JPS63171087A - Orthogonal transform coding method for picture data - Google Patents

Abstract

PURPOSE:To raise the efficiency of compressing of picture data by sequentially adding converted data in blocks, comparing a frequency at the time when the added value reaches a specified value with a threshold frequency and sampling the converted data according to the compared result. CONSTITUTION:Halftone picture data is orthogonally converted in terms of two dimensions in an orthogonal transform circuit 11 and the converted data (y) is arranged in vertical and horizontal directions in order of the sequence of conversion function in the respective blocks. Next, a bit distribution decision part 20 connected to a coding circuit 12 sequentially adds the absolute value of the data (y) from a high frequency component side to a low frequency component side and obtains the converted data y0 when the added value reaches the specified value, then compares the corresponding frequency F0 with the threshold frequency F'. When F0 is higher than F', the area on the low frequency side is set as a zone sampling area and when F0 is lower than F', the area nearer to the low frequency side than to a circular arc centering the data (y) and passing through the data Y0 is set as the zone sampling area. And the code length of a specified bit is assigned to each data (y) of the respective areas and information Q expressing the bit distribution is transmitted to the circuit 12 to encode the data (y).

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image data encoding method for the purpose of data compression, and more particularly to an image data encoding method using orthogonal transformation.

(Prior Art) An image signal carrying a halftone image, such as a TV signal, has a huge amount of information, so a broadband transmission path is required for its transmission. Conventionally, attention has been paid to the fact that such image signals have large redundancy, and various attempts have been made to compress image data by suppressing this redundancy. Recently, it has become common practice to record halftone images on optical disks, magnetic disks, etc., and in this case, image data compression is widely applied to efficiently record image signals on the recording medium. ing.

As one of such image data compression methods, one that utilizes orthogonal transformation of image data is well known. This method divides digital two-dimensional image data into blocks with an appropriate number of samples, and orthogonally transforms the numerical sequence consisting of the sample values for each block.This transformation concentrates energy in a specific component, so the energy By assigning a long code length to a large component of
This reduces the number of codes per block. The above-mentioned orthogonal transformation includes Fourier (F).

ur 1er) transformation, cosine (Cosine) transformation,
Although Hadamard transform, Karhunen-Loeve transform, Haar transform, etc. are often used, the above method will be explained in more detail by taking Hadamard transform as an example. First, as shown in FIG. 2, digital two-dimensional image data is divided into two blocks in a predetermined one-dimensional direction to form the blocks. When the two sample values x(0) and x(1) in this block are expressed in a rectangular coordinate system, they are highly correlated as described above, so x(1) = x as shown in Figure 3. They are mostly distributed near the straight line (0). Therefore, this orthogonal coordinate system is transformed by 45° as shown in Figure 3 to create a new V (0) −V (
1) Define the coordinate system. In this coordinate system, y(0) indicates the low frequency component of the original image data before conversion, and y(0) is a slightly larger value (approximately f2 times) than X(0) and x(1). However, on the other hand, it indicates the high frequency components of the original image data.

y(1) will be distributed only in a very narrow range close to the y(0) axis. So, for example, the above X(0),
If 7-bit code length is required for each encoding in 1), then y(0) will require 7-bit or 8-bit code length.
On the other hand, y(1) is, for example, 4 bits.
Encoding can be performed with a code length comparable to that of a pit, and in the end, 1
The code length per block is reduced and image data compression is achieved.

Above, we have explained the second-order orthogonal transformation that constitutes one block for each two pieces of image data, but as the order increases, the tendency for energy to concentrate on a specific component becomes stronger, increasing the effect of reducing the number of bits. Can be done. In general, the above transformation can be performed using an orthogonal function matrix, and in the extreme, if the eigenfunction of the target image is selected as the orthogonal function matrix, the transformed image becomes its eigenvalue matrix, and the pair of matrices This means that the original image can be expressed using only the angular components. Also, in the above example, image data is grouped into a block in only one dimension, but this block may be composed of several pieces of image data spanning two dimensions, in which case one-dimensional orthogonal transformation is performed. A more remarkable effect of reducing the number of bits can be obtained than in the case where the number of bits is reduced.

The transformation data obtained by the above-mentioned two-dimensional orthogonal transformation is determined by the sequence (O
(the number that crosses it) is arranged in order. Since this sequence corresponds to the spatial frequency, each converted data is arranged in frequency order in the vertical and horizontal directions, as shown in FIG. Therefore, a relatively long code length is assigned to the transformed data that carries the low frequency component (the data on the upper left side of Figure 4). ), a relatively short code length is assigned to the converted data that carries high frequency components (the data on the lower right side of Figure 4), or
Alternatively, the code length per block is reduced by truncating.

The above-mentioned truncation of the converted data is performed by so-called zone sampling, as is well known.

In other words, as shown in FIG. 4, the converted data are arranged vertically and horizontally in order of frequency within each block B, so the arc R centered around the converted data y(1,1) of sequence 0
Converted data y of the same frequency component is arranged above.

Therefore, if, for example, the converted data y in the shaded zone Z in this M4 diagram is sampled and only the sampled converted data y is encoded, significant data compression can be achieved.

(Problem to be Solved by the Invention) However, since the distribution state of the transformed data in the block B is different for each image and for each block, if the above-mentioned zone Z is fixed, in a certain block, Conversion data that is useful for image reproduction may be truncated, degrading the image quality of the reproduced image, or conversely, conversion data that is not very useful in a certain block may also be sampled, making it impossible to improve the image data compression effect. Become. In order to solve this kind of problem, conventionally, multiple sampling zone patterns are prepared, an appropriate zone pattern is selected after recognizing the conversion data distribution state of each block, and the processing is performed based on the selected pattern. It is also being considered to perform zone sampling.

However, in such a method, there is a practical limit to the number of zone patterns that can be prepared, and therefore, it is natural that zone sampling may be performed based on a zone pattern that is not very appropriate.

The present invention has been made in view of the above circumstances, and provides an image data compression method that can maximize the image data compression effect for any image data block without impairing the image quality of the reproduced image. The purpose of this invention is to provide an orthogonal transform encoding method.

(Means for Solving the Problems) The first method of orthogonal transform encoding of image data of the present invention is to convert the characteristic values corresponding to the respective absolute values of the transform data in each block from the high frequency component side to the low frequency component side. In each block, the frequency corresponding to the converted data when this added value reaches a predetermined value is compared with a pre-specified limit frequency, and the above frequency is higher than the limit frequency. In this case, only the converted data on the lower frequency side than the limit frequency is zone-sampled, and when the frequency is below the limit frequency, only the converted data on the lower frequency side than the frequency is zone-sampled and encoded. It is characterized by the fact that

The second orthogonal transform encoding method of image data of the present invention basically performs zone sampling in the same way as the first method, but there is a difference in density (so-called block Specifically, when the difference between the sampling areas in adjacent blocks exceeds a predetermined value, the sampling area is adjusted so that the difference is within the predetermined value. It is characterized by being corrected.

Note that as the above-mentioned "characteristic value corresponding to the absolute value of the converted data", the absolute value of the converted data itself, the square value of the converted data, etc. can be used.

(Function) When converted data that is not very important for image reproduction is relatively widely distributed in the high frequency region on the lower right side in FIG. Converted data where the added value of the characteristic values reaches a predetermined value exists on the lower frequency side. Therefore, in this case, according to the above method, the zone sampling area is set to be relatively narrow, and much of the less important conversion data is discarded, thereby increasing the image data compression effect.

On the other hand, when there are many converted data with large absolute values in the high frequency region shown in Figure 4 (they are important to some extent for image reproduction), the converted data where the sum of the above characteristic values reaches a predetermined value is , exists on the higher frequency side.

Therefore, in this case, the zone sampling area is set relatively wide, and although the image data compression effect is reduced to a certain extent, converted data useful for image reproduction will not be discarded.

Also, as mentioned above, if the frequency at which the sum of the above characteristic values is reached is higher than the limit frequency, the converted data on the lower frequency side than this limit frequency is zone sampled, so When converted data that is not very important for image reproduction is widely distributed, the zone sampling area is not expanded unnecessarily, and the data compression rate is kept high.

(Example) Hereinafter, the present invention will be described in detail based on an example shown in the drawings.

FIG. 1 schematically shows an apparatus for carrying out the first method of orthogonal transform encoding of image data according to the present invention. Image data (original image data) X representing a halftone image is first passed through a preprocessing circuit 10 and subjected to preprocessing such as smoothing to remove noise and other steps to improve data compression efficiency. The image data X that has undergone this preprocessing is passed through the orthogonal transformation circuit 11 and first undergoes two-dimensional orthogonal transformation. For example, as shown in FIG. 6, this two-dimensional orthogonal transformation is performed for each rectangular block of the number of samples (number of pixels) MXN in the halftone image G indicated by the image data X. Note that as this orthogonal transformation, for example, the aforementioned Hadamard transformation is used. This Hadamard transformation is
Since the transformation matrix consists of only +1 and -1, it can be performed by a simpler transformation circuit compared to other orthogonal transformations. Furthermore, as is well known, two-dimensional orthogonal transformation can be reduced to one-dimensional orthogonal transformation. In other words, by applying one-dimensional orthogonal transformation in the vertical direction to the image data regarding MXN pixels in the two-dimensional block B, and then applying one-dimensional orthogonal transformation in the horizontal direction to the obtained MXN transformation data, A two-dimensional orthogonal transformation is performed. Note that the order of vertical and horizontal conversion may be reversed.

The transformation data y obtained by the above two-dimensional orthogonal transformation is, as shown in FIG.
alsh functions, trigonometric functions in Fourier transform, etc.) are arranged vertically and horizontally in sequential order. As described above, this sequence corresponds to the spatial frequency, so the transformed data y are arranged in the vertical and horizontal directions in the order of the spatial frequency (that is, in the order of the density of the detail components of the image). In this Figure 4, the conversion data y (1, 1> in the topmost leftmost column is sequence 0 (zero)).
As is well known, this conversion data V
(1, 1> indicates the average image density within block B.

The converted data y arranged in this manner is sent to the encoding circuit 12 and encoded as shown in FIG.

This encoding circuit 12 encodes each converted data y in block B with a code length (number of bits) according to an allocation bit allocation table created as described later.

The above bit allocation table allocates a unique number of bits to each sequence, as shown in FIG. 7, for example.
As mentioned above, energy is concentrated in the low frequency components of the converted data y, so a relatively long code length is given to the low frequency components with high energy, while a relatively short code length is given to the high frequency components with low energy. By providing , the required number of bits per block B is reduced and image data compression is achieved. Furthermore, as shown in FIG. 7, by assigning 0 bits to the converted data y that carries particularly high frequency components and truncating it, and sampling only the other converted data y (so-called zone sampling),
The image data compression effect can be further enhanced.

However, if the zone sampling area is too wide, it will be impossible to sufficiently increase the image data compression effect, and on the other hand, if the zone sampling area is too narrow,
Much of the conversion data y useful for image reproduction will be discarded. Hereinafter, features of the method of the present invention for appropriately setting the zone sampling area in accordance with the distribution state of the transformed data y, which differs from image to image and from block to block, will be explained.

A bit allocation determining unit 20 connected to the encoding circuit 12 as shown in FIG. Add sequentially toward the components. In other words, this addition starts from the sequence-maximum conversion data V (M, N) and follows an arc R centered on the conversion data y (1° 1), as shown by the arrow H in Fig. 4. This is done by adding up the absolute values of each converted data y along the route. Further, the bit allocation determining unit 20 stores a predetermined limit frequency F' as shown in FIG. I remember it. This arc R' defines the maximum zone sampling area in which a data compression rate higher than a desired value can be obtained.

In other words, if the zone sampling area is expanded beyond this arc R' to the higher frequency range side, there is a possibility that the desired data compression rate may not be achieved. Note that such a circular arc R' can be obtained experimentally or empirically.

The bit allocation determining unit 20 obtains converted data 'Jo' when the above-mentioned addition value reaches a predetermined value. Next, the bit allocation determining unit 20 determines which is higher, the frequency FO corresponding to the converted data ya or the limit frequency F'. When the frequency FO corresponding to the converted data yo is higher than the limit frequency F' (the state shown in FIG. 4), the bit allocation determining unit 20 sets the region on the lower frequency side of the arc R' as a zone sampling region. Let it be Z. On the contrary, if the frequency FO is less than the limit frequency F', that is, the fifth
As shown in the figure, the above conversion data y. is closer to the converted data y(1,1) than the converted data y' on the arc R', the bit allocation determining unit 20 distributes the converted data around the converted data y(1,1>). Zone sampling of the region on the lower frequency side than the arc Ro passing through 'io
Let it be am. Then, the bit allocation determining unit 20 allocates a code length of 1 bit or more only to each converted data y of the zone sampling area Z determined in this way, and allocates O (zero) bits to other converted data y. . The number of bits to allocate to each converted data y given a code length of 1 pit or more is as follows:
The determination may be made according to a pre-stored allocation table, or the actual value of each conversion data y may be checked and the determination may be made accordingly. The flow of the process for determining the bit allocation in the bit allocation determining section 20 described above is shown in FIG. 8 in an easy-to-understand manner.

Information Q indicating the bit allocation determined as described above is sent to the encoding circuit 12, and the encoding circuit 12 encodes each converted data y according to this bit allocation information Q. At this time, the converted data y to which O pits are assigned as described above is truncated, and only the converted data y to which a code length of 1 bit or more is given is sampled and encoded.

Here, the area of the converted data y to which O bits are assigned was determined as described above, so in block B, the converted data y is relatively wide in the high frequency area (which is not very important for image reproduction, that is, the absolute value is small). If they are distributed, the zone sampling area is set to be relatively narrow. Therefore, encoding less useful transform data y is avoided and the required number of bits per block B is greatly reduced.

On the other hand, in block B, there is a relatively important (
In other words, if the converted data y (which has a relatively large absolute value) is distributed over a wide range up to a high frequency region, the zone sampling region is set relatively wide. Therefore, in this case, although the number of bits required per block B is greater than in the above case, the converted data y useful for image reproduction will not be truncated. However, even if the image data y, which is considered to be relatively important, is distributed over a wide range of high frequency regions, the zone sampling region is not expanded unnecessarily, and only regions whose frequencies are at most below the critical frequency F' are sampled. Therefore, a high data compression rate can always be achieved.

The image data Q(V) encoded as described above and the position information of the arc RO are recorded on a recording medium (image file) such as an optical disk or a magnetic disk by the recording/reproducing device 13 as shown in FIG. be done. As described above, this image data g(y) has been significantly compressed with respect to the original image data X, so a large amount of images can be recorded on a recording medium such as an optical disk. During image reproduction, the image data g(y) and the position information of the arc RO are read out from the recording medium and decoded into the converted data y in the decoding circuit 14. The transformed data y thus decoded is sent to the inverse transform circuit 15 and undergoes the inverse transform with the two-dimensional orthogonal transform. Thereby, the original image data X is restored, this original image data X is sent to the image reproduction device @16, and the image carried by the data X is reproduced.

Next, an embodiment of the second method of the present invention will be described. This second method can also be carried out by an apparatus having basically the same configuration as the apparatus shown in FIG. However, in this case, the bit allocation determining section 20 is provided with a storage means for storing the zone sampling area of the adjacent block B. The bit allocation determining process will be described below with reference to FIG. 9, which shows the flow of the bit allocation determining process, and FIG. 10, which explains sampling area modification. The bit allocation determining unit 20 determines the zone sampling area and bit allocation in each block B in the same manner as in the first method, but the difference between the sampling areas in two adjacent blocks is kept within a predetermined value. Make corrections at any time. To explain this, for example, as shown in FIG. 10, the block Brl-
It is assumed that the zone sampling area of 1 is an area on the lower frequency side than the circular arc R, and the zone sampling area of the nth block Bn subjected to orthogonal transformation is an area on the lower frequency side than the circular arc Rn. When determining the bit allocation for block Bn, the bit allocation determining unit 20 stores information indicating the circular arc Rn-+ in the storage means, and calculates the difference in radius between the inner arc Rn-+ and the circular arc Rn. Then, if the difference between the two radii exceeds a predetermined value r, the bit allocation determining unit 20 sets the arc R so that the difference becomes the self.
Correct n. In FIG. 10, the radius of the arc Rn is
Since it exceeds the value of the radius of -1 plus r, it is shown that it is corrected to an arc Rnl with this value as the radius, but the radius of the arc Rn is the radius of the arc Rn-+ minus r. Even if the radius is less than this value, the radius of the arc Rn is corrected to this value. The bit allocation determining unit 20 selects the arc Rnl or the arc Rn modified in this way.
A zone sampling area is defined by (if no modification is required), and bit allocation in that area is determined. Then, the encoding circuit 12 encodes the converted data y according to this bit allocation. Further, information indicating the circular arc Rn or Rn+ is stored in the storage means in place of the information indicating the circular arc Rn-+ of the block B n-+, and the zone sampling area of the block B n@ to be subjected to the next orthogonal transformation is stored in the storage means. ,
It will be used to make corrections if necessary.

As described above, by ensuring that the zone sampling areas of adjacent blocks do not deviate greatly, it is possible to prevent the occurrence of so-called block distortion, in which density differences occur at the boundaries of each block in the reproduced image. .

In the embodiment described above, the absolute value of the converted data y is added in each block and the added value is compared with a predetermined value, but the characteristic value added in this way corresponds to the above absolute value. It is sufficient if the sign is constant; therefore, for example, the square value of the converted data y may be added. Furthermore, as mentioned earlier, the lower the frequency of the transform data y, the more important it is for image reproduction, so if we use a value weighted by the reciprocal of the spatial frequency carried by each transform data y as the above characteristic value, , more preferred.

Further, the predetermined value to be compared with the added value of the above characteristic values may be an optimal value determined in advance, or may be a value obtained by accumulating all the characteristic values of block B and multiplying this cumulative value by a predetermined ratio. Moreover, all other characteristic values in block B are accumulated except for the characteristic value related to the converted data V (1, 1) of sequence O, which usually takes a particularly large value, and a predetermined ratio is applied to this accumulated value. It may also be a multiplied value.

In addition, in order to prevent the zone sampling area from being set very narrow, a second limit frequency lower than the above-mentioned limit frequency F' is set, and the converted data when the added value reaches a predetermined value is set. Even if M2 is lower than the second limit frequency, the zone sampling region may be set to a region lower in frequency than the limit frequency of M2.

(Effects of the Invention) As explained in detail above, the orthogonal transform encoding method for image data of the present invention is useful for performing zone sampling of transform data obtained by orthogonally transforming image data block by block. This allows the sampling area to be set particularly small after all data has been sampled. Therefore, according to the method of the present invention, it is possible to significantly increase the data compression rate without deteriorating the quality of images reproduced through data compression.

Particularly in the second method of the present invention, since the sampling areas of adjacent blocks are not far apart, it is possible to prevent block distortion from occurring in the reproduced image.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a schematic configuration of an apparatus for carrying out the method of the present invention, FIGS. 2 and 3 are explanatory diagrams explaining orthogonal transformation according to the present invention, and FIGS. 4 and 5 are diagrams illustrating the orthogonal transformation according to the present invention. FIG. 6 is an explanatory diagram explaining the block division of an image according to the present invention; FIG. 7 is an explanatory diagram explaining the bit allocation table according to the present invention; FIGS. 9 is a flowchart showing the flow of bit allocation determination processing in the first and second methods of the present invention, respectively. FIG. 10 is an explanatory diagram illustrating zone sampling in the second method of the present invention. 11... Orthogonal transform circuit 12... Encoding circuit 2
0... Bit allocation determining unit F'... Limit frequency B s B n-+ s 8 n N B nn-block of image data R1RO% Rn-+v Rn 1R
n'...Circular arc r centered on the conversion data of sequence ○...Difference in the radius of the arc X...Original image data y...Conversion data yo...Additional value of characteristic values is a predetermined value Converted data a (y) when reached... Encoded image data Z... Zone sampling area Figure 6 Figure 7 Figure 5

Claims (7)

[Claims] (1) After applying orthogonal transformation to the two-dimensional image data for each block based on a predetermined orthogonal function, the transformed data arranged in the sequence of the orthogonal function within the block after undergoing this transformation is uniquely In the orthogonal transform encoding method for image data that is encoded with a code length of In the block, the frequency corresponding to the converted data when this added value reaches a predetermined value is compared with a pre-specified limit frequency, and if the frequency is higher than the limit frequency, the frequency is lower than the limit frequency. Sample only the side conversion data,
Orthogonal transformation of image data, characterized in that when the frequency is below a limit frequency, only the transform data on the lower frequency side than the frequency is sampled, and each of the sampled transform data is encoded with a unique code length. Encoding method. (2) The orthogonal transform encoding method for image data according to claim 1, wherein the characteristic value is an absolute value of the transformed data. (3) The orthogonal transform encoding method for image data according to claim 1, wherein the characteristic value is a square value of the transform data. (4) The image data according to any one of claims 1 to 3, wherein the characteristic value is weighted by the reciprocal of the spatial frequency carried by the corresponding transformation data. Orthogonal transform encoding method. (5) The predetermined value is a value obtained by multiplying a cumulative value of all characteristic values in the block by a predetermined ratio, according to any one of claims 1 to 4. An orthogonal transform encoding method for image data. (6) The predetermined value is sequence 0 within the block.
Claims 1 to 4, characterized in that the value is the cumulative value of all other characteristic values except for the characteristic value related to conversion data of (zero) multiplied by a predetermined ratio. An orthogonal transform encoding method for image data. (7) After applying orthogonal transformation to the two-dimensional image data for each block based on a predetermined orthogonal function, the transformed data arranged in the sequence of the orthogonal function within the block after undergoing this transformation is uniquely In the orthogonal transform encoding method for image data that is encoded with a code length of In the block, the frequency corresponding to the converted data when this added value reaches a predetermined value is compared with a pre-specified limit frequency, and if the frequency is higher than the limit frequency, the frequency is lower than the limit frequency. Sample only the side conversion data,
If the frequency is below the limit frequency, only the converted data on the lower frequency side than the frequency is sampled, each of the sampled converted data is encoded with a unique code length, and the sampling in adjacent blocks is performed. 1. An orthogonal transform encoding method for image data, characterized in that, when a difference between regions exceeds a predetermined value, the sampling regions are modified so that the difference is within a predetermined value.