JPH10126624A - Picture encoding device and picture decoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve encoding efficiency and to dissolve erroneous substitution by substituting a picture for code information for respective divided small block areas and selecting the picture of the preserved small block area which is most similar. SOLUTION: Inputted picture information is inputted to a picture area division means 201 and it is area-divided into the units of a character and a graphic and they are segmented. A picture recognition means 203 executes a recognition processing on the notice picture which is divided and inputted. A picture substitution means 204 substitutes the picture for code information of a pertinent character code and the like and it is outputted to an encoding information generation means 209. When the picture cannot be substituted for the character code and the like, a similar pictures selection means 205 selects and extracts the picture similar to the notice picture from the picture preserved in an area picture preservation means 202. A similar picture encoding means 207 encodes the notice picture by using the similar picture. When the similar picture does not exist, a prediction/encoding means 208 encodes the notice picture by a regular method.

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 an image decoding apparatus used for compressing the amount of data at the time of transmission and storage in an apparatus which handles images, such as a facsimile apparatus and an image filing apparatus. It concerns the device.

[0002]

2. Description of the Related Art Image coding methods are roughly classified into binary image coding and multi-valued image coding. The former are MH, MR, MMR.
There are coding methods and the like, which are standard facsimile coding methods. Recently, an encoding method using an arithmetic code has attracted attention, and there is an encoding method called a QM coder. This QM coder is JBIG (Joint Bi-l), which is being studied as a standard method by ITU-T (International Telegraph and Telephone Consultative Committee) and ISO (International Organization for Standardization).
evel Image Coding Experts
(Group) method. On the other hand, the latter includes a discrete cosine transform (hereinafter referred to as DC), which is studied as a standard method by ITU-T and ISO.
T) using ADCT (Adaptive Discrete Cosine Transform) and JPEG (Joint Photographic).
There is a so-called “Coding Experts Group” method, which encodes information after converting image information into frequency information.

[0003]

However, the above-mentioned image coding method has already reached a certain limit in terms of coding efficiency (compression rate), and a dramatic improvement in efficiency in the future cannot be expected much.

[0004] In recent years, most documents used in offices and the like have been created by personal computers (PCs) and word processors (word processors), and documents handwritten have almost disappeared. A document created by a PC or a word processor is usually not code information handled by facsimile but code information composed of layout information such as character codes and ruled lines. The advantage of this code information is
The point is that the amount of data is extremely small, and that processing such as editing is simple. For example, in the case of image information such as a facsimile, when characters and layouts are edited, processing must be performed on the image, so that the processing amount and the work area required for the processing become very large. . Due to such a problem, processing such as editing is conventionally performed on image information such as an output document of a facsimile, in other words, editing processing is rarely performed on a received document or the like.

As a method of converting image information into code information used in a PC or word processor, there is OCR (optical character recognition). OCR is a technology that automatically identifies the characters in a scanned document and converts them into character codes that can be handled by a computer. Scanning, copying machines, facsimile machines, etc. generally use OCR to extract characters from images. Cut out,
The extracted character is replaced with a character code.

However, the OCR has a major problem of erroneous replacement, which has not been successfully solved even at present.
The erroneous replacement means that a character is erroneously replaced with a different character. For example, a numeral "1" and an alphabet "l", a symbol "o" and an alphabet "O". In some cases, such erroneous replacement may cause an extreme drop in image quality, which is a major problem in OCR. In the current OCR, the character recognition rate is not 100% due to the influence of noise or the like in an image. Therefore, there is a possibility that the document after the character recognition may be erroneously replaced. Therefore, after all, the whole document after the conversion is often reviewed. This is also a major problem in OCR.

Accordingly, the present invention has been made to solve such a problem, and an image coding apparatus and a picture coding apparatus capable of dramatically improving coding efficiency and solving the problem of erroneous replacement. It is an object to provide the decoding device.

[0008]

In order to achieve the above object, an image coding apparatus according to a first aspect of the present invention divides input image information into small block areas in units of characters and figures. Image area dividing means, area image storing means for storing an image divided by the image area dividing means, image recognizing means for performing image recognition for each small block area divided by the image area dividing means, When the image can be replaced by code information such as a character code by the recognition means, image replacement means for replacing the image of the small block area with the corresponding code information; and code information such as a character code by the image recognition means. If the image cannot be replaced with the image of the small block region of interest from among the previously appeared small block region images stored in the region image storage unit, Similar image selecting means for selectively extracting an image in the lock area, and similar image determination for determining whether the similar image selected and extracted by the similar image selecting means is effective as a similar image of the image of the small block area of interest Means for predicting and encoding the image of the small block area from a similar image when the similar image determining means determines that the image is valid; and the similar image determining means. When it is determined that the image is invalid in the small block area, a prediction encoding unit that performs prediction encoding without using a similar image for the image of the small block area, and a result of the image replacement unit and the two prediction encoding units And code information generating means for generating code information to be output based on the code information.

Further, according to a second aspect of the present invention, in the image encoding apparatus of the first aspect, by setting a high recognition rate of the image recognizing means, erroneous replacement does not occur in the image replacing means. Only the pattern is replaced with the code information.

According to a third aspect of the present invention, in the image coding apparatus according to the first or second aspect, the similar image determining means includes a similar image selected and extracted by the similar image selecting means. It is determined whether or not the image is valid as a similar image based on the matching rate with the image of interest.

[0011] On the other hand, an image decoding apparatus according to claim 4 is
Whether the input code information has been replaced with code information,
Code identification means for identifying whether the image has been encoded using a similar image or not, and a replacement image for creating an original small block image from information replaced with code information Creation means, similarity prediction decoding means for predicting and decoding from a similar image based on a previously reproduced small block area image, and predictive decoding means for performing preliminary decoding independently of the already reproduced small block area image A reproduction area storage unit for storing a reproduction image reproduced by the replacement image generation unit and the two prediction decoding units; and a reproduction image reproduced by the substitution image generation unit and the two prediction decoding units. And a reproduced image combining means for reproducing the image.

Further, according to a fifth aspect of the present invention, in the image decoding apparatus according to the fourth aspect, the reproduced image includes:
Identification information for identifying an image reproduced by the replacement image generating means and an image reproduced by the other two predictive decoding means is added.

According to a sixth aspect of the present invention, in the image decoding apparatus according to the fifth aspect, a different color is assigned to one image from the other as the identification information.

According to a seventh aspect of the present invention, in the image decoding apparatus of the fifth aspect, an arbitrary pattern is added to the background of one image as the identification information.

According to an eighth aspect of the present invention, in the image decoding apparatus according to the fifth aspect, an arbitrary pattern such as an underline is added to one image as the identification information.

[0016]

Embodiments of the present invention will be described below in detail with reference to the drawings.

Here, as an embodiment of each invention of the present application, a case where the present invention is applied to a facsimile machine will be described as an example.

FIG. 1 shows a block diagram of a facsimile apparatus when each invention of the present application is specifically implemented. First, on the transmitting side, an image reading unit 101 reads a document using a CCD image sensor or the like, and then performs processing for converting transmission data into appropriate data in an image processing unit 102, and finally performs encoding in an encoding unit 103. And sends the generated code to the transmission path. On the other hand, when the image is reproduced on the receiving side, the encoded data is
04, the image processing unit 105 performs image processing suitable for the output device, and outputs the image data to an image output unit 106 such as a plotter to obtain a hard copy.

Examples of processing performed by the image processing units 102 and 105 include resolution conversion for a binary image,
There is a size conversion and the like, and in a multi-valued image including a color, a color (color component) conversion, a resolution conversion, a size conversion and the like can be mentioned. Examples of the encoding method used in the encoding unit 103 and the decoding unit 104 include MH, MR, MMR methods conventionally used in facsimile and JBIG method using arithmetic codes for binary images. Yes, adaptive discrete cosine transform (ADCT) for multi-valued images
There is a JPEG system that uses JPEG. 2 and 3 show the internal block diagrams of the encoding unit (image encoding device) 103 and the decoding unit (image decoding device) 104, respectively.

First, an example of the encoding device according to the present invention will be described with reference to the block diagram of FIG.

An image area dividing means 201 for dividing input image information into small block areas in units of characters and figures, an area image storing means 202 for storing divided images, a divided small block Image recognition means 203 for performing image recognition for each area; image replacement means 204 for replacing the image of the small block area with the corresponding code information; Similar image selecting means 205 for selecting and extracting the image of the most similar small block area, similar image determining means 20 for determining whether or not the selected and extracted similar image is effective as a similar image of the image of the small block area of interest
6, similar prediction encoding means 207 for predicting and encoding an image of a small block area from a similar image, predictive encoding means 208 for predictively encoding an image of a small block area without using a similar image, It comprises a code information creating means 209 for creating a code based on the results of the replacing means 204 and the two predictive encoding means 207 and 208. Note that each of the above-mentioned units uses an FCU (facsimile control unit) or a memory constituting a facsimile apparatus in which the encoding apparatus is used, or a processor or a memory provided exclusively for the encoding apparatus and a decoding apparatus described later. This is realized by executing the processing shown by the flowchart in FIG.

Next, the operation of the encoding apparatus having the above configuration will be described with reference to the flowchart shown in FIG.

First, the image information input for encoding is input to the image area dividing means 201, where the image information is divided into units such as characters and figures and cut out (process 401). The divided image information is input to the area image storage unit 202, the image recognition unit 203, and the similar image selection unit 205, respectively.

The area image storage means 202 stores the divided area image (process 402). Image recognition means 2
In 03, a recognition process is performed on the divided and input target image, and it is determined whether or not the image can be replaced with code information such as a character code (decision 403). As an example of the recognition processing, there is a method of determining whether or not an image is a character and, if the image is a character, recognizing the type of the character from its skeleton information.

If the image recognizing means 203 determines that the image can be replaced with code information such as a character code, the image is replaced by the corresponding code information by the image replacing means 204 and the code information is replaced with the code information. Creation means 20
9 (YES in decision 403 → process 404 → process 409).

On the other hand, if the image recognizing means 203 determines that the image cannot be replaced with a character code or the like, then the similar image selecting means 205 stores an image similar to the target image in the area image storing means 202. (Step 403: NO)
5). As an example of a method of selecting a similar image, there is a method of giving priority to an image having the same size, giving priority to an image having a similar number of black pixels, and selecting an image having a similar shape.

When a similar image is selectively extracted, the similar image determining means 206 determines whether the selected image is valid as a similar image (determination 406).
There are various determination methods, but if the matching ratio is determined to be valid when the matching ratio is equal to or greater than a predetermined value based on the matching ratio between the selected and extracted similar image and the target image, the configuration is simplified. become.

If it is determined that the selected similar image is effective as a similar image of the image of interest, the similar image predicting means 207 performs prediction using the similar image and encodes the image of interest (decision 406). YES → Process 40
7). On the other hand, when it is determined that the selected similar image is invalid as a similar image of the image of interest, the prediction encoding unit 208 performs prediction using a normal method that does not use similar images and encodes the image of interest ( NO in decision 406 → process 408).

As an example of predictive coding, there is an arithmetic coding method as shown in FIG. In the arithmetic coding method, in the coding, the input image data is converted into a prediction information creation unit 50.
The prediction information is obtained by performing Markov separation of an information source on the basis of the situation of the pixel to be coded in a portion referred to as “1” (also referred to as a template) and surrounding pixels, and an arithmetic coding unit is performed based on the prediction information. At 502, encoding is performed while dynamically evaluating prediction information. In decoding, decoding is performed by performing Markov separation of an information source according to the situation of pixels around the pixel to be decoded with the pixels already decoded by the same prediction information creating unit 504 as the encoding data at the time of encoding. The prediction side information of the data to be generated is created, and the arithmetic decoding section 503 performs decoding while dynamically evaluating the prediction side information based on the prediction information.

The arithmetic coding method generally has higher coding efficiency than the conventional run-length coding method (MH, MR). The encoding method is based on the corresponding section on the number line of [0, 1] (binary decimal [0.0... 0, 0.1
... 1]) are divided into unequal lengths in accordance with the occurrence probabilities of the respective symbols, the target symbol sequence is assigned to the corresponding subsections, and the section obtained by repeating the division recursively. The coordinates of the points contained within are represented by binary decimal numbers that can be distinguished from at least other sections, and are used as codes.

The concept of arithmetic coding will be described with reference to FIG. 6 using the symbol sequence '0100' as an example. First, when coding the first symbol, the ratio of the occurrence probabilities of the symbols '0' and '1' in all sections Is divided into A (0) and A (1) according to
Section A (0) is selected by the occurrence of '0'. next,
When encoding the second symbol, A (0) is further divided by the occurrence probability ratio of both symbols in that state,
A (01) is selected as a section corresponding to the generated symbol sequence. Coding proceeds by repeating such division and selection processing. On the other hand, in the decoding, a process completely opposite to that of the encoding is performed, and the symbol is reproduced based on the binary decimal number indicated by the code. The important thing here is that
This is the width of a number line when encoding a symbol. If the width of the number line does not match between the start of encoding and the start of decoding, the symbol cannot be accurately reproduced. Normally, the width of this number line is set to 1 on the encoding side and the decoding side.

With the above processing, the input image is encoded, and the results of the image replacing means 204 and the two predictive encoding means 207 and 208 (processing 404, processing 407,
Based on the result of the process 408), code information is created by the code information creating means 209 (process 409), and this is performed until the input image is completed (NO in decision 410 → loop of process 401).

As described above, the information of characters and the like is much smaller in the amount of data and the versatility of code information such as character codes than in the case of image data. Is used to replace the image with code information, and the non-replaceable part performs efficient predictive coding by the similar predictive coding means 207 and the predictive coding means 208. The encoding becomes possible.

By preventing erroneous replacement in the replacement by the image replacement means 204, the information amount is increased, but the reliability can be increased. That is, when the recognition rate is increased in the image recognition, the erroneous replacement is reduced, but the number of unrecognizable images is increased. Conversely, when the recognition rate is lowered, the amount of information is reduced, but erroneous replacement is increased. Misplacement is a major problem for OCR, and in order to solve it, processing must be performed so that misplacement does not occur at all. As described above, there are patterns with many mis-replacements, such as the numeral “1” and the letter “l”, and the symbol “O” and the letter “O”, but there are also patterns without mis-replacement. Present to some extent. Therefore, this problem can be solved by replacing only such a pattern with the code information.

As described above, since the image replacing means 204 performs replacement only on a pattern that does not cause erroneous replacement at all, no erroneous replacement occurs, so that there is no deterioration in image quality and the image can be replaced with code information. The part is replaced with code information, and the part that is not replaced is subjected to efficient predictive coding, so that coding can be performed with extremely high efficiency.

Next, an example of the decoding apparatus according to the present invention will be described with reference to the block diagram of FIG.

The structure is a code identification for identifying whether the input code information is replaced code information, a code coded using a similar image, or a code coded without using a similar image. Means 301, replacement image creating means 3 for creating an original small block image from information replaced with code information
02, similar prediction decoding means 303 for performing prediction and decoding based on the image of the already reproduced small block area, predictive decoding means 304 for performing preliminary decoding regardless of the image of the already reproduced small block area, Reproduction area storage means 305 for storing the reproduced images reproduced by the creating means 302 and the two predictive decoding means 303 and 304, and reproduced image combining means 3 for combining the reproduced images individually and reproducing the entire image
06. Note that each of the above-described units also uses an FCU or a memory constituting a facsimile apparatus in which the decoding apparatus is used, or a processor or a memory provided exclusively for the decoding apparatus and the encoding apparatus, as shown in the flowchart of FIG. It is realized by executing the processing shown. In the present embodiment, the code identifying means 301 is provided independently, but each of the replacement image creating means 302, the similar predictive decoding means 303, and the predictive decoding means 304 may be provided with a dedicated code identifying means.

Next, the operation of the decoding apparatus having the above configuration will be described with reference to the flowchart shown in FIG.

First, it is determined whether the code of interest is the replaced code information, the code coded using the similar image, or the code coded without using the similar image from the input codes. Identification is performed by the identification means 301 (determination 701). If it is the replaced code information, the replaced image creating means 302 reproduces the original image from the code information and outputs it to the reproduction area storage means 305 (processing 702 → processing 705).

In the case of a code that has been encoded using a similar image, the similar predictive decoding unit 303 replaces the image of a similar small block region that has previously appeared and is stored in the reproduction region storage unit 305. The image is reproduced using the image data and output to the reproduction area storage unit 305 (processing 703 → processing 7).
05).

Further, in the case of a code encoded without using a similar image, the predictive decoding unit 304 reproduces the image without using a similar image and outputs it to the reproduction area storage unit 305 (step 704 → Process 705).

The reproduction area storage means 305 stores the reproduced image for each small block area and outputs the image to the reproduction image combining means 306 in order to create the entire image (processing 705 → processing 706). This is repeated until the input code information is completed (NO in decision 707).
Upon completion of the loop (determination 701 loop), the reproduced image combining means 306 finally combines the reproduced images to create the entire image and outputs it.

As described above, in the case of the replaced code information, the original image is reproduced from the code information, and in the case of the code encoded using the similar image, the image of the similar small block area is used. If the code is encoded without using a similar image, the image is reproduced without using a similar image, and each reproduced image is finally combined to create an entire image Therefore, it is possible to reproduce an image without error from the highly efficient encoded code created by the encoding device described above.

It is to be noted that, when composing the entire image, the user can easily identify both the code-replaced image and the other image by reproducing the image with identification information so that the image can be distinguished from the other image. Become like As the identification information, a color different from the other image is added to one image within a range that does not impair the color of the original image, or a background pattern such as a thin mesh is added to the background of one image, Alternatively, a unique character pattern such as an underline may be added to one of the images. This makes it possible to easily identify the image replaced with the code information and the image not replaced with a color, a background pattern, an underlined character pattern, or the like.

Finally, regarding the format of the code information used in the present invention, as one example, since there are three types of code processing methods, three states can be identified, and in addition, the position of each image can be identified. A method of adding information is conceivable. There are various methods of expressing three types of states and methods of expressing position information. For example, three types of expressions use two bits, A = “00”, B = “10”, and C = “11”.
It is conceivable that the position information is represented by an offset position or the like from the immediately preceding encoded image.

[0046]

As described above, according to the first aspect of the present invention, at the time of encoding image information, the image information is divided into small block areas such as characters and figures and divided. Image recognition is performed for each small block area. If the image can be replaced with code information such as a character code, the image of the target block area is replaced with the corresponding code information.
Select the image of the small block area most similar to the image of the small block area of interest from among the images of the small block area that appeared before, and enable the selected similar image as a similar image of the image of the small block area of interest After determining whether or not the image is valid, the image of the small block area is predicted from a similar image and coding is performed. If the image is invalid, the image of the small block area is not used and the similar image is not used. The parts that can be replaced with code information are replaced with code information, and the parts that are not replaced are subjected to efficient predictive coding, so that coding can be performed with extremely high efficiency. There is a possible effect.

Further, according to the second aspect of the present invention, only the pattern which does not cause the erroneous replacement is replaced, so that the erroneous replacement does not occur, so that the image quality is not degraded, and the image is replaced with the code information. Are replaced with code information, and the parts that are not replaced are subjected to efficient predictive coding, so that coding can be performed with extremely high efficiency.

According to the third aspect of the present invention, in the above, it is determined whether or not the selected similar image is valid as a similar image of the image of interest by determining a coincidence rate between the similar image and the image of interest. Since the determination is made based on the original, there is an advantage that the configuration is simplified in addition to the above effects.

On the other hand, according to the invention of claim 4, when the code coded as described above is decoded, if the code information is replaced code information, the original image is reproduced from the code information and the similar image is reproduced. In the case of a code that has been encoded using, the image is reproduced using an image of a similar small block area that appeared earlier, and in the case of a code that has been encoded without using a similar image, the similar image is reproduced. Since the images are reproduced without using them and each reproduced image is finally combined to form an entire image, it is possible to reproduce an image without error from a code coded with high efficiency as described above. Will be possible.

Further, according to the fifth aspect of the present invention, at the time of the decoding, the image is reproduced by attaching identification information capable of discriminating between the image replaced with the code information and the image not replaced. , The user can easily identify both.

According to the sixth aspect of the present invention, since one image is given a different color from the other as the identification information, the image replaced with the code information is replaced with the image not replaced. Can be easily identified by color.

According to the seventh aspect of the present invention, an arbitrary pattern is added to the background of one image as the identification information. Therefore, the image is not replaced with the image replaced with the code information. The image and the background pattern can be easily identified.

According to the eighth aspect of the invention, an arbitrary pattern such as an underline is added to one of the images as the identification information, so that the image is not replaced with the image replaced with the code information. The image can be easily identified by the added pattern such as the underline.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a facsimile apparatus when each invention of the present application is specifically implemented.

FIG. 2 is a block diagram showing a configuration of an image encoding device according to the present invention.

FIG. 3 is a block diagram showing a configuration of an image decoding apparatus according to the present invention.

FIG. 4 is a flowchart showing the operation of the image encoding device shown in FIG. 2;

FIG. 5 is a block diagram showing a configuration of an arithmetic coding system.

FIG. 6 is an explanatory diagram showing the concept of arithmetic coding.

FIG. 7 is a flowchart showing the operation of the image decoding apparatus shown in FIG. 3;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 101 Image reading part 102, 105 Image processing part 103 Encoding part 104 Decoding part 106 Image output part 201 Image area division means 202 Area image storage means 203 Image recognition means 204 Image replacement means 205 Similar image selection means 206 Similar image determination means 207 Similar prediction encoding means 308 Prediction encoding means 209 Code information creation means 301 Code identification means 302 Replacement image creation means 303 Similar prediction decoding means 304 Prediction decoding means 305 Playback area storage means 306 Playback image combining means 501, 504 Prediction information creation section 502 Arithmetic encoder 503 Arithmetic decoder

Claims (8)

[Claims] 1. An image area dividing means for dividing input image information into small block areas in units of characters, figures, and the like; an area image storing means for storing an image divided by the image area dividing means; An image recognition unit that performs image recognition for each of the small block regions divided by the image region division unit; and if the image recognition unit can replace the image with code information such as a character code, the image of the small block region is applied. Image replacement means for replacing the code information to be replaced with an image of a previously appeared small block area stored in the area image storage means when the image cannot be replaced with code information such as a character code by the image recognition means. A similar image selecting means for selectively extracting an image of the small block area most similar to the image of the small block area of interest from among the selected small block areas; A similar image determination unit that determines whether the output similar image is valid as a similar image of the image of the small block area of interest; and if the similar image determination unit determines that the similar image is valid, the small image A similar prediction encoding unit that predicts and encodes the image of the block region from a similar image, and, if the similar image determination unit determines that the image is invalid, the similar region is similar to the image of the small block region. A predictive encoding unit that performs predictive encoding without using an image; and code information creating unit that creates code information to be output based on the results of the image replacing unit and the two predictive encoding units. Image encoding device. 2. The method according to claim 1, wherein the recognition rate of the image recognition means is set high so that only the pattern which does not cause erroneous replacement by the image replacement means is replaced with code information. Image coding device. 3. The similar image determination unit determines whether or not the similar image is valid as a similar image based on a coincidence rate between the similar image selected and extracted by the similar image selection unit and a target image. The image encoding device according to claim 1 or 2. 4. A method according to claim 1, wherein the input code information is replaced with code information, or encoded using a similar image.
Code identification means for identifying whether or not the image has been encoded without using a similar image; replacement image creation means for creating an original small block image from information replaced with code information; Similar prediction decoding means for predicting and decoding from a similar image based on the same, predictive decoding means for performing preliminary decoding independently of an image of a small block area which has already been reproduced, Means for storing a reproduced image reproduced by the means, and reproduced image combining means for combining the reproduced images reproduced by the replacement image creating means and the two predictive decoding means to reproduce the entire image. An image decoding device characterized by the above-mentioned. 5. The reproduction image according to claim 4, wherein identification information for identifying an image reproduced by said replacement image generating means and an image reproduced by the other two predictive decoding means is added. The image decoding device according to any one of the preceding claims. 6. The image decoding apparatus according to claim 5, wherein one color of the one image is different from that of the other image as the identification information. 7. An arbitrary pattern is added to the background of one of the images as the identification information.
The image decoding device according to any one of the preceding claims. 8. The image decoding apparatus according to claim 5, wherein an arbitrary pattern such as an underline is added to one of the images as the identification information.