JPH1155528A - Image coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the image coder that realizes a high compression rate regardless of its low cost. SOLUTION: The image coder consists of a binarization means and a coding means. The binarization means is provided with a fixed threshold level binarization section 11, a dither method binarization section 12 and a selector 13 that selects either of binarized data Ic, Id based on a selection signal S. The coding means is provided with a typical prediction TP section 21 that provides an output of a signal for a TP, an adaptive template AT section 22 that provides an output of a signal denoting coordinate information of an AT pixel, a model template 23 that generates a context, and an arithmetic coding section 24 that conducts Markov model coding. Since the selection signal S is given to the AT section 22, the binarization method of binary image data I received by the coding means is known and an optimum AT pixel coordinate depending on a pixel is selected without checking the property of the binary image data I and then rational coding is conducted at a low cost.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding apparatus, and more particularly, to an apparatus for handling binary images, such as a facsimile, which can achieve a higher compression ratio regardless of the type of image. The present invention relates to an image encoding device that determines an encoding parameter.

[0002]

2. Description of the Related Art In a facsimile, an image captured by a scanner is binarized and transmitted. In order to binarize a captured image with high quality, it is general to binarize a character region and a photograph region in the image by the following methods. That is, characters are binarized using a fixed threshold value (fixed threshold method) because the resolution is important, and photographs are binarized using the dither method because it is important to produce gradation. Widely used dither methods include systematic dither method and error diffusion method. In the error diffusion method, the threshold value of the pixel of interest is already 2
It changes depending on the original density of the pixel whose value has been converted. Therefore, the output binary image has no periodicity and contains a large amount of information, and thus the encoding efficiency when encoding is low. On the other hand, in the systematic dither method, since the magnitude of the threshold value of each pixel does not change and is constant as in the error diffusion method, the output binary image becomes periodic, and the error diffusion method , The amount of information is smaller than that of the binary image, and the coding efficiency when coding is large.

FIGS. 5 and 6 are views for explaining a Bayer type dither method which is one of the systematic dither methods. The threshold period of the commonly used Bayer-type dither method is 4
× 4, the density range of the input multi-valued image is 0 to 25
In the case of 5, the threshold matrix is, for example, as shown in FIG. FIG. 6 shows an example in which an image having a constant density is binarized using a threshold as shown in FIG. 5, and it can be seen that a periodic pattern in which white pixels and black pixels alternate is generated.

Incidentally, as an option of the facsimile coding method, JBI according to ITU-T recommendation T.82 is used.
There is a G (Joint Bi-level Image Coding Experts Group) system. The JBIG method realizes Markov model coding of a binary image, and is lossless coding in which the code amount is close to entropy. According to information theory, the entropy is maximum when the ratio of white to black appearance is 0.5, and the ratio is 0
Or, the entropy is smaller as it is biased toward 1. A group of coded pixels around the target pixel to be coded is referred to as a context. In the JBIG method, it is estimated whether the target pixel is white (0) or black (1) by referring to the context. Since the estimation probability is accurately captured, the code amount is smaller than when no context is referred to.

FIGS. 7 (A) to 7 (C) show a group of pixels, in which a pixel marked with "?" Is set as a target pixel, and the other 10 pixels are set as a context. For example, FIG.
When the context is pure white as in (A), the probability that the target pixel is white is high. In the context as shown in FIG. 7B, the pixel of interest is estimated to be a part of the black line of the character, so that the probability of being black is high. When the context is a part of the dither pattern as shown in FIG. 7C, the pixel of interest has a high probability of being black to supplement the pattern. Thus, referring to the context, it can be seen that whether the target pixel is black or white can be more accurately estimated, and the code amount can be reduced.

In the context of the JBIG system, FIG.
There are a two-line model shown in FIG. 8A and a three-line model shown in FIG. The pixel marked with “?” Is the pixel of interest, and the pixel marked with “F” is a floating pixel described later. Before sending the image data, a decision is made between the sending and receiving facsimile machines to determine whether to use the two-line model or the three-line model. The floating pixel shown in FIG. 8 has a higher correlation with the target pixel than the floating pixel, and can be replaced with a pixel that can accurately estimate the color (black or white) of the target pixel. This is called an AT (Adaptive Template), and a pixel to be replaced with a floating pixel is called an AT pixel. A
Since information on whether to use T and information on the coordinates of the AT pixel when the AT is used are to be attached to the header of the code, there is no need to search for the coordinates of the AT pixel on the decoding side.

[0007]

However, in order to provide an AT function, it is necessary to calculate a correlation between an AT candidate pixel and all target pixels, and a device capable of realizing the correlation is very complicated. ,There will be a cost. However, in facsimile, when encoding binary data, which binarization method is used, for example, a fixed threshold method or a dither method is selected. Since the binarization device and the encoding device are provided in the same facsimile terminal, the AT device can be quickly determined by knowing the binarization method without searching for the AT candidate pixel each time by the encoding device. can do.

For example, the sender can select either a text mode or a photo mode by using a button. When the text mode is selected, the sender is binarized by the fixed threshold method, and when the photo mode is selected, 2 by Bayer-type dither method
It is assumed that the value is to be converted into a value and the threshold period of the Bayer dither method is 4 × 4. When binarization is performed by such a Bayer-type dither method, the pixels four pixels before and eight pixels before the target pixel have the same threshold value, so that there is a high probability that the two pixels have the same color (white or black). . That is, these pixels have a high correlation with the pixel of interest, and are optimal as AT pixels to be changed. Therefore, in the two-line model and the three-line model, the AT pixels are preferably selected as shown in FIGS. 9A and 9B, respectively. Therefore, when the sender selects the character mode, the AT is not used, and when the photograph mode is selected, the AT pixel may be the pixel shown in FIG. In addition, as a method of binarizing pixels without worrying about the type of the original, image area separation is performed to determine whether the target pixel corresponds to a character area or a photographic area. There is a method of performing binarization and performing binarization in a photographic area by a dither method. In this case, if the statistics of the ratio between the character region and the photograph region are set in advance, and the AT pixels are set to be suitable for the binarization method performed on the region having a high statistical ratio at the time of encoding, Encoding efficiency can be improved. As described above, even if the AT pixel is determined,
If the information that T is used and the coordinate information of the AT pixel are encoded in a format according to Recommendation T.82, the information can be decoded, so that no problem occurs.

The present invention has been made in view of the above points, and has a low cost by determining an AT pixel based on a binarization method without using a complicated and expensive device. It is an object of the present invention to provide an image encoding device that realizes a high compression ratio while achieving a high compression ratio.

[0010]

According to a first aspect of the present invention, there is provided an encoding method wherein a selector selects from a plurality of binary image data and encodes the input binary image data based on a preset encoding method. Means for inputting a selection signal for controlling the selector to the encoding means together with the selector, wherein the encoding means specifies an encoding condition based on the input selection signal, and specifies the specified condition. In performing the encoding, the encoding means does not need any special means for examining the properties of the binary image data, and can perform the encoding efficiently at low cost. It is made possible.

According to a second aspect of the present invention, in the first aspect of the present invention, as the plurality of binary image data, input multi-valued image data to be encoded is binary-coded by a plurality of binary methods. The encoding means does not require any special means for examining the properties of the binary image data, so that encoding can be performed efficiently at low cost. .

According to a third aspect of the present invention, in the first or second aspect of the present invention, the encoding means controls the selector to perform an operation of controlling the selector as means for designating an encoding condition based on the input selection signal. Statistical means for taking the statistics of, characterized in that the coding conditions are specified based on statistical data output from the statistical means,
In this way, encoding can be performed efficiently.

According to a fourth aspect of the present invention, in the first aspect of the present invention, the binarizing means inputs multi-level image data to be encoded, and selects the selector based on the multi-level image data. Is provided with a selection signal generating means for generating a selection signal for controlling the operation of the image data, so that the binarization can be performed without regard to the type of the document image.

According to a fifth aspect of the present invention, in accordance with any one of the second to fourth aspects, the plurality of binarization methods are a fixed threshold method and a dither method. Is specifically given.

The invention of claim 6 is characterized in that, in the invention of any one of claims 1 to 5, the encoding method is Markov model encoding, and the encoding method is specifically given. Things.

According to a seventh aspect of the present invention, in the first aspect of the present invention, a parameter corresponding to the coordinates of an AT pixel is used as the encoding condition.
Since it is possible to quickly determine AT pixels without using a complicated AT device for searching for AT candidate pixels in the encoding device,
This makes it possible to execute reasonable encoding at low cost.

According to an eighth aspect of the present invention, in the seventh aspect of the present invention, the encoding means is configured to determine that pixel data of a line to be encoded in the binary image data and pixel data of a line immediately above the line to be encoded are included. A typical prediction unit that performs a typical prediction of skipping encoding when they match, an AT unit that outputs coordinate information of an AT pixel corresponding to the specified parameter, and a coordinate information output from the AT unit. A model template that forms a context based on the model template, and an arithmetic coding unit that performs Markov model coding with reference to the model template, so that a more specific configuration of the coding apparatus is provided. Things.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION

(Embodiment 1) FIG. 1 is a block diagram for explaining an embodiment of an image coding apparatus according to the present invention. The configuration of the device according to the present embodiment and its operation will be described with reference to FIG. The image encoding device according to the present embodiment includes a binarizing unit and an encoding unit. The binarizing means inputs the multi-value image data G, binarizes it with a fixed threshold value, and outputs the binary image data Ic. A dithering binarizing unit 12 for binarizing and outputting binary image data Id; a fixed threshold binarizing unit 11 and a dithering binarizing unit according to a selection signal S of either a character mode or a photograph mode A selector 13 selects one of the binary image data Ic and Id output from the selector 12 and outputs it as the binary image data I. On the other hand, the encoding means receives the binary image data I (Ic or Id) output from the binarization means, and outputs signals SLNTP, TP (typical prediction).
A TP unit 21 that outputs TPBALUE, an AT unit 22 that outputs ATMOVE representing coordinate information of a pixel having a higher correlation with the target pixel based on the selection signal S, a model template that forms a context for encoding including the AT pixel 23, and an arithmetic coding unit 24 for performing Markov model coding with reference to the model template 23. Arithmetic encoding section 24 outputs encoded data C.

(Embodiment 2) FIG. 2 is a block diagram for explaining another embodiment of the image encoding apparatus according to the present invention. The configuration of the device according to the present embodiment and its operation will be described with reference to FIG. The image encoding apparatus according to the present embodiment includes a binarizing unit and an encoding unit, and the binarizing unit determines a character area or a photograph area for each pixel based on the input multi-valued image data G. An image area separation unit 14 that outputs a signal (selection signal) Sp representing the result of the image area separation;
A fixed threshold binarization unit 11 that inputs multi-valued image data G and binarizes it with a fixed threshold, outputs binary image data Ic, inputs the multi-valued image data G, binarizes it by a dither method, Either the binary image data Ic or Id is selected for each pixel by the dither binarization unit 12 that outputs the image data Id and the signal Sp representing the result of the image area separation, and is output as the binary image data I. It is constituted by a selector 13.

On the other hand, the encoding means sums up the results of image area separation performed by the binarization means and performs encoding corresponding to a fixed threshold binary image or performs encoding corresponding to a dither image. The counter 25 outputs a selection signal S indicating the selection result, and the binary image data I (Ic or Id) output from the binarization means is input to the counter 25 for TP (typical prediction). A TP unit 21 that outputs signals SLNTP and TPVALUE, an AT unit 22 that outputs ATMOVE representing coordinate information of a pixel having a higher correlation with the pixel of interest based on the selection signal S, and forms a context for encoding including the AT pixel. It is configured by a model template 23 to be performed and an arithmetic coding unit 24 that performs Markov model coding with reference to the model template 23.

(Comparative Example) FIG. 3 is a block diagram showing an example of an image coding apparatus for generating an AT according to Recommendation T.82 for comparison with the present invention. The image encoding apparatus includes a binarizing unit and an encoding unit. The binarizing unit inputs the multi-valued image data G, binarizes the multi-valued image data with a fixed threshold, and outputs a binary image data Ic. A binarizing unit 11, a multi-valued image data G, which is binarized by a dithering method and binarized by a dithering method, and a binarizing unit 12 which outputs a binary image data Id; Selection signal S
And a selector 13 for selecting any one of the binary image data Ic and Id output from the fixed threshold binarization unit 11 and the dither method binarization unit 12 and outputting as the binary image data I. .

On the other hand, the encoding means inputs the binary image data I (Ic or Id) output from the binarization means,
Signals SLNTP, TPVA for TP (typical prediction)
The TP unit 21 that outputs the LUE, obtains statistics of the binary image data I, obtains and outputs an ATMOVE representing the coordinate information of the pixel having a higher correlation with the pixel of interest, and outputs the ATMOVE 22, the encoding unit including the AT pixel It comprises a model template 23 that forms a context, and an arithmetic encoding unit 24 that performs Markov model encoding with reference to the model template 23. As described above, in the image coding apparatus according to Recommendation T.82, the binarizing means and the coding means are completely independent, and the coding means is any binary coding means.
Since it is not known whether or not the binarization has been performed, in order to determine the AT pixel, statistics of the binary image data must be taken to examine the properties of the binary image. It is presumed that the device for this becomes complicated and leads to an increase in cost.

In the image coding apparatuses of the embodiment shown in FIG. 1 and the embodiment shown in FIG. 2, since the coding means is notified of the type of the binary image data,
The optimum AT pixel can be selected accordingly.
For example, if the binarization method for the three-line model and the two-line model is the Bayer dither method, the pixels at the positions shown in FIGS. 8A and 8B are set as AT pixels. , Can greatly improve the compression ratio. In addition, since such a configuration can be easily changed, there is an advantage that the cost is not increased.

FIG. 4 is a block diagram for explaining an example of the decoding means in the JBIG system. The decoding means includes an arithmetic decoding unit 31, a model template 32, and a TP unit (typical prediction unit) 33. As shown in the figure, decoding can be performed only by using the coded data C and ATMOVE, and is not affected by the AT execution method. Therefore, the image data encoded by the image encoding device of the present invention can be decoded by decoding means according to Recommendation T.82.

[0025]

According to the first aspect of the present invention, when encoding is performed, information on a method of binarizing the binary image data to be encoded is notified to the encoding unit, so that the encoding unit performs the binary encoding. No special means for examining the properties of image data is required, and encoding can be performed efficiently at low cost.

Advantageous effect of the second aspect: In addition to the effect of the first aspect, the present invention can be suitably applied to an apparatus having a binarizing means and an encoding means. By transmitting the selection signal of the binary image data in the binarizing means at the time of performing to the encoding means, the encoding means does not need special means for examining the properties of the binary image data, Coding can be performed efficiently at low cost.

Effect of the invention of claim 3: claim 1 or 2
In addition to the effects of the present invention, the selected binary image data
By taking statistics of the ratio of the binarization method and performing encoding based on this statistical data, encoding can be performed efficiently.

Advantageous Effects of the Invention of Claim 4: Claims 1 to 3
In addition to the effects of any one of the aspects of the invention, by providing a selection signal generation unit that generates a selection signal for selecting binary image data to be transmitted to the encoding unit based on the multi-valued image data, Binarization can be performed without concern.

Effect of the invention of claim 5: claims 2 to 4
In addition to the effects of any one of the aspects of the invention, by limiting to a means to which a plurality of binarization methods can be specifically applied, an image coding apparatus that provides an effect is provided.

Effect of the invention of claim 6: claims 1 to 5
In addition to the effects of any one of the aspects of the invention, by limiting the coding method to means to which the coding method can be specifically applied, an image coding apparatus that provides an effect is provided.

Effect of the invention of claim 7: claims 1 to 6
In addition to the effect of any one of the aspects of the invention, a coding condition is specifically given. AT pixels can be quickly determined without the necessity of using a complicated AT device for searching for an AT candidate pixel in the encoding device, so that low-cost and reasonable encoding can be performed.

Effect of the invention of claim 8 In addition to the effect of the invention of claim 7, by providing a more specific configuration of the encoding means, there is provided an image encoding apparatus that provides an effect.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an embodiment of an image encoding device according to the present invention.

FIG. 2 is a block diagram illustrating another embodiment of the image encoding device according to the present invention.

FIG. 3 shows A according to Recommendation T.82 for comparison with the present invention.
It is a block diagram which shows an example of the image coding apparatus which produces T.

FIG. 4 is a block diagram illustrating an example of a decoding unit in the JBIG system.

FIG. 5 is a diagram showing an example of a matrix of thresholds for explaining the Bayer dither method.

FIG. 6 is a diagram showing an example of a dither pattern for explaining a Bayer dither method.

FIG. 7 is a diagram illustrating an example of a context pattern.

FIG. 8 is a diagram for explaining an AT applied to a context.

FIG. 9 is a diagram illustrating an example of selecting a position of an AT pixel applied to a context.

[Explanation of symbols]

11: fixed threshold binarization unit, 12: dither method binarization unit, 1
3 selector, 14 image area separating unit, 21 TP unit, 22
.. AT part, 23 model template, 24 arithmetic coding part, 25 counter, 31 arithmetic decoding part, 32 model template, 33 TP part, C coded data,
G: multi-valued image data, I (Ic, Id): binary image data, S, Sp: selection signal.

Claims (8)

[Claims] An encoding means for selecting a binary image data from a plurality of binary image data and encoding the input binary image data based on a preset encoding method, and controlling the selector to perform a control operation. A signal, which is input to the encoding means together with the selector, wherein the encoding means specifies an encoding condition based on the input selection signal, and performs encoding in accordance with the specified condition. Encoding device. 2. The multi-valued image data to be encoded, which is binarized by binarization means using a plurality of binarization methods, as the plurality of binary image data. 2. The image encoding device according to 1. 3. The coding means includes, as means for designating coding conditions based on the input selection signal, statistics means for obtaining statistics of a selection signal for controlling the selector, and outputting the statistics from the statistics means. 3. The image encoding apparatus according to claim 1, wherein the encoding condition is specified based on statistical data to be performed. 4. The apparatus according to claim 1, wherein said binarizing means includes a selection signal generating means for inputting multi-valued image data to be encoded and generating a selection signal for controlling and operating said selector based on said multi-valued image data. The image encoding device according to claim 1, wherein 5. The image coding apparatus according to claim 2, wherein said plurality of binarization methods are a fixed threshold method and a dither method. 6. The image encoding apparatus according to claim 1, wherein said encoding method is Markov model encoding. 7. The method according to claim 1, wherein a parameter corresponding to the coordinates of an AT pixel is used as the encoding condition.
7. The image coding apparatus according to any one of claims 6 to 6. 8. A typical prediction unit that skips encoding when pixel data of a line to be encoded in the binary image data coincides with pixel data of a line immediately above the line to be encoded. A typical predictor that performs
An AT unit that outputs coordinate information of an AT pixel corresponding to the specified parameter, a model template that forms a context based on the coordinate information output from the AT unit, and a Markov model code with reference to the model template. The image encoding apparatus according to claim 7, further comprising an arithmetic encoding unit that performs the encoding.