GB2154826A

GB2154826A - A method and apparatus for condensing image signals

Info

Publication number: GB2154826A
Application number: GB08504425A
Authority: GB
Inventors: Takashi Sakamoto
Original assignee: Dainippon Screen Manufacturing Co Ltd
Current assignee: Dainippon Screen Manufacturing Co Ltd
Priority date: 1984-02-22
Filing date: 1985-02-21
Publication date: 1985-09-11
Also published as: DE3505796C2; GB2154826B; DE3505796A1; JPS60176365A; GB8504425D0; JPH0423869B2; FR2559979B1; FR2559979A1

Abstract

Image signals defining colour in the elements of a scanned picture are transmitted, with economy in information, by deriving all but one of the colour signals at all but one of the nodes of each local (e.g. 2 x 2) matrix from all the colour separation signals (e.g. cyan, magenta, yellow, black C00, M00, Y00, K00) at the remaining element of the matrix and the change in level of the remaining master colour (e.g magenta) on moving to the other elements in the matrix. The relationship between the steps in level of the master colour and the derived steps (m') in the other in non-linearly quantized (Fig. 3), providing closer step spacing at low gradients of brightness. <IMAGE>

Description

SPECIFICATION A method and apparatus for condensing image signals THIS INVENTION relates to a method and apparatus for condensing image signals, in particular a method and apparatus for condensing ditigal image signals obtained by scanning a colour original image using an apparatus, such as a colour scanner, for scanning and recording images.

Generally, where a colour image having variable colour density is to be reproduced, it is necessary for variations in tone to be represented in the reproduction image. Even in case of a digital image, it is said that a storage capacity is required to enable the storing of not less than two hundred different tones for each of the primary colours, that is red (R), green (G) and blue (B) or for each of the complementary inks cyan (C), magenta (M), yellow (Y) and black (K). In order to enable up to two hundred different tones to be represented, each colour separation signal of each picture element of the original image needs to be represented by a 8-bit digital signals.

The number of bits of memory required to store the signals representing each picture element is therefore so great that, when signals representing each colour separation signal of each picture element within a certain scanning region of the original image are to be stored, a storage memory of considerable capacity is required.

In order to minimize the total storage capacity required, the pitch at which the original image is sampled during scanning is increased or coarsened to the maximum possible limit which is allowable if the reproduction image is not to seem unatural to the human eye. Accordingly, the correlation between adjacent picture elements is not so close or high, and even by condensing the image signal for each colour separation signal, it is usually difficult to condense the image signals and at the same time avoid producing a reproduction image which is of poor visual quality.

In view of the foregoing problem, the Applicants have already proposed a method for recording colour images in which the storage capacity of the memory required to store the signals representing the images is reduced to a certain extent without adversely affecting the visual quality of the reproduced image. Thus as is disclosed in Japanese laid open (unexamined) Publication No. Sho 55-22708 (Japanese Patent Application No. 53-94507, advantage is taken of the fact that the human eye is much more sensitive to changes in the brightness of small areas than to changes in colour of such a minute area.In the previously proposed method for condensing image signals and for reproduction of the image represented by the condensed images signals for certain of the picture elements, signals representing only the brightness of one (or possibly more) of the colour separation signals are retained in memory.

Referring now to Fig. 1 of the accompanying drawings, there is illustrated one mode of condensing each of the four colour separation signals for the respective colour plates C, M, Y, K disclosed in the Japanese laid open Patent Publication No. Sho 55-22708.

Thus as shown in Fig. 1, the original image is divided into the desired number of picture elements and the colour separation signals for each of the colour separation plates C, M, Y, K are recorded for only one representative picture element in each 2 X 2 matrix of picture elements, the magenta colour separation signal only is recorded for the other picture element and is used to determine the brightness thereof, magenta being closest to the human visual sense.

The digital signal representing the magenta colour separation signal for the colour plate M comprises 8 bits whether it is a representative picture element or one of the other picture elements.

In order to reproduce the signal for the colour plates C, Y, K for the picture elements other than the representative picture elements, the following formulae are used, in which the subscripts 00, 01, 10 and 11 represent, for each 2 x 2 matrix of picture elements, the positional relationship of the other picture elements to the representative picture element.

For example, for a given representative picture element, the colour separation signals for the picture element 01 of that 2 x 2 matrix are obtained in the following manner: C01 = COO + M,, - Moo M01 = M01 = Moo + M01 - M00 Yol = YOO + M01 - M00 K01 = K,, + M,, - Moo This method only works as far as variations in brightness in each unit region of 2 x 2 picture element matrix are concerned.However, as the factor "M01 - M00,, common to the above four formulae gives the difference in brightness between the two picture elements, the method has an advantage of being able to work for each minimum unit of the picture element.

Although the above formulae illustrate a method in which the colour separation signals COO, YOO, K00 of the represntative picture elements-are used to calculate the colour separation signals C01, Yo1 Ko1, it is also possible to use an interpolation factor obtained from the respective colour separation signals for the four representative picture elements surrounding the picture element in question, as is described in afore-mentioned Japanese laid open Patent Publication Sho.

55-22708. This method will not, however, be described herein in detail.

In the above method, colour information is stored only for the representative picture element of each unit region or 2 x 2 picture element matrix and the information for the other picture elements in each region is not stored. The storage capacity required is thus reduced because, if the 8-bit image signals C, M, Y, K for each picture element are stored, a storage capacity of 1 6 bytes is normally required for each 2 X 2 matrix of picture elements when four colour separation signals are required whereas, if colour separation signals C, Y, K are stored only for the representative picture element, the storage capacity required per 2 x 2 matrix of picture elements is reduced to: 2 x 2 picture elements X brightness value + 3 colours = 7 bytes.

However, even where the above-described method, is adopted, although the image signals required to be stored to produce the colour separation plates C, Y, K are reduced by a quarter in the above example and could be reduced by a ninth for a 3 X 3 matrix of picture elements, the degree to which the image signals are condensed to reduce the required storage capacity is not satisfactory when all four colour separation signals are considered because the magenta colour separation signal representing brightness is stored for each picture element.

It is an object of the present invention to provide a method and apparatus for condensing image signals which overcomes or at last mitigates the above-mentioned problems.

According to a first aspect of the present invention, there is provided a method of condensing image signals comprising colour separation signals for each of a plurality of picture elements of an original image, which method comprises using the colour separation signals of a representative picture element to represent the colour image signals of a region of picture elements, selecting a respective single colour separation signal of each of the other picture elements of the region to represent the brightness of that picture element and encoding the selected colour separation signals to produce representative brightness signals.

In a second aspect, the present invention provides a method for condensing image signals in which a unit condensing region comprising a plurality of picture elements is established, all of a plurality of image signals corresponding to colour signals necessary for reproducing a colour image are provided by a representative picture element of the unit condensing region, and only an image signal corresponding to a brightness signal of the plurality of image signals is provided by each picture element other than the representative picture element of the unit condensing region so that the total number of bits required for storing the image signals obtained by scanning a colour original image may be reduced, in which method at least the image signals representing the brightness of each picture element other than the representative picture element in each unit region are encoded to produce a representative brightness signal for the picture element.

The present invention also provides apparatus for condensing image signals comprising colour separation signals representing each of a plurality of picture elements of an original image,which apparatus comprises means for using the colour separation signals of a representative picture element to represent the colour image signals of a region of picture elements, and means for encoding a selected single colour separation signal representing the brightness of each of the other picture elements of the region to produce representative brightness signals.

For a better understanding of the present invention, and to show how the same may be put into effect, reference will now be made, by way of example, to the accompanying drawings, in which: Figure 1 illustrates a previously proposed method for condensing image signals; Figure 2 illustrates a method in accordance with the invention for condensing image signals; Figure 3 is a graph illustrating the relationship between representative values and predicted error signals providing a transfer characteristic for the application of non-linear quantization; Figure 4 shows graphically an example of the tone characteristics reproduced where an image signal is reproduced using the transfer characteristic of Fig. 3; Figure 5 illustrates an alternative method in accordance with the present invention for condensing image signals; ; Figure 6 is a block diagram showing colour scanning and recording apparatus using a method in accordance with the invention; Figure 7 illustrates apparatus for carrying out the method illustrated in Fig. 2; Figure 8 illustrates apparatus for carrying out the alternative method illustrated in Fig. 5; Figure 9 illustrates apparatus for reproducing an image from image signals condensed by the apparatus of Fig. 7; and Figure 10 illustrates apparatus for reproducing an image from image signals condensed using the apparatus of Fig. 8.

Referring now to the drawings, Fig. 2 illustrates a method in accordance with the present invention for condensing image signals, including the brightness value utilizing the redundancy between brightness values in the picture elements of an original image making up a unit region (in this case a 2 X 2 matrix of picture elements) for which the signals are to be condensed.

As shown in Fig. 2, the colour separation signals C, M, Y, K for one representative picture element of each unit region, or 2 X 2 matrix, of picture elements are recorded. The subscripts 00, 01 10 and 11 in Fig. 2 give the positional relationship to the respective representative picture element of the other picture elements in the same matrix. The difference in brightness between each picture element of a unit region and the respective representative picture element is obtained and the brightness signal for the picture element is condensed by applying a nonlinear quantization characteristic to the difference signal.More specifically, in the first place, difference signals are obtained for each picture element, other than the representative picture element, as follows: mO1 = M01 - M00, m10 = M10 - M00 and m,1 = M11 - M00, where Moo M M,o and M" are the magenta colour separation signals for the representative picture element and the other picture elements of a unit region and m01, m10 and m11 are signals representing the difference between the magenta signal Moo for the representative picture element and each of the other picture elements of the unit region. mO1, m10 and m represent predicted or predictive errors.

When a redundancy exists in the colour signals Moo, Mo1, M10, M", the signals m01, m10, m become small values in most cases, and accordingly the condensing step is carried out by applying a non-linear quantization to those predictive errors mO1, m10, m11 to form representative values signals rio1, tri10, run11. It is quite usual to employ an encoding of variable length in such a case, but as it is considered that the separation from any optional position in relation to the original image is an essential factor in this embodiment, an encoding of fixed length is described hereinafter as an example.

Thus, Fig. 3 shows a transfer characteristic for non-linear quantization in which, when the predicted error signals mO1, m10, m,l approximate to zero, the step height is small, and when the predicted error signals are larger, the step height increases and the representative values rn',, rnio, m11 are prepared with larger step differences. At this stage, the total number of steps is far less than the tone value in total, which means that the number of bits is reduced to that extent.

The step of reproduction when the non-linear quantization is used is described hereinafter with reference to Fig. 4 which illustrates the variation in tone between picture elements.

Where as shown in portion A of Fig. 4 the variation in tone between picture elements is small and the predicted error is therefore small, the prepared representative values can be reproduced in a fine and exact manner. However, where as shown in portion B of Fig. 4, the variation between picture elements is large, and the predicted error is therefore large, the deviation from the most approximate representative value may be large, and accordingly the reproduced image may be a little different from the original image. Further, if the predictive error is too large, as it may happen that the representative value is not prepared, the deviation from the representative value becomes larger.With the non-linear quantization characteristic shown in Fig. 3, as the representative values rr1o1, rhio, are established to 5 bits (32 levels) and the value ihii to 6 bits (64 levels), when the absolute values of the predictive errors with respect to the values rnO1, tri10 are extremely large those representative values become close at a certain value. However, as the positions 01, 10 of the picture elements of a unit region are nearer to the position 00 of the representative picture element of the region than thepicture element at the position 11 in the unit region, it rarely happens that the representative values become close.

When the colour signals Mo1, M10, M11 representing brightness (that is the magenta colour signals other than the magenta colour signal Moo for the representative picture element) are not condensed, each colour separation signal requires 8 bits of memory so that the signals Mo1, M10 and M11 require 3 bytes of memory in total. However, when the total number of bits is condensed as described above, the signals rhol and rnrO occupy 5 bits and the signal rn" occupies 6 bits of memory so that 2 bytes of storage capacity are quite sufficient.

Referring now to Fig. 5, an alternative method in accordance with the invention is shown. In this method, the Hadamard conversion is used to condense the signals representing brightness.

Thus, if M is the signal representing brightness before conversion, m is the signal representing brightness after conversion and the subscripts 01, 10 and 11 indicate the positional relation of each picture element in a unit region (as described above) to the representative picture element and the subscript 00 indicates the position of the representative picture element, it is known that the two-dimensional Hadamard conversion is reduced to the following one-dimensional Hadamard conversion::

It is known that the modified or converted brightness signals mOO, m01, maO and m11 are closely correlated to the original signals Moo, Mo1, M10 and M11, and the total number of bits necessary for storing the signals mOO, mO1, m10 and m11 is related to the total number of bits necessary for representing the signals or values Moo, Mo1, M10 and M11.The extent to which the number of bits required is reduced depends on the intensity of the correlation between the signals Moo, Mo1, Mao and M,1, but in the case of a colour image read by the method of sampling pitch employed in the colour scanner used in graphics art printing, the correlation coefficient is not always high, and therefore, in order to obtain a quality sufficient for practical use, at least 8 bits are required for the signal mOO, 6 bits for the signal m11, and 5 bits each for the signals mO1 and m O.

The original signals Moo, Mo1, M10, M11 are easily reproduced from the signals mOO, mO1, m10, and m11, by making use of the orthogonality of the Walsh coefficient used in the Hadamard conversion as follows:

Although the above embodiment is described for the case where four colour separation signals are used for four colour plates C, M, Y, K the method is not limited to that case, but is also applicable to the case where three colour separation signals red(R), green(G), and blue(B), are used; in which case the colour separation signal G is used as the signal representing brightness.

In addition, although the above arrangement is described where the unit region or matrix of picture elements for condensing is a 2 x 2 matrix of picture elements it is also possible to use as a unit area a 3 X 3, 4 X 4 and so on, matrix of picture elements.

Fig. 6 is a block diagram illustrating colour scanning and recording apparatus using a method in accordance with the invention. The apparatus may be in the form of a layout scanner system for example, comprising an original image scanning part 1 and a reproduction image recording part 2. The original image scanning part 1 comprises an original image cylinder 3 for carrying an original image to be reproduced, a motor 4 for rotating the cylinder 3 in a main scanning direction, a rotation angle encoder 5 for detecting the angle of rotation of the cylinder 3, a onerevolution encoder 6 for detecting each revolution of the cylinder 3, a scanning head 7 for scanning the colour original image mounted on the cylinder 3, a feed screw 8 for moving the scanning head 7 in a subsidiary scanning direction parallel to the axis of rotation of the cylinder 3, and a drive motor 3 for rotating the feed screw 8.

The recording part 2 comprises a recording cylinder 10 for carrying a film to be exposed, a motor 11 for rotating the cylinder 10 in a main scanning direction abouts its axis, a rotation angle encoder 1 2 for detecting the angle of rotation of the cylinder 10, a one-revolution encoder 1 3 for detecting each revolution of the cylinder 10, a recording head 1 4 for recording a reproduction image on a film or the like mounted on the cylinder 1 0, a feed screw for moving the recording head 14 in a scanning direction substantially parallel to the axis of the cylinder 10, and a drive motor 1 6 for rotating the feed screw 1 5.

A colour separation device 1 7 is provided on the scanning head 7 and the colour separation device 1 7 outputs image signals obtained by scanning a colour original image mounted on the cylinder 3 as three colour separation signals, for example red (R), green (G), and blue (B) colour separation signals, and an unsharp signal (U). The three colour separation signals and the unsharp signal (U) output from the colour separation device are input to a tone operation circuit 1 8 which performs a series of necessary processes such as logarithmic conversion, colour correction, tone correction, detail emphasis and magnification conversion on the signals.The processed signals are output as four colour separation printing signals namely cyan (C), magenta (M), yellow (Y) and black (K) colour separation printing signals, each signal being related to each colour in the area of the original image represented thereby so that the tone of the original image can be represented in the reproduction printed image.

A timing pulse generating circuit 26 is provided to generate a timing pulse P1 and a onerevolution pulse P2, respectively in accordance with the pulses output by the rotation angle encoder 5 and the one-revolution encoder 6.

The magenta colour separation printing signal M is output from the tone operation circuit 18 to an analogue to digital (A/D) converter 20, which is controlled by the timing pulse P1 corresponding to the pitch or dimensions of the picture element in the main scanning direction.

The digitized signal M is stored in a first-in-first-out (FIFO) buffer memory in sequential order via a switch or latch 21 controlled by a clock signal derived via a divide-by-two circuit from the onerevolution pulse P2.

The cyan C, yellow Y and black K colour separation signals output from the tone control circuit 1 8 are digitized by an A/D converter 20 controlled by a clock signal derived via a divideby-two circuit from the timing pulse P1 so that these signals are digitized on a cycle twice as long as that on which the magenta colour separation signals M are digitized. Thus the rate at which the magneta signals are digitized is twice that for the other colour separation signals. The digitized signals C, Y, and K are input to a multiplexer 24. However, the multiplexer 24 is controlled so as not to be actuated by the one-revolution pulse P2 during the time when the magenta signal is written into the buffer memory 22, so that the C, Y, and K signals which correspond to one scanning line are cut off.

When the magenta signal M corresponding to the next or second scanning line is output from the tone operation circuit 18 through the A/D converter 19, the switch 21 is turned or operated by the one-revolution pulse P2 so that the magenta signal M is directly input to a condensing circuit 23. The magenta signal M of the preceding scanning line is input to the condensing circuit 23 from the buffer memory 22 in order in which the signal was stored synchronously with the magenta signal M actually being input directly to the condensing circuit 23.

Accordingly, the magenta signals M corresponding to a pair of picture elements adjacent each other in the subsidiary scanning direction are input as a pair to the condensing circuit 23. The condensing circuit 23, under control of the clock signal supplied to the A/D converter 20 and the clock signal P1, carries out the condensing or encoding process utilizing the non-linear quantization or the Hadamard conversion for each unit region or 2 x 2 matrix of picture elements and supplies condensed signals to the multiplexer 24.

Because the signals C, Y, K are input to the multiplexer 24 on a cycle twice as long as the pitch of the picture elements in the subsidiary scanning direction, the image signals output from the multiplexer 24 form the set of signals Moo, m01, m10, m", COO, YOO, K00 or the set of signals mOO, mO1. mrO, m", COO, YOO, K,,, as described above, and are written one after another into a picture image memory 29, such as magnetic disk storage means, via a buffer memory 25.

Fig. 7 is a block diagram showing one example of the condensing circuit or device for use where the condensing is carried out using non-linear quantization.

As shown in Fig. 7, the magenta signals for adjacent scanning lines input in parallel to the condensing circuit 23 from the buffer memory 22 and the A/D converter 19, respectively, are supplied to data selectors 40, and 402 from which the signals Moo, Mo1, M,o, M" for the magenta colour separation plate for each unit region, or 2 x 2 matrix of picture elements, are output in parallel.The signals Moo, Mo1, M10, M" are then input to adder-subtractor circuits 411, 412 and 413 as shown to provide the predictive error signals (Ma-Moo) (M1o-Moo) and (Mo1-Moo) Because the data selectors are clocked by the clock signal P1 which has a rate corresponding to the pitch of the picture elements to be scanned, the signals Moo, Mo1, M10, M11, of M plate are not output simultaneously from the data selectors. This problem is, however, overcome by providing appropriate sample and hold circuits or delaying devices (not shown) for the data selectors 40, and 403.

The predictive error signals (M"-Moo), Mzo-Moo) and (Mo1-Moo) output from the addersubtractor circuits 411, 412 and 413 are input to respective table memories 421, 422 and 423 as address signals. The table memories 421, 422 and 423 store the representative values as data at addresses given by the predictive error signals, the relationship between each address and the corresponding data being in accordance with the non-linear quantization described above with reference to Fig. 3. Thus the tables memories output the representative values m", m10 and my1 .

As described above, the signal m1, is a 6-bit signal while the signals mO, and m1O are 5-bit signals.

The representative value signals are supplied to a multiplexer 44 to which the signal Moo (an 8-bit signal) is also supplied via a sample and hold circuit 43. The multiplexer 44 outputs a condensed 24-bit (3 bytes) image signal under control of the same clock signal used to control the A/D converter 20. Thus, the multiplexer 24 outputs signals at a rate corresponding to the pitch of the unit region, that is the 2 x 2 matrix, of picture elements.

Fig. 8 is a block diagram of an arrangement for a condensing circuit in which the image signal is condensed using the Hadamard conversion. As shown in Fig. 8, the magenta signals M for adjacent scanning lines are input in parallel to data selectors 451 and 452 from the buffer memory 22 and the A/D converter 19 (via the switch 21), respectively, and the data selectors 451 and 452 output signals Moo, Mo1, M10 and M11 for each unit region as described in relation to Firure 7.The signals Moo, Mo1, M10 and M11 input to respective adder-subtractor circuits 46 to 464 as shown in Fig. 8 to derive the signals (M10-M11), (M10 + M11) (M00 - M01), (Moo + Mo1) which are input to adder/subtractor circuits 471 to 474 as shown.The adder-subtractor circuits 471 to 474 output the signals: (Moo - M01 - M10 + Mix) = 2m11; (Moo - M01 + M10 + Mix) = 2my1; Moo + M01 - M10 - M11) = 2m10;and (Moo + M01 + M10 + M11) = 2mOO,respectively In this case, as the signals Moo, Mo1, M10, M11 are all 8-bit signals, the processed signals output from the adder-subtractor circuits 471 to 474 are 10-bit signals.

The signals mO0 = i[Moo + M01 + M10 + M11] corresponds to the average of the respective tones of the signals Moo, Mo1, M10 and M11, and therefore it is necessary to have at least sufficient storage to represent the tone to a level almost the same as the original signal. To that end, the lower bits are removed in a respective bit reduction circuit 48 while the eight higher bits are output to a multiplexer 49 as the signal mO0.

Because two of the signals Moo, Mo1, M10, M11 are added and the remaining two values are subtracted, none of the values mO1, m10 andm11 is very large when there exists a certain correlation between the signals Moo, Mo1, Mo and M,1, as is the case to which the method in accordance with this invention is applied. However, if the signals mO1, m10, m11 are larger than a certain value, the higher bits of these signals are removed after carrying out the process of substituting them with the certain value using bit reduction circuits 481 to 483.The bit reduction circuits 482 and 483 output the lower 5 bits as the signals M01 and M10, respectively, while the bit reduction circuit 48 outputs the lower 6 bits as the signal M11 to the multiplexer 49. In this connection, as the reduction in the number of bits with respect to the lower 5 bits and the higher 6 bits is decided according to the correlation between the signals Moo, Moo1, M10 and M11, when 10 bits are obtained by the adding and subtracting operation carried out on the signals Moo, Mo1, M10 and M11, it is also possible for the values m11, mO1 to be formed by the intermediate 5 bits by cutting off the lower 1 bit and the higher 4 bits, for example.

Furthermore, it is also possible to achieve the required reduction in the number of bits by means of a table memory for non-linear quantization in the same manner as described in relation to Fig.

7.

With the image signals of a plurality of original images condensed and stored in the image memory 27 as described above, such processes as the relocation of the image data according to the desired position in the reproduction image are carried out either in the image memory 27 or in another image memory by means of an editing device (not illustrated) in accordance with the layout assignment to be used.

The edited data is then stored in an image memory 27' ready for the recording of a prescribed reproduction image.

In order to record the reproduction image, the image signal representing a unit region or 2 x 2 matrix of picture elements is read out of the image memory 27' and is supplied via a buffer memory 30 to a demultiplexer 31 where it is divided into cyan, yellow and black signals, C, Y, and K and the condensed magenta signal M. The signals C, Y, K are input directly to an operation circuit 33 for producing colour separation plates, while the condensed image signal m is input to the operation circuit 33 via a reproducing device 32.The signals C, M, Y, K are divided in the operation circuit 33 into signals representing separate picture elements of the 2 x 2 matrix, the magenta signal being used to determine the relationship between the representative picture element and the other picture elements, and the signals for the four colour plates corresponding to two scanning lines, for example, are input in parallel to a buffer memory 34. The buffer memory 34 comprises at last two FIFO or line memories. The four colour signals corresponding to the scanning line to be recorded are directly output to an output control circuit 35, while the four colour signals of the picture element corresponding to the next scanning line to be recorded are stored in the buffer memory 34.

Figs. 9 and 10 illustrate embodiments of the reproduction device 32 for use with the condensing circuits Figs. 7 and 8, respectively.

Referring now to Fig. 9, the condensed 24 bit (3 bytes) image signal m, input to the reproduction device 32 is divided in the signal Moo and the signals rn', m10 and mO1, corresponding to the representative values, by means of a data selector 50. Each of the signals rn11, m10 and mO, is input to a respective table memory 511, 512 and 513 as an address signal.

The table memories 511, 512 and 513 store the predictive error signals as data at addresses in accordance with the non-linear quantization shown in Fig. 3. Thus, the table memories 511, 512 and 513 output as data the predictive error signals (M11-Moo) (Mro-Moo) and (Mo1- M00) corresponding to the address signals m11, m10 and mO1 respectively. Each of the signals (M11 M00), (M10-Moo) and (Mo1-Moo) is added to the signal Moo in a respective adder circuit 521 522 and 523 so that the signals M11, M10, M01 and Moo are output from the reproduction device.

Referring to Fig. 10, there is shown a reproduction device 32 for use with the condensing circuit of Fig. 8. As shown, the 24 bit (3 bytes) signal m is divided into the signals m11, m10, mO, and mOO by means of a data selector 53, and the signals m11, m10, m01 and mOO are input as address signals to the reproduction table memories 541, 543, 543 and 544 where the higher bit(s) or lower bit(s) or both, as appropriate, are added to each of the signals m11, m10, m01 and mOO to convert the signals into 8-bit signals.The 8-bit signals output from the reproduction table memories 541 to 544 are supplied to adder-subtractor circuits 551 to 554 which output signals (- m11 + m01), (m1, + mOO), (- m10 + m00) and (m00 + m10), respectively. These signals are combined in adder-subtractor circuits 561 to 564 which produce the following output signals: ( + m11 - rn01 - rn10 + mOO) = M11.

( - rn11 + rn01 - m10 + mOO) = M10, (- m11 + mOo + m10 - m01) = M01,and (m11 + mOO + m10 + m01) = MOO,respectively.

As described above, the signal mOO has 8 bits representing the average tone of the signals Moo, Mo1, M10 and M11 and therefore the signal mOO is examined in view of the bit-reduction which occurred in the adding or subtracting process so that the reproduced values Moo, Mo1, M10 and M,1 have 8 bits representing tone.

Thus, contary to the previously proposed method wherein the signals M,,, Mo1 M10 and M11 are recorded as 8 bits each (i.e. 4 bytes in total), when the method described above is used, where predictive encoding is employed, the necessary number of bits is 8 bits for signal Moo, 5 bits each for the signals mO1, m10 and 6 bits for rnii' being 3 bytes in total, and, where the Hadamard conversion is used, the required bits of storage is 8 bits for the signal mOOm 6 bits for the signal m" and 5 bits each for the signals m01, m10 making 3 bytes in total, which means the storage capacity required for the brightness signal is reduced to three-quarters of that required by the previously proposed method.

Thus, with a method and apparatus embodying the invention, the degree to which the image signals as a whole are condensed is greatly increased since the image signal representing the brightness is condensed while assuring the quality of image sufficient for practical use and the average number of bits per picture element of the signal representing the brightness in the unit region to be reduced without bringing about any unnatural feeling to the visual sense of the reproduced colour image, thereby reducing the storage capcity required to within the bounds of the present state of the art for data storage.

Furthermore, as the condensing step is carried out on the basis of correlation of brightness between adjacent picture elements there is less possibility of adversely affecting the detail in the unit region and an exact tonal reproduction can be produced so that substantially the same image as the original image can be reproduced,.

Besides, since the condensing and reproduction steps are carried out independently for each unit region, when any graphic form or pattern is separately taken out of the original image establishing an unit region, the image of the separated portion can be reproduced exactly simply from the separated portion as in the previous method, and, in addition, by establishing the unit region to be a minimum unit of image for the editing operation, the speed of editing can be increased.

Claims

1. A method of condensing image signals comprising colour separation signals for each of a plurality of picture elements of an original image, which method comprises using the colour separation signals of a representative picture element to represent the colour image signals of a region of picture elements, selecting a respective single colour separation signal of each of the other picture elements of the region to represent the brightness of that picture element and encoding the selected colour separation signals to produce representative brightness signals.

2. A method for condensing image signals in which an unit condensing region comprising a plurality of picture elements is established, all of a plurality of image signals corresponding to colour signals necessary for reproducing a colour image are provided by a representative picture element of the unit condensing region, and only an image signal corresponding to a brightness signal of the plurality of image signals is provided by each picture element other than the representative picture element of the unit condensing region so that the total number of bits required for storing the image signal obtained by scanning a colour original image may be reduced, in which method at least the image signals representing the brightness of each picture element other than the representative picture element in each unit region are encoded to produce a representative brightness signal for the picture element.

3. A method according to Claim 1 or 2, wherein the encoding is carried out using the relationship between the colour separation signals representing brightness and the corresponding colour signal of the respective representative picture element.

4. A method according to Claim 1, 2 or 3, wherein the representative brightness signals are produced by determining the difference between each respective colour separation signal representing brightness and the corresponding colour separation signal of the respective representative picture element to produce difference signals and relating the difference signals to the representive brightness signals using a predetermined relationship.

5. A method according to Claim 4, wherein the predetermined relationship is a non-linear quantization.

6. A method of condensing image signals comprising colour separation signals for each picture element of an original image, substantially as hereinbefore described with reference to, and as illustrated in Figs. 2, 3, 4, 6, 7 and 9 of the accompanying drawings.

7. A method according to Claim 1, 2 or 3, wherein the representative brightness signals are derived from the selected colour separation signals using a matrix conversion factor.

8. A method according to Claim 7, wherein, for a unit region comprising four picture elements, the matrix conversion is

where mO, to m1, are the representative brightness signals for each of the picture elements and M01 to M11 are the corresponding selected colour separation signals while Moo is the corresponding colour separation signal for the representative picture element and mOO is the related brightness signal.

9. A method of condensing image signals comprising colour separation signals representing each picture element of original image, substantially as hereinbefore described with reference to, and as illustrated in Figs. 2, 5, 6, 8 and 10 of the accompanying drawings.

10. A method of reproducing an image from image signals condensed using a method in accordance with any one of Claims 1 to 9, which method comprises reversing the encoding step to decode the representative brightness signals, determining the colour separation signals for the picture elements other that the representative picture element using the decoded brightness signals and recording the colour separation signals of each picture element.

11. A method of reproducing an image from image signals condensed using a method in accordance with any one of Claims 1 to 6, substantially as herein before described with reference to Figs. 2 to 4, 6, 7 and 9 of the accompanying drawings.

1 2. A method of reproducing an image from image signals condensed using a method in accordance with any one of Claims 7 to 9, substantially as hereinbefore described with reference to Figs. 2, 5, 6, 8 and 10 of the accompanying drawings.

1 3. Apparatus for condensing image signals comprising colour separation signals for each of a plurality of picture elements of an original image, which apparatus comprises means for using the colour separation signals of a representative picture element to represent the colour image signals of a region of picture elements, and means for encoding a repective selected single colour separation signal representing the brightness of each of the other picture elements of the region to produce representative brightness signals.

14. Apparatus according to Claim 1 3, wherein the encoding means comprises means for obtaining difference signals representing the difference between the colour separation signal of the representative picture element corresponding to the selected colour separation signals and each selected colour separation signal and means for relating the difference signals to representative brightness signals.

1 5. Apparatus according to Claim 14, wherein the obtaining means comprises a plurality of adder-subtractor circuits.

16. Apparatus according to Claim 14 or 15, wherein the relating means comprises a table memory or memories in which the representative brightness signals are stored as data and for which the difference signals are used as address signals, there being a predetermined relationship between each addresss and the corresponding data stored in the memory or memories.

1 7. Apparatus according to Claim 16, wherein the predetermined relationship comprises a non-linear quantization.

18. Apparatus according to Claim 13, wherein the encoding means comprises a matrix transformation means.

19. Apparatus according to Claim 18, wherein the matrix transformation means is arranged to effect the following transformation:

where mO1 to m,1 are the representative brightness signals for each of the picture elements and M01 to M11 are the corresponding selected colour separation signals, while Moo is the corresponding colour separation signal for the representative picture element and mOO is the related brightness signal.

20. Apparatus according to Claim 1 8 or 19, wherein the matrix transformation means comprises a plurality of adder-subtractor circuits.

21. Apparatus for condensing image signals comprising colour separation signals for each of a plurality of picture elements of an original image, substantially as herein before described with reference to, and as illustrated in, Figs. 2, 3, 4, 6, 7 and 9 of the accompanying drawings.

22. Apparatus for condensing image signals comprising colour separation signals for each of a plurality of picture elements of an original image substantially as hereinbefore described with reference to, and as illustrated in, Figs. 2, 5, 6, 8 and 10 of the accompanying drawings.

23. Apparatus for reproducing an image from image signals condensed using apparatus in accordance with any one of Claims 1 3 to 22, which apparatus comprises means for reversing the operation of the encoding means to decode the representative brightness signals, means for determining the colour separation signals for the picture elements other than the representative picture element using the decoded brightness signals and means for recording the colour separation signals for each picture element.

24. Apparatus for producing an image from image signals condensed using apparatus in accordance with any one of Claims 13 to 17 and 21, substantially as hereinbefore described with reference to, and as illustrated in, Figs. 2 to 4, 6, 7 and 9 of the accompanying drawings.

25. Apparatus for producing an image from image signals condensed using apparatus in accordance with any one of Claim 1 8 to 20 and 22, substantially as hereinefore described with reference to, and as illustrated in, Figs. 2, 5, 6, 8 and 10 of the accompanying drawings.

26. Colour scanning and recording apparatus whenever using a method in accordance with any one of Claims 1 to 1 2 and/or apparatus in accordance with any one of Claims 1 2 to 25.

27. A reproduction image whenever produced using a method in accordance with any one of Claims 1 to 1 2 and/or apparatus in accordance with any one of Claims 1 2 to 25.

28. Any novel feature or combination of features described herein.