NZ229271A - Image compression pixel transcoding - Google Patents

Description

229 2 7 1 2V 5;.S?>; Priority Oate(s): Complete Speci'oation aas,: .VA-2 Publication bsis: ...... 2.8 MAY..1991...

P.O. Journal, M, RUE COPY NEW ZEALAND PATENTS ACT 1953 COMPLETE SPECIFICATION "IMAGE INFORMATION CODING" WE, STANDARD TELEPHONES AND CABLES PTY. LIMITED, A Company of the State of New South Wales, of 252-280 Botany Road, Alexandria, New South Wales, 2015* Australia, hereby declare the invention for which we pray that a patent may be granted to us, and the method by which it is to be performed, to be particularly described in and by the following statement: 1 22 9 2 71 The Invention is concerned with a transcoding method for bitmap image digital data, with special reference to the Information system providing on request images of documents which have previously been stored in memory in a digital 5 form.

The rapid growth of the possibilities offered by personal computers has made it possible for certain techniques previously only accessible to specialists to be popularised, particularly in the publishing industry. Laser and other 10 similar printers now make it possible to reproduce Images which had previously been stored in a memory by a computer in a digital form, with the aid of a scanner.

Thus, the users of such systems themselves can now produce documents which bring together image reproductions 15 and previously stored texts, with the assistance of routinely available specialised software. However, the memory storage of image data from a scanner Implies the availability of large computer memory capacities. For example, these can be in the order of one megaoctet per A4 at standard page 20 as read by a scanner, with a resolution of 300 ppi.

The idea, therefore, is to transcode the bitmap image data provided in binary form by the scanners in order to reduce the volume of data to be stored in memory, whilst mak- 2 22 9 2 7 ing It possible to fully reconstruct the images after storage in the memory.

Certain codes, such as the HUFFMAN code, are already well known: They provide a substantial compression of the 5 volume of binary data which is required for the storage of images from a "plain" coding made by a scanner, and considering each image in bitmap form, that is as a set of elementary points.

The data which has been transcoded in order to be stored must be re-transcoded afterwards in the reverse direction when they are being used for reproduction. This operation often requires very long decoding times with the above mentioned codes when specific auxiliary equipment Is not available.

Thus, the invention offers a method for the compressed transcoding of bitmap Image digital data, making it possible, In virtue of its structure, to quickly retrieve the initial plain code from a "memory storage containing the compact data of the stored, images.

The method offered by the invention aims at providing a compressive transcoding of the bitmap image digital data which is generated in the shape of a succession of bits, and which translate into a first code, often called the initial 2202 code or the plain code, the respective states of the successive pixels of at least one image supposed to be' read through linear scanning.

According to one characteristic of the invention, the bits which in the first code correspond to the successive Image pixels are transcoded Into a compressed code octet in succession.

Accordingly there Is described a method of transcoding "bitmap" image data from a first code in the form of a succession of bits representative of the states of successive pixels, wherein the bits are transcoded in bytes of known length in succession into a compressed code, the method including classifying each byte or group of two or more bytes representing contiguous pixels into one of a two or more classes depending on selected characteristics of the byte or group of bytes, transcoding each byte or group into a compressed code in accordance with a transcoding alghorithm depending on the class in which the byte or group of bytes is classified, and Incorporating a segment Identifier with the transcoded byte or group of bytes to indicate the class of the byte or group of bytes. 2292 7 The bits which, in the first code, correspond to the successive image pixels are transcoded octet by octet and in succession, Into modules of a compressed code. Each module includes a segment identifier which is possibly followed by 5 unchanged octets from the first code. Each segment identifier is composed of one or two octets which determine the nature of the coded segment and the length of the segment via two distinct groups of bits. The segment of the code, which is composed of Identical successive octets themselves 10 made up of identical bits and corresponding to monochromatic segments of the image, are transcoded in the shape of modules of a first type composed of a segment identifier.

The first code segments composed of at least one octet comprising bits that correspond to at least two different 15 pixels in a succession of non-repetitive octets, are transocded in the form of modules of a second type composed of a segment identifier .followed by the unchanged octet or octets of the first code".

The first code segments, the constituent octets of 20 which are identical and in the same line position as the corresponding octets of a previous scanning line, are transcoded in the shape of modules of a third type, which are composed of segment identifiers. 229 2 The first code segments, the successive octets of which are of identical heterogeneous composition, are transcoded in the shape of modules of a fourth code, which are made up of a segment identifier followed by the first code octet 5 which supports the repetition.

According to a first alternative, the successive Identical octets of the bits corresponding to successive identical pixels are transcoded In the shape of modules of a compressed code of a first type. Each module of the first 10 type, generally based on at least two successive identical octets, is composed of a segment identifier Including at least a first octet where at least two bit positions, each having a specific role, respectively translate the identity of the pixels concerned and that state, given that at least 15 some of the other bit positions of that module specify, according to their value and their rank, on the one hand the number of octets of the segment identifier of the module, and on the other hand the number of successive first code octets which are concerned, by this module. The first code 20 octets which individually include bits corresponding to at least two different pixels, are transcoded in the form of modules of the compressed code of the second type. Each second type module is composed of a segment identifier of at 2 2 9 2 7 least one octet and one additional octet, the first octet including two bit positions which respectively translate the existence of different pixels amongst the pixels concerned, and the continuation of the code in respect of at least the 5 said following octet. At least part of the other bit positions of the segment identifier of this second type module specify, according to the values and the rankings of the bits, on the one hand the number of octets of the segment Identifier of the module, and on the other hand the number 10 of successive octets In the first code that the module includes .

The invention, its characteristics and its advantages are described as follows, together with the figures listed be-low: Figure 1 represents the basic diagram of a classic mi cro publishing station.

Figures 2A and 2B represent two structures of segment identifier for composed code module.

Figure 3 presents a synoptic table of the structures of 20 the modules in respect of an example of compressed code.

The micro publishing station presented on Figure 1 is dedicated to the production of printed documents. It is aided by so-called computer assisted publishing techniques 22 9 2 7 which imply the memory storage, in numerical form, of texts and images to be printed.

To that end, such a micro publishing station Is organised around at least a computer 1, a PC for example, 5 classically equipped with active or passive memories (not shown), a visual display unit 2, and an alpiianumerical keyboard 3> possibly linked to a mouse, which is not shown. The texts and Images to be printed are stored in advance in the computer memory 1 in numerical form, in order to be 10 organised by the user with the aid of appropriate software, and in order to make it possible to have them printed via a printer 4, of the laser type for example, which is connected to the computer 1 for that purpose.

The texts are classically entered in numerical form 15 into the computer memory, thanks to the ancillary equipment of the computer. Such equipment Includes, for example, the keyboard 3, a disc/diskette drive (not shown), on which text and image data have previously been recorded, a modem (not shown), which makes it possible to receive the numerical 20 data transmitted by a telecommunication link 5> which connects, for example, with a service centre or a graphic scanner 6. Such a scanner 6 is responsible for the conversion of the image carried by a face of a document into a sue- 229 2 cession of bits which represent image pixels, the image itself being read by linear scanning. The bits representing the successive pixels on the successive image lines, are classically regrouped in a succession of octets so that they 5 can be accounted for. One to eight bits, according to the primary code selected, are therefore provided by the scanner for each pixel. A single bit is enough to ensure the plain coding of the state of a pixel of one monochromatic image at two levels. Several bits dedicated to the same pixel make 10 it possible to obtain variations of the colour shades of that same image, weaving effects, or else different shades of grey in black and white pictures. As well known, the succession of octets obtained for one image is memorised in a file permitting access to binary image data via 15 specialised software for the preparation of documents to be printed. This kind of software works from texts and/or images which have been stored on file.

As explained above, one of the objectives of the method employing the invention is. to transcode the binary signals 20 the first code, which, for example, are provided by a scanner in the case of an image being treated according to the bitmap method, into a compressed code which makes it possible to consequently reduce the volume of memory re- 9 22 9 2 quired for their conservation, whilst authorising a quick retrieval of the Initial code without any loss of information. This is made possible by the brevity of the memory access time necessary for their retrieval during the 5 operation.

To that end, a processor takes charge of each octet of the first code. This processor is generally located either at the level of the scanner 6, or at the level of the computer 1. Each first code octet is then considered for 10 transcoding, either In isolation, or, preferably, in combination with the neighbouring successive octets, so as to form a module of the compressed code, which, too, is made up of octets. Such a choice of format facilitates the treatment of the bitmap image binary data by the processor(s), as 15 it does not imply any translation inside the registers of a processor, nor any index handling, thereby avoiding any corresponding losses of time.

The successive octets of the first code, of one or several successive images are transcoded into modules of the 20 compressed code. Such modules comprise a whole number of octets generally much smaller than the number of octets of the first code.

Each bitmap image data file created in order to memorise in the compressed code numerical data which is provided in the first code, includes a definition heading followed by the Image data.

This file heading specifies in a familiar way some characteristic image and coding data. In the present case, this includes inter alia the number of image lines, the number of octets per image line of the first code, the number of valid pixels per line, an indication, either implicitly or explicitly, concerning the code used, a possible indication of A DITHER matrix for keyboarding, and, in this case, the dimensions of that matrix, and, in particular, the number of lines concerned with the image.

Each compressed code module also includes a segment identifier composed of one or two octets, which are possibly followed by a number of octets which is then defined in the module identifier.

Pour types of modules of the compressed code, respectively labelled Ml, M2, M3. and M4 may be used, in the transcoding method.

The first compressed code module, Ml, is dedicated to the transcoding of successive, identical octets of first code bits which correspond to successive, identical pixels, 22§ t that is to say those which belong to one or several lines of a uniform image segment.

The second module of the compressed code, M2, is dedicated to the transcoding of first code octets which corre-5 spond either to at least two different pixels each, or, independently, to an octet of identical pixels inserted between octets of different pixels.

The third module of the compressed code is dedicated to the transcoding of first code octets of a scanning line, 10 which are identical to, and in the same line position as the corresponding octets of a preceding scanning line.

The fourth module of the compressed code, M4, is used for transcoding successive first code octets which are of an identical heterogeneous composition, and which correspond to 15 a repetition of a brief pattern, such as particularly In the case of colour coding, or in the case of the coding of shades of grey which require the use of several bits per pixel, or, also, in the case of weaving effects. Each pixel is then represented by several bits, classically between two 20 and eight per colour plan. Each pixel has definition bits in the different plans.

The first octet of the identifier of each of the modules of the compressed code mentioned above includes, ac- 12 229 2 7 cording to the case concerned, three or four bit positions of specific rank, the contents of which determine the type of module and the number of octets of its identifier.

In the example of an octet of the module identifier 5 shown on Figure 2A, the bit position in rank 7 makes it possible, according to its contents, that is to say its binary state, to determine whether the Identifier which includes it extends over one or two octets.

The bit position of rank 6 makes it possible, according 10 to its contents, to determine whether the module which includes it is of the first type (Ml) or not, that is whether the module corresponds to a range of identical pixels or not.

The contents of bit positions of rank 5 and 6, which 15 are Independent, determine either the state of the identical pixels of the module of the Ml type which includes them in a first combination, or the fact that the two positions of rank 5 and 6 pertain to an M2 module via a second combination, or else to a module of the M3 or M4 types through a 20 third combination.

The contents of the bit position of rank 4 makes it possible to differentiate between the M3 and M4 modules when the contents of the bit positions of rank 5 and rank 6 cor- 13 22 9 2 7 respond to the third combination mentioned above. Bit positions of ranks 3> 2, 1, and 0, that is to say here on the right, of an octet of a module (Ml) or second (M2) module, enable the definition of the number of first code octets 5 which are transcoded by this module of the compressed code.

In the case of an identifier of a module of the compressed code made up of two octets (Figure 2B), the second octet is utilised in order to increase the transcoding capacity in terras of the number of first code octets of the 10 module which includes it. The eight bit positions of the second identifier octet will then add to the positions that are reserved for that purpose in the first octet of the Identifier in question.

Figure 3 presents an example of realisation of the in-15 termediary code adapted to transcoding according to the invention, and makes it possible to determine the way in which such a code can be operated by a processor, such as that of computer 1 at Figure 1.

For example, the graphic scanner 6 generates a suc-20 cession of octets of bitmap image which is plainly coded as the analysis of an image progresses when translated, for example, in monochrome on two levels, that is with one bit per pixel. An Image file heading is typically created by the 14 229 2 7 user before the file receives a series of modules of the compressed code generated by the treatment of the successive octets of the straight code by the processor as described above.

If, for example, the succession of bitmap data received by the scanner corresponds to a series of a number of homogeneous and identical octets, that is to say that they are composed of bits of the same value, which correspond to a uniform range of pixels, the processor described above will 10 produce a segment Identifier for the module Ml. The identifier appears after the identical octets have been counted, either up to the first heterogeneous octet, that is the one which includes two different bits, or up to the specific limit number n of octets.

This number n of octets is for example that which is likely to be coded over 13 bits and which therefore corresponds to more than eight thousand octets of the straight code.

The Ml module is composed of only one octet if the num-20 ber of homogeneous and identical octets is likely to be counted in five bits. This lone octet is its identifier octet. 22 9 2 7 In the example which is being described, the identifier octet signals that it constitutes a module Ml by itself, due to its rank 7 bit at the zero stage, that it corresponds to a group of homogeneous and identical octets of the straight 5 code, due to its rank 6 bit at state 1, and which is the binary state (i.e. 0 or 1), common to the pixels, thanks to its rank 5 bit.

The five other bits of this unique octet determine the number of octets concerned with the straight code, that is 10 the area of the uniform segment concerned with the monochromatic image referred to above.

Those skilled in the art are easily able to reconstitute the initial code from such an octet of the Ml module with the aid of a processor duly programmed in a classical 15 way, which will not be described here, as it has no immediate connection with the object of this invention.

A segment identifier of two octets is constituted if the number n of homogeneous, identical and successive octets requires more than five bits binary coding. The first one 20 of those two octets is differentiated from the first octet referred to above by the state of its rank 7 bit, i.e. "1", which specifies that the Ml module is composed of two octets, and that 13 of these bits of the module of the first 16 2 2 Q 2 7 type are utilised for the translation of the number of octets of the straight code which are thus transcoded.

The addressee would also have no difficulty to retranslate into the straight code such a module of the compressed 5 code with the aid of a processor.

The appearance of an heterogeneous octet in the straight code after a series of identical homogeneous octets, or the appearance of a homogeneous octet whose bits have an inverse value of those of the bits of the previous 10 octet, or also the counting of the limit number n of homogeneous, successive and identical octets generate the production of an Ml module of the first type by the processor responsible for transcoding.

In the case where the segment of uniform pixels could 15 not be coded by one Ml module only, a second Ml module should be produced in the same conditions. The same situation occurs when identical, homogeneous octets of the straight code appear in succession, with bits of opposite value to those of the previous octets.

A module of a second M2 type is produced in the event of the appearance of an heterogeneous octet of the straight code whose bits correspond to at least two pixels of different states. 17 22^27 In the example suggested in relation to Figure 3> an M2 module includes an identifier composed of one or two octets. The first heading octet determines by its bit of rank 7 at state 0 or 1, whether the module which includes it does have 5 one or two identifier octets. The combination of contents 0 or 1 of the bit positions of rank 6 and 5 indicates that there is a change of state in an octet in the straight code, and that at least that straight code octet is reproduced identically in intermediary code, following the identifier 10 which is being considered.

There also is no difficulty for the addressee to return to the straight code from an M2 module.

In an alternative realisation also shown on Figure 3, it is possible to code specifically the series of octets 15 which correspond to identical conflgurations on successive image scanning lines. This implies the temporary storage of data for at least two successive image lines.

This solution is also implemented when the image has been captured with a DITHER matrix, and when the same con-20 figuration of the image lines repeats itself with a module corresponding to the number of lines of the matrix concerned . 18 22 9 2 7 The Identification of the configurations of octets of the first code of an image line which are identical and in the same line positions as their counterparts of a previous image line, leads to the production of a third type of M3 5 module of the compressed code.

The identification of the lines is easily achieved for example by counting a specific number of octets of the first code, or by a corresponding determination from the indications of the number of octets contained in the modules of 10 the compressed code.

Such an identification can also be achieved by the insertion of a specific code, a code of end of line, for example, composed of two octets whose bits have a zero value, when transcoding in the compressed code.

In the example proposed in relation to Figure 3, an Identity between the successive lines or portions of lines is transcoded via an M3 module. Such an M3 module includes a segment Identifier on one or two successive octets.

As was previously the. case, the first octet of the seg-20 ment identifier specifies, due to its bit 7, the number of octets of the identifier which contains it, and, by way of a specific combination of bits of ranks 6, 5 and 4, here all being at zero, that the coding is bi-dimenslonal and re- 19 229 2 7 produces, at least partially, that of a preceding image line.

The other four bit positions of an octet of an identifier, or the other twelve bit positions of two successive 5 octets of an identifier specify the number of successive octets of the first code of an image line that are concerned .

In another alternative realisation also presented on Figure 3> it is necessary to code specifically the succes-10 sive octets of the first code which have identical heterogeneous compositions.

This applies particularly when a number "p" of bits is being used for coding a pixel, "p" being greater than 1 when shades are provided at the level of the images. 15 The scanner then provides a first code in which each pixel is coded by a group of "p" bits, a same configuration of the "p" bits being used for the identical pixels.

Successive and Identical pixels are then translated by a succession of heterogeneous and identical octets in the 20 first code, and the identification of such a succession leads to the generation of a fourth type of M4 module in the compressed code. 22 9 2 7 Such an M4 module Includes a segment Identifier composed of one or two successive octets.

As in the case of the previous ones, the first octet of the identifier specifies the number of octets of the ldenti-5 fler which contains it, via its rank 7 bit; it also defines, by a specific combination of its rank 6, 5 and 4 bits, here respectively at 0, 0 and 1, that there is a repetition of heterogeneous octets.

The other four bit positions of the octet of the lden-10 tifier or the other thirteen bit positions of the successive octets of the identifier specify the number of successive octets of the first code which are concerned by this repetition. The heterogeneous octet of the first code which is to be repeated is then reproduced identically, following the 15 identifier of the M4 module being considered.

Once again, the reconstitution of the first code from such a module of the compressed code with the aid of a processor is easy, thanks to simple programs whose disclosure is without immediate connection with the invention. 20 The different modules Ml, M2, M3 and M4, likely to co exist in the same image data file, or even in a same image line, make it possible to recall the number and the consti- 21 229 2 7 tution of the octets of the first code of an image, whilst it is possible to memorise this image within a small volume.

The compressed code according to the invention offers the advantage of not being expansive, as it implies a raaxi-5 mum of one octet or a tw-octet word additionally per line in the most unfavourable case where all the first code octets of a line were to remain in the first code after being transcoded into the compressed code. 22

Claims (17)

2292 What we claim is:

1. A method of transcoding "bitmap" image data from a first code in the form of a succession of bits representative of the states of successive pixels, wherein the bits are transcoded in bytes of known length in succession into a compressed code, the method Including classifying each byte or group of two or more bytes representing contiguous pixels into one of a two or more classes depending on selected characteristics of the byte or group of bytes, transcoding each byte or group into a compressed code in accordance with a transcoding alghorlthm depending on the class in which the byte or group of bytes is classified, and incorporating a segment identifier with the transcoded byte or group of bytes to indicate the class of the byte or group of bytes.

2. A method of transcoding as claimed in claim 1 wherein a first class is characterized by two or more successive bytes being identical.

3. A method as claimed in claim 2 wherein the segment identifier is of a first type and includes a plurality of bits indicating the position of the pixels corresponding to the bytes before transcoding the state of the pixels, the number of bytes in the group before transcoding, and the length of the transcoded group. 229271

4. A method as claimed in claim 2 or claim 3 wherein the segment identifier includes a first bit which indicates whether the segment identifier relates to one or more than one transcoded byte; a second bit which indicates that the segment identifier relates to a group of identical successive pixels; a third bit which indicates the binary state which the segment Identifier represents; and a plurality of further bits which indicate the number of successive pixels to which the segment identifier relates.

5. A method as claimed in claim 4, wherein the segment Identifier relates to more than one transcoded byte.

6. A method as claimed in any one of claims 1 to 5, wherein a second class is characterized by at least two differing successive bytes.

7. A method as claimed in claim 6 wherein the bytes of the second class remain unchanged in the first code after transcoding and are incorporated with a segment identifier of a second type and at least one additional transcoded byte.

8. A method as claimed in any one of claims 1 to T, wherein a third class is characterized by bytes representing identical pixels in successive rows and in the same line position . 229271

9. A method as claimed in claim 8, wherein the bytes of the third class are transcoded as segment identifiers of a third type.

10. A method as claimed in any one of claims 1 to 9, wherein a fourth class is characterized by the repetition of identical sequences of differing bytes.

11. A method as claimed in any one of claims 1 to 10, wherein the step of classifying each byte or group of bytes is performed by a processor.

12. A method as claimed in claim 11, wherein the processor performs the step of classifying by comparing bytes with contiguous bytes.

13. A method as claimed in claim 12, wherein the processor stores at least one byte for comparison with the Immediately succeeding byte.

14. A method as claimed in claim 12 or claim 13, wherein the processor stores bytes representing at least one line of pixels and an immediately succeeding line of pixels whereby the processor compares bytes representing pixels of the immediately succeeding line with the corresponding pixels occupying the same line position in the stored line.

15. A method as claimed in any one of claims 1 to 14 ,wherein the segment identifier includes data indicating the 229271 *;number* of successive pixels to which the transcoded byte is applicable.

16. A method of transcoding bitmap image data generated by linear scanning of an image substantially as herein described with reference to the accompanying drawing.

17. A data transcoder employing the method of any one of claims 1 to 16. STANDARD TELEPHONES AND CABLES PTY. LIMITED P.M. Conrick Authorized Agent P5/1/1703