JPH012475A - Image processing method and device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a method and apparatus for digital data processing for determining digital data representing the color content of adjacent pixel blocks of an image.

Below is the margin [Prior art and problems to be solved by the invention]-e
Images, such as paintings or text, are often encoded into digital data that represents the color content of the image's pixels. Monochromatic images are encoded into a single set of pixels that define the gray level content of each pixel, whereas color images are encoded into each color component (colo
ur component) into several digital "separated" (color separated) images. Digitization of this image leads to the generation of very large amounts of data. For example, a typical pixel resolution is 6
．． Per 45afl (per square inch) 300x3
It is 00. It is often necessary to store this data or transmit the data to a remote station, and various techniques have been developed to compress the data in order to reduce storage space and transmission time.

One method is to divide the image into several image blocks, and the data in each block is then compressed. The inventor has discovered that although this technique is generally satisfactory, image blocks of the following types cannot be compressed.

Therefore, it is undesirable to apply a compression algorithm to these blocks, since the data that is compressed as a result of compression differs only slightly from the original data in terms of quantity.

[Means for Solving the Problems and Effects] According to one aspect of the present invention, there is provided a method for processing digital data representing the color content of adjacent blocks of pixels defining the outline of an image, the method comprising: i) determining, for each block, a measure of variation in the color content of pixels within the block; and (i1) digital data defining the outline of the block when the variation is within a predetermined range. processing the data according to a predetermined compression algorithm;
Provided is a method for processing digital data, the method comprising:

What the discoverers have discovered is that there are certain well-defined types of pixel blocks that can be compressed effectively, even though there are other types that cannot. Typically, a form of incompressible pixel 1' lock is one in which the color content varies significantly from pixel to pixel, meaning that in the frequency domain the block requires a large number of blocks to accurately represent it. This requires high frequencies.

Such blocks can include abrupt changes in color (edges). Another type of pixel block that cannot be effectively compressed is one in which the color content varies very slightly, but is not the same for each pixel. An example of such a pixel block is one that defines a gradual vignette (ie, a gradual change in color). In the frequency domain such a block would require a large number of low frequencies to represent it. Therefore, the present invention
Before the algorithm is actually passed, it is detected whether a block of pixels is suitable to pass the compression algorithm.

A typical compression algorithm that could be applied would involve transforming a block of pixels into the frequency domain and then encoding the resulting frequency coefficients according to a Huffman-type code.

Step (i) can be performed in the frequency domain by determining the energy of the block. This energy can be determined by adding the squares of all AC coefficients or just some of the high frequency coefficients. However, step (i) is preferably performed on a block of pixels in the spatial domain, since depletion removes the need to first transform the block into the transform domain.

In the spatial domain, the method may include: (a) for each set of pixels within a window having a predetermined dimension smaller than the block of pixels, determining a difference value between the color content values of the pixels; , the maximum difference value (W'six) determined based on the determined difference value
(b) the difference value (D) between the maximum and minimum color content values in the pixel block is equal to the threshold value (T, , T3); the second predetermined limit value (
Tz, Tz); and (c) determining whether the difference value (W'max
) is less than the first threshold (T1) and the stage (b
) obtained in the second predetermined range (Tz,
When outside T3), the digital data defining the block may be subjected to data processing according to a predetermined compression algorithm.

What is determined in step (a) is whether the color content of the block varies at relatively high frequencies indicating that the block should not be compressed; ) is to determine whether the color content also varies at relatively low frequencies indicating that the block should not be compressed. Of course, blocks can be compressed if the color content is the same throughout the block.

Typically the window will have dimensions of 2x2 pixels.

The first threshold value and the second limit value can be set directly and can be expressed in any suitable way. For example, the second limit value can be defined by a single value and a range.

According to a second aspect of the invention, an apparatus for processing digital data representing the color content of adjacent blocks of pixels defining the outline of an image comprises: (i) measuring variations in the color content of pixels in each individual of the blocks; and for each block, (i1) when the variation is within a predetermined range, subjecting the digital data defining the block to data processing according to a predetermined compression algorithm; The data processing means (2, 3) are provided.

Typically, the processing means will be defined by a suitably programmed computer. [Embodiments] A method of digital data processing and an apparatus for carrying out the method as an embodiment of the present invention will be described with reference to the accompanying drawings. The present invention relates to the processing of digital data representing the color content of adjacent blocks of pixels that determine the contours of an image. This digital data could be generated using an input scanner, for example from our Magna-scan series, or could be generated electronically by an operator without any scanning step. This latter method poses special problems when it is desired to compress digital data for subsequent storage or transmission. The reason for this is
This is because electronically generated images can display very sharp changes in contrast that are difficult to transform or represent in the frequency domain, given the need for a very large number of high frequency components. be. Generally, representing an image in the frequency domain, eg, after passing through a Fourier transform, leads to a set of data that can be relatively easily compressed, eg, by Huffman coding.

It is also difficult to screen sections of an image where the color content varies slowly, for example in slowly varying vignettes. In this case, if this part of the image were transformed into the frequency domain, it would require very low frequencies to represent it, which again is difficult to encode. However, if the color content is the same throughout a pixel block, this color content can be encoded very easily in the frequency domain by a single DC term.

In a typical compression system, the original image, which can be a multicolored or monochrome picture or text, or a combination of both, is typically rectangular and divided into many small blocks of pixels, e.g. 8x8 pixels. be divided. In that case, each block is transformed to the frequency domain and the data is compressed.

To address the above problem, the digital data representing the image and stored in this example in the disk memory 1 (FIG. 1) is sent to a compression selector circuit 2, which The contents of each block of pixels of the image are retrieved and the block of pixels is transferred either to the compression circuit 3 so that the contents of the block can be compressed, or directly to the multiplexer circuit 4 in an uncompressed state.

The output from the compression circuit 3 is also sent to a multiplexer 4 which generates a single output (compressed or uncompressed) which is sent to a memory or transmission circuit in the usual manner.

Although the circuit 2.3 may be implemented by hardware components, it will typically be defined by a suitably programmed computer.

Evaluation of the contents of a pixel block may be performed based on a representation of the block, either in the spatial domain or in the frequency domain. It is preferred to operate in the spatial domain, since this avoids the need to transform the data to the frequency domain, and to subsequently return to the spatial domain if it is determined that compression is not appropriate. .

FIG. 4 is a flow diagram illustrating the series of steps performed for each pixel block represented in the spatial domain. A 3.times.3 pixel block is considered for this example, and three examples of such blocks are illustrated in FIGS. 2(A)-2(c).

In the case of a monochromatic image there will be one set of pixel blocks, whereas in the case of a polychromatic image there will be multiple sets, one for each color component. In this latter case, the method to be described is performed for each block of each color component representation. To further simplify this description, the data values within each block are represented by decimal numbers varying between O and lO, whereas in practice the range is typical for each color component. between O and 255.

In the first step 5 of Figure 4, a 2x2 window is passed around the pixel block. At each position of the window within that block, the digital data value of the pixel in the window white is determined, and the difference between the maximum and minimum digital values W, , 8 is determined (step 6);
h If we consider the pixel block of FIG. 2(A) if the window is arranged as indicated by reference numeral 7A, the difference W□,=1. Similarly for the blocks shown in FIGS. 2(B) and 2(c), the values are 9 and 6, respectively, when the windows are at positions 7B and 7C.

A series of different values for Wm11X are generated for each of the four window positions. In the case of FIG. 2(A), these values would be respectively 1.1.0.1. The maximum value for WIl,,l is then determined and in step 8 (W',,1 ). In the case of the block in FIG. 2(^), W' m m 11 =1.

For the blocks of FIG. 2(B) and FIG. 2(c), W'
The values for 1° are 9 and 6, respectively.

In step 9, the value of W's a 11 is compared with a threshold T,. W l, □ is T, greater than or T
, this indicates that the block is unsuitable to be compressed since it appears to require some high frequency terms to be represented in the frequency domain. The value of T is set empirically and is shown in this example to have a value of 8 (Figure 3). Therefore, the results of step 9 are shown in Figure 2 (
The block in B) was found to be unsuitable for compression, whereas the blocks in FIG. 2(A) and FIG. 2(c) must undergo further processing. . If the block in Figure 2(B) is evaluated in this way,
The data is sent directly to multiplexer 4 so that it is passed through uncompressed (step 10).

Blocks that satisfy the condition W+, □<T, undergo another evaluation. Initially, the maximum and minimum pixel values within a complete block are determined in step 11, and then the difference (D) between these pixel values is calculated. Figure 2 (A
), D=1, and D=9 in FIG. 2(B).
, and D=7 in FIG. 2(c).

In step 12, D is set to two relatively close thresholds T2
, Tz, but this threshold is again determined empirically (Tz = 1. T3 = 3 in the example of Fig. 3) and due to the very low frequencies at which it is necessary to represent the block in the frequency domain. It limits the range of block energies that are inappropriate for data compression.

In this particular example, the block of FIG. 2A falls within its range and therefore the digital data within that block should be passed to multiplexer 4 uncompressed. However, in Figure 2 (B) and
The data in the block of diagram (c) exists above the threshold T3 and is therefore suitable for compression, and the compression algorithm is passed in step 13. This may take any conventional form, but typically involves transforming the block of data into the frequency domain and then converting the Huffman coding to coefficients.

As mentioned above, in some cases it may be desirable to assess the compressibility of a pixel block from its representation in the frequency domain. In the frequency domain, digital data is represented by a DC term and a number of AC coefficients, where the total number of data values is equal to the number of pixels in the block (although the majority of these cannot be O). In this method, a measure of the energy of a pixel block is determined in step 14 (FIG. 5) by summing the squares of all AC coefficients.

That energy (E) is then compared to a first threshold T, (step 15) which gives a first indication whether the energy is generally low or high. If E is smaller than T a (low energy), then E is a pair of lower thresholds T,, rt(L threshold ↑2.T.

(corresponding to ) (step 16) to determine whether the low energy of the block falls within a special region that should not be compressed. If E is within this region, the data is passed uncompressed (step 17). Otherwise, the data is sent to the compression circuit 3 and a compression algorithm is applied (step 18).

If step 15 determines that the energy of the block is generally high, the energy E is then compared to a high threshold T, (corresponding to threshold T,) in step 19. If the energy is greater than the threshold T, the data is passed uncompressed (step 20), otherwise the data is compressed.

Typically, in this case a representation of the pixel block in the spatial domain is maintained such that if the data is to be passed uncompressed, this spatial representation can be passed directly to the multiplexer 4. Alternatively, the transformed data can be transformed back to the spatial domain.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a digital data processing device for expressing color content of pixels of an image as an embodiment of the present invention. Figure 2(A), Figure 2(B) and Figure 2(c) are diagrams illustrating three different types of pixel blocks; Figure 3 shows the conditions under which pixel blocks are compressed or not compressed; FIGS. 4 and 5 show diagrammatically; and FIGS. 4 and 5 show, respectively, a flowchart illustrating a method of processing a block of pixels in the spatial domain and the transform domain, respectively. 1...Memory, 2...Pressure selector, 3...Compression circuit, 4...Multiplexer.

Claims (1)

[Claims] 1. A method for processing digital data representing the color content of adjacent blocks of pixels defining the outline of an image, which method comprises: (i) for each block, determining the color content of pixels within the block; (ii) when the variation is within a predetermined range, data processing the digital data defining the block according to a predetermined compression algorithm. A digital data processing method characterized by: 2. The method of claim 1, wherein step (i) is performed by determining the energy (E) of the block by adding the squares of all alternating current coefficients in the frequency domain. 3. The method of claim 1, wherein step (i) is performed on blocks of pixels in the spatial domain. 4. (a) for each set of pixels in a window having a predetermined dimension smaller than the block of pixels, determine a difference value between the color content values of the pixels, and determine based on the determined difference value; (b) the difference value between the maximum and minimum color content values in the pixel block; (D) is within a second predetermined limit value (T_2, T_3), with the lower of the limit values (T_2, T_3) being greater than zero; and (c) The difference value (W′_m_a_
x) is smaller than the first threshold (T_1), and the difference value (D) obtained in step (b) falls within a second predetermined range (T_1);
2. The method of claim 3, comprising the step of: subjecting the digital data defining the block to a data processing according to a predetermined compression algorithm when outside T_3). 5. An apparatus for processing digital data representing the color content of adjacent blocks of pixels defining the outline of an image, the apparatus comprising: (i) determining a measure of variation in the color content of the pixels in each individual of the blocks; and for each block (ii) when the variation is within a predetermined range, the digital data defining the block is subjected to data processing according to a predetermined compression algorithm; ), a digital data processing device comprising: 6. The apparatus of claim 5, wherein said data processing means comprises a suitably programmed computing device.