Raster Format Converter Circuit

1. Field of the Invention

As this invention is used to store data to frame memory when the data converted into block unit on digital HDTV(ATV) shall be displayed and also used as an address generator to convert data format into raster scan method via frame memory when, by using blocktransform including DCT(DISCRETE COSINETRANSFORM), VQ(VECTOR QUANTIZATION) etc., data format with raster scan method coming from camera shall be changed to data format of block unit, which is convenient to condense, through frame memory or data composed of block unit shall be displayed on screen after restoring the condensed data, it is concerning raster format converter circuit of HDTV applicable for all systems, which use blocktransform or DCT, VQ etc., i.e. HDTV(High Definition Television), ATV(Advanced Television), MPEG(Moving Picture Experts Group), JPEG(joint Photographic Experts Group), Video Phone etc.

2. Description of the Prior Art

In general, ATV(Advanced Television) is aimed to realize highly precise and large sized screen with increasing scanning line number more than twice and also aspect ratio comparing with the existing television.

With basic frame of 1125 scanning line number, 1025 active scanning line, 60 Hz field frequency, 2:1 interlaced scanning, 16:9 aspect ratio, pixel number, 1920 luminance signal, 960 color difference signal, its standardization is being led by CCIR(Comittee Consutatif International des Radiocommunications).

For one example of such HDTV art, the American patent application Ser. Nr. 07/721515 by The Glass Valley Group Inc. can be illustrated.

As a video signal communication method which transmits together data appointing the deleted part on wide screen and according to the transmit ted signal possibly receives via general TV with narrow screen, when receiving in the form of regular order after each time extruding of luminance component(Y) & color component(C1, C2) corresponding to TV signal of wide aspect ratio, this is developed under consideration of the interchangeability with MTSC television system.

Accordingly, as such existing art is for the interchangeability with NTSC television system, it could not suggest the art to convert data format into raster scan method through frame memory.

The purpose or this invention is to offer raster format converter system used as an address generator in frame memory, in HDTV condensing image data by using blocktransform including DCT(DISCRETE COSINE TRANSFORM), VQ(VECTOR QUANTIZATION) etc., and then to offer the method of luminance signal processing & color difference signal processing.

For this, raster format converter system of this convention consists or: 10 bit counter that uses "ON" signal which activates as "High" during data existing period so that can be counted only during the period while does as "Low" during data non-existing period, also uses macroblock start signal(MSC) which resets in the starting point of macroblock, and is used to count one macroblock; Y processor making horizontal & vertical address of luminance signal(Y) by using the said counter & slice number(SVP), macroblock number(MACRO #); C processor making horizontal, vertical address & luminance address of color difference signal(C) by using the said 10 bit counter and slice number(SVP), macroblock number(MACRO #); the 1st multiplexer that selects vertical address and offer Y address during luminance signal period and C address during color difference signal period; the 2nd multiplexer that selects and offers vertical address with selecting Y address during Y period and C address during C period.

Furthermore, the luminance signal processing method of this invention consists of: the 1st process that increases luminance vertical address until 8 line data is inputted with repeated input of 4 parallel data by 4 slices in regular order; the 2nd process that, after 8 line data input, first increases horizontal address until 4 block datas are inputted and also increases other illuminance vertical address and then increases until 8 block datas are all inputted; subsequently the 3rd process that, by using macroblock number, increases the rest luminance horizontal address until 44 macroblocks are all inputted: the 4th process that then repeats to increase the rest luminance vertical address by 4 slices in turn.

Moreover, the color signal processing method of this invention consists of: the 1st process that repeats to input 4 parallel datas by 4 slices in regular order and then increases color difference address until 8 line data is inputted: the 2nd process that, after 8 line data input, first increases other color difference horizontal address until 4 block datas are inputted and then counts 4 blocks: subsequently the 3rd process that, by using macroblock number, increases the rest color difference horizontal address until 44 macro blocks are all inputted; then the 4th process that repeats to increase the rest color difference vertical address by 4 slices in turn.

FIG. 1 is the schematic diagram of macroblock.

FIG. 2 is the schematic diagram of frame.

FIG. 3 is the schematic diagram of frame memory.

FIG. 4 is the first block diagram from slice 1 to slice 4 in frame.

FIG. 5 is the raster format converter circuit diagram of this invention.

FIG. 6 is the Y processor schematic diagram producing Y address in raster format converter of this invention.

FIG. 7 is the C processor schematic diagram producing C address in raster format converter of this invention.

FIG. 8 is the flow chart to explain the operating order of Y processor.

FIG. 9 is the flow chart to explain the operating order of C processor.

As FIG. 5 is the schematic diagram of raster format converter which is used as an address generator of frame memory in the system condensing image data by using blocktransform including DCT(DISCRETE TRANSFORM), VQ(VECTOR QUANTIZATION) etc., this consists of: 10 bit counter(1) which uses "ON" signal which activates as "High" during data exisitng period so that can be counted only during the period, while does as "Low" during data non-existing period, also uses macroblock start signal(MSC) which resets in the starting point of macroblock, and is used to count one macroblock: Y processor making horizontal & vertical address of luminance signal by using 10 bit counter(1) & slice number(SVP), macroblock number(MACRO #): C processor(3) making horizontal, vertical address of color difference signal(C) by using 10 bit counter and slice number(SVP), macroblock number(MACRO #): the 1st multiplexer(4) that is used for offering vertical address with selecting Y address during Y period and C address during C period: the 2nd multiplexer(5) that is used for offering horizontal address with selecting Y address during Y period and C address during C period.

As FIG. 6 and 7 are the schematic diagrams of Y processor & C processor(3) consisting of this invention, for Y processor(2) slice number(SVP) shall be used as max. 4 bit VY(9).about.VY(6) of Y vertical address while for C processor(3) as C vertical address(VC(8).about.VC(5)); in C processor(3) the max. bit(MSB) of 10 bit counter(l) shall be used as signal(YC) to distinguish Y and C period: in Y processor(2) macroblock number(MACRO #) shall be used as max. 6 bit(HY(7).about.HY(2)) of Y horizontal address, while in C processor(3) macroblock number (MACRO #) as min. 6 bit(HC(5).about.HC(O)) of C horizontal address. And in Y processor(2) in order to increase 4 slices in turn 2 bit(K, L) of 10 bit counter(l) shall be used as Y vertical address (VY(5), VY(4)) while in C processor(3) for the same purpose as the said 2 bit(K, L) of counter(1), used as C vertical address(VC(4)), VC(3)): in Y processor(2) in order to count 8 lines within block middle bit(G, H, I) of 10 bit counter(1) shall be used as min. 3 bit(VY(2), VY(1), VY(O)) of Y vertical address, while in C processor(3) for the same purpose as the aforementioned middle bit(G, H, I) of 10 bit counter(1) used as min. 3 bit(VC(2), VC(1), VC(O)) of C vertical address.

Furthermore, it is also composed of: in Y processor(2) so as to count horizontal 4 blocks middle bit(E, F) of 10 bit counter(1) shall be used as min. 2 bit(HY(1), HY(O)) of Y horizontal address: in order to distinguish between 4 top and 4 bottom blocks of illuminance signal Y, D of 10 bit counter (1) shall be used as Y vertical address(VY(3)); in C processor (3) so as to store color difference signal(U, V) data in frame memory in different field from illuminance signal (Y), max. 2 bit HC(7), HC(6) of C horizontal address shall be set as "High" and in order to store U & V datas in frame memory in different field from each other, F of 10 bit counter(l) be used as max. bit VC(9) of C vertical address.

This invention with the above composition can be explained as follows.

The video data for HDTV in this invention shall have a hierarchical structure so as to be advantageous to DCT(DISCRETE COSINE TRANSFORM) and motion compensation. i.e. the structure from small unit to block, macroblock, slice, frame. 1 block is composed of horizontal 8 pixel, 8.times.8 of vertical 8 line and 1 macroblock is of luminance signal 8 block like FIG. 1, color difference signal 2 block(U, V, each 1 block). Consisting of 44 macroblocks, 1 slice is of same frame as FIG. 2. Then Y(luminance signal), U(color difference signal), V(color difference signal) shall have 60 slices resp. and each slice have 44 macroblocks. As the memory memorizing such frame can be illustrated like FIG. 3, by using max. bit of horizontal & vertical address, the addresses of Y, U, V shall be distinguished as follows:

In this invention, in order to do data processing speed fast the circuit is composed of 4 step parallel processing method and the memory, considering motion compensation, is divided into even & odd memory. FIG. 4 shows the first block(block 1) from slice 1 to slice 4. The first 4 pixels of slice(1.about.4), i.e. a in slice 1, b in slice 2, c in slice 3, d in slice 4, are stored in even memory, while the second 4 pixels of slice(1.about.4), i.e. A in slice 1, B in slice 2, C in slice 3, D in slice 4 stored in odd memory. The storage order to memory is in order of a, b, c, d, A, B, C, D, e, f, g, h, E, F, G, H, and a, b, c, d, e, f, g, h, to even memory while A, B, C, D, E, F, G, H, to odd memory are stored resp.

In this way, store 1.about.8 block in FIG. 1 in turns and, as block 9 is U(color difference signal), fix max. 2 bit(HC(7), HC(6) in FIG. 7) of horizontal address as `11`, also fix max. 1 bit of vertical address as `0` and then store in the same way as FIG. 4. As block 10 is V(color difference signal), furthermore, fix max. 2 bit of horizontal address as `11`, also fix max. 1 bit of vertical address as `1` and then store in the same way as FIG. 4.

44 macroblocks shown in FIG. 2 shall be stored by 1 macroblock in memory shown in FIG. 3 in this way.

As frame data in this invention is stored in memory after 4 step parallel processing, which 4 slices shall be stored in case of 44 macroblock storage, next time store slice 5.about.slice 10 in this way.

Since the address implementation so as to store pixel data of frame to memory is carried out by raster format converter circuit in FIG. 5. ON signal shows the data existing period and, so that only necessary datas can be stored to memory as counter is not increased during data non-exisitng period.

That is, 10 bit counter(1), as counter counting 1 macroblock, shall be reset by macroblock start(MSC) signal. And slice number(SVP) & macroblock number(MACRO #) are supposed to be made in advance and the implementation is also possible by counter etc.

In the concrete, this can be explained together with Y processor(2) shown in FIG. 6 and C processor(3) in FIG. (7). Y processor(2) makes Y horizontal address(HY(7).about.HY(O)) and Y vertical address(VY(9).about.VY(O)) by using output of 10 bit counter(1) & slice number(SVP), macroblock number(MACRO #), and the processing progress is described in FIG. 8.

C processor(3) makes C horizontal address(HC(7).about.HC(O)) and C vertical address(VC(9).about.VC(O)) by using output of 10 bit counter(1) & slice number(SVP), macroblock number(MACRO #), and the processing progress is described in FIG. 9.

Like this, Y, C addresses made by Y processor(2) and C processor(3) select multiplexer (4) & (5) by using MSB marked C bit of 10 bit counter(I), which distinguishes Y and C, as YC section signal, so that the output of Y address can be made during Y period and the output of C address be done during C period. Thus, by using only one of 10 bit counter(1), the address which stores datas by block unit shall be made.

In this invention, the operation of Y processor(2) for Y processing is done as FIG. 8. Reset(A1) 10 bit counter(I), then by the bit(K, L) increment to Y processor(2) due to data input, also be increased(A2) the luminance vertical address (VY(4.about.5)), check whether 4 parallel data(4 slice) is inputted and, if not, carry out(A3) again the said step(A2). As, by repeating this, 4 parallel data is increased to bit(G, H, I) to Y processor(2), vertical address(VY(O.about.2)) is also increased(A4). Subsequently, count whether 8 line data is inputted, if not, carry out the said step(A2) and, if so, carry out the next step(A5).

If 8 line data is inputted, as EF bit of Y processor(2) is increased, horizontal address(HY(O.about.1)) is also risen(A6).

Check whether 4 block data is inputted with such increment, if not, repeat to carry out from the said step(A2) and, if so, carry out the next step(A7). If 4 block data is inputted, D bit to Y processor(2) is increased, accordingly so is(A8) vertical address(VY(3)). Check whether 8 block datas are all inputted with such increment, if not, repeat to carry out from the said step(A2) and, if so, carry out the next step(A9).

Then, if 8 block datas are all inputted, by using macroblock number(MACRO #) shown in FIG. 5, increase(A10) vertical address(HY(2.about.7)). Thus, if 44 blocks are not all inputted, repeat to carry out from the said step(A1) and, if all inputted, carry out(All) the next step.

Subsequently, by using slice number(SVP) shown in FIG. 5, increase vertical address(VY(6.about.9)) and carry out(A12) Y processing.

On the other hand, in C processing of this invention, if resetting(B1) 10 bit counter(l) like FIG. 9, as data is inputted, bits(K, L) to C processor(3) are increased and so is(B2) vertical address(VC(3.about.4)). Thus, check whether 4 parallel datas are inputted, if not, carry out the said step(B2) again and, if so, carry out the next step(B3). If 4 parallel datas are inputted, as bits(G. H, I) to C processor(3) are increased, so is(B4) the vertical address (VC(0.about.2)). Thus, check whether 8 line datas are inputted, if not, repeat to carry out from the said step(B2) and, if so, carry out the next step(B5). If 8 line datas are inputted, as F bit to C processor(3) shown in FIG. 5 is increased, so is(B6) the vertical address(VC(9)).

Judge whether 4 block datas are inputted with such increment of vertical address, if not, repeat to carry out the said step(B2) and, if so, carry out the next step(B7).

In the aforementioned, if 4 block datas are inputted, C processor(3) increases(B8) horizontal address(HC(0.about.5)) by using macroblock number(MACRO #). Judge whether 44 macroblock datas are all inputted with the increment of horizontal address, if not, repeat to carry out again from the said step(B1) and, if so, carry out the next step(B9). If 44 macroblock datas are all inputted, increase vertical address (VC(5.about.8)) by using signal(SVP) shown in FIG. 5 and carry out(B10) C processing with repeating to carry out this from the said initial step(B1).

Like the above, as composing address generation circuit, which leads or lights the data consisted of block unit in HDTV with digital type to frame memory, by using only one 10 bit counter, this invention can get the saving effect of the cost for circuit implementation.