixel of the initial aggregate, in the case of pixel reading/writing, or about the position in the picture of the pixel aggregate to which a processing result is associated, in the case of reading/writing of processing results;

information about the number of pixel aggregates involved; and

, in the case of a reading of a region that extends over multiple aggregates and involves only a subset of pixels in some of these aggregates, information about the size of the subset involved;

said control circuit (ISC), which is to;

follow the evolution of the operations in the memory according to the information provided by the first and second combinatory circuits (OCD, FCA), by the command generating network (SSG) and by the internal controller (GC);

provide said command generating network (SSG) with the signals indicating whether the operation is a reading, a writing or a refresh and with the information (N_-- BURST, BURST_-- L) about the amount of data; and

manage the dialogue with the component (IM) concerning the transfer of pixels or results of processing operations between the memory (SD) and the component (IM);

a third and a fourth combinatory logic circuit (ROW, COL) which, on the basis of the configuration signals and of control signals provided by the control circuit (ISC), convert position information within a picture, provided by the second combinatory circuit (FCA), into row and respectively column addresses for the memory (SD) and provide the command generating network (SSG) with such addresses, in the presence of a respective request for row or column address (NEWRAS, NEWCAS).

38. Circuit as claimed in claim 37, characterised in that said control circuit (ISC) comprises:

a counting circuit (CSC) which is to evaluate the number of pixel aggregates in the picture portion involved in a read/write operation, by processing, on the basis of information about the type of operation and of the configuration signals provided by the first combinatory circuit (OCD), the information about such number of aggregates provided by the second combinatory circuit (FCA), and which is also intended to follow the evolution of writing/reading operations starting from the address requests coming from the command generating network (SSG);

a fifth combinatory circuit (BUE), which is to generate and to send at least to said command generating network (SSG) the information (N_-- BURST, BURST_-- L) on the amount of data to be written or read, obtained from the position information within the picture provided by the second combinatory circuit (FCA) and from the information about the number of pixel aggregates provided by the counting circuit (CSC);

a first sequential logic circuit (SIG), which, by using the information about the type of operation provided by the first combinatory circuit (OCD), information about the progress of the read/write operations provided at least by the counting circuit (CSC) and the address requests provided by the command generating network (SSG), generates a group of signals (RD, WR, REF) for starting or stopping read, write and refresh operations in the memory (SD), to be provided to the command generating network (SSG), and a signal enabling either data transfer direction over said second internal bus (16);

a second sequential logic circuit (ENG) which, by using the information about the type of operation, configuration signals coming from the first combinatory circuit (OCD), the information (N_-- BURST, BURST_-- L) about the amount of data involved in an operation provided by the counting circuit (CSC) and the address requests coming from the command generating network (SSG), generates;

signals indicative of the progress of the operations, provided to the first sequential logic circuit (SIG) and to a third sequential logic circuit (AGG), also belonging to the control circuit (ISC), and said signal indicative of whether the operation concerns luminance or chrominance pixels, sent to said second combinatory circuit (FCA) and to the first sequential logic circuit (SIG);

signals for managing the dialogue between the external memory (SD) and the component (IM);

said third sequential logic circuit (AGG) which, in the case of data transfer from the memory (SD) to the component (IM), generates control signals for data storage into the operating units (M1, M2) of the component, on the basis of the information about the type of operation provided by the first combinatory circuit (OCD), of position information provided by the second combinatory circuit (FCA) and of information about the number of pixel aggregates involved in the operation, provided by said counting circuit (CSC).

39. Circuit as claimed in claim 36, characterised in that said command generating network (SSG) comprises:

an operating circuit (SSG1) which receives from the address generating network (SAG) the addresses, the signals indicating the type of operation and the information about the amount of data involved in the operation and emits the command signals for the memory (SD) and the requests for new addresses, and

a control circuit (SSG2) which supervises the operations of the operating unit and causes emission of the various command and request signals according to the evolution of the operations themselves.