JP2010027051A - Image compression apparatus and image compression method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image compression apparatus and an image compression method attaining high-speed image compression processing by a dictionary system. <P>SOLUTION: The image compression apparatus has an analysis section 12 analyzing an input image and outputting object layout information and page attribute information; a segmenting section 13 segmenting the image based on the object layout information and outputting a component image ; a dictionary section for selecting an image dictionary corresponding to the page attribute information out of existing image dictionaries, comparing the selected image dictionary with the component image to determine the corresponding image dictionary, and outputting the determined image dictionary and a dictionary index; and an encoding section 15 encoding the image dictionary and the dictionary index. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image compression apparatus and an image compression method for extracting object arrangement information and page attribute information from image information and compressing an image based on the extracted information.

  Conventionally, in order to efficiently compress a binary image, JPEG2 is known as a standardized technique for converting an image bitmap into a dictionary and using an index (corresponding dictionary number and its arrangement) and a dictionary as compressed data. . In JPEG2, in order to increase the compression ratio, for example, a configuration is shown in which encoding is performed by another pair of encoding means based on shape information (including position) and halftone dot information, which form a braided dot character, for example. .

  Since image information has a large capacity, it is generally compressed and stored / used. Usually, character images and the like are binarized to perform lossless compression, and photographs and the like are subjected to multi-value irreversible compression.

  However, a method of irreversibly compressing a binary character image has been proposed and standardized as ISO / IEC14492 (JPEG2).

  Characteristic of this method is that many of the same characters appear in the document image etc., so by making a bitmap image of the character into a dictionary, the same character is not a bitmap, but a dictionary index and position information on the image This is to reduce the amount of bitmap to be compressed and compress it. For such compression technology, for example, the following patent documents are known.

  Patent Document 1 (Japanese Patent Application Laid-Open No. 2006-254329) discloses not only simple line-drawing characters but also shape information and pattern information constituting the characters so as to efficiently compress characters represented by patterns such as halftones. A method of efficiently compressing characters and the like represented by patterns by separating and compressing them is disclosed.

  Patent Document 2 (Japanese Patent Laid-Open No. 2007-174008) discloses an example in which dictionary-based compression is used for character area compression in a system that identifies characters and photo areas of an image and applies compression suitable for each. A method of high-speed compression is disclosed by excluding information that can be regarded as non-text from character area data input before dictionary compression from a lexicographic compression target.

  Patent Document 3 (Japanese Patent Laid-Open No. 2005-301663) is a technique for setting a bitmap dictionary unit based on a character unit acquired by OCR (Optical Character Recognition) and preventing an image from being partized more than necessary. Disclosure.

  However, in the prior arts of these patent documents 1 to 3, the page image attribute information of documents having various document formats and font types is not considered in the matching process of the component images, and an efficient image dictionary search / There is a problem of not registering.

  An object of the present invention is to provide an image compression apparatus and an image compression method capable of speeding up image compression processing using a dictionary method.

One embodiment to solve the problem is:
An analysis unit that analyzes the input image and outputs object placement information and page attribute information;
Based on the object arrangement information, the image is converted into a component, and a component converting unit that outputs a component image;
An image dictionary corresponding to the page attribute information from the analysis unit is selected from among existing image dictionaries, and the corresponding image dictionary is determined by comparing the selected image dictionary with the component image from the componentizing unit. A dictionary forming unit for outputting the determined image dictionary and a dictionary index of the component image;
An encoding unit for encoding the selected image dictionary and the dictionary index;
An image compression apparatus comprising:

  In the matching process between the component image and the image dictionary when performing the image compression process, the image dictionary corresponding to the page attribute information (for example, document, presentation, map, etc.) of the image to be compressed is given priority and the image dictionary not corresponding is excluded. Etc. to make the matching process more efficient. Thereby, the image compression processing can be speeded up.

1 is a block diagram showing an example of the configuration of an image compression apparatus according to a first embodiment of the present invention. The block diagram which shows an example of a structure of the layout analysis part of the said image compression apparatus. Explanatory drawing explaining an example of operation | movement of the layout analysis part of the said image compression apparatus. The block diagram which shows an example of a structure of the page attribute determination part of the said image compression apparatus. Explanatory drawing of an example of the page determination data of the said image compression apparatus. The block diagram which shows an example of a structure of the image componentization part of the said image compression apparatus. Explanatory drawing explaining an example of operation | movement of the image componentization part of the said image compression apparatus. The block diagram which shows an example of a structure of the image dictionary part of the said image compression apparatus. Explanatory drawing explaining an example of operation | movement of the image dictionary part of the said image compression apparatus. The block diagram which shows the structure of the image compression apparatus which is a modification of 1st Embodiment. The block diagram which shows an example of a structure of the layout analysis part of the said image compression apparatus. The block diagram which shows an example of a structure of the page attribute determination part of the said image compression apparatus. Explanatory drawing which shows an example of the inclination calculation auxiliary | assistance figure for the said image compression apparatus. Explanatory drawing of an example of the page determination data of the said image compression apparatus. The block diagram which shows an example of a structure of the image dictionary part of the said image compression apparatus. The block diagram which shows an example of a structure of the image compression apparatus of 2nd Embodiment of this invention. The block diagram which shows an example of a structure of the character processing part of the said image compression apparatus. The block diagram which shows an example of a structure of the image compression apparatus of 3rd Embodiment of this invention. The block diagram which shows an example of a structure of the image dictionary part of the said image compression apparatus. Explanatory drawing explaining the search order switching effect of the dictionary of the said image compression apparatus. The block diagram which shows an example of a structure of the image dictionary part of the said image compression apparatus. Explanatory drawing explaining the function of the bitmap count part of the said image compression apparatus.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  The gist of the present invention is that page attribute information is extracted from an input image, the image is converted into a part into a plurality of image dictionaries, and one is selected from a plurality of image dictionaries based on the extracted page attribute information Thus, the present invention provides an image compression apparatus and an image compression method that realizes quick compression processing by making dictionary matching processing efficient.

  By adopting such a configuration, a dictionary adapted to attributes is generated, so that a dictionary can be generated at a high speed and a compression ratio can be high, and a compressed file of an index method can be generated.

  In addition to the layout analysis unit, the dictionary switching information can be switched using user instruction means such as a document mode. Also, the switching method can be switched not for each page but for each object in the page.

  Further, this switching (selection) of the dictionary can be performed for each page, for each page information of a document type such as a map, for each document direction information of a page unit such as a skew angle and top and bottom, and for each character information. Each can generate a high-speed and high-compression dictionary compressed file.

(First embodiment)
FIG. 1 is a block diagram showing an example of the configuration of the image compression apparatus according to the first embodiment of the present invention. An image compression apparatus 1 according to an embodiment of the present invention includes a control unit 10 that controls the overall operation, a scanner 11 that reads an original image, a layout analysis unit 12 that is supplied with an image signal m1 from the scanner 11, and a scanner. 11 includes an image component converting unit 13 to which the image signal m1 is supplied from 11 and the object arrangement information m2 is supplied from the layout analyzing unit 12. Further, the image compression apparatus 1 receives the page attribute information m3 from the layout analysis unit 12, the image dictionary unit 14 that receives the component image m4 from the image component unit 13, and the image dictionary signal m5 and the dictionary index from the image dictionary unit 14. An image encoding unit 15 to which m6 is supplied, a storage unit 16 to which code data m7 is supplied from the image encoding unit 15, and an interface unit 17 provided at the subsequent stage of the storage unit 16 are provided.

  In the image compression apparatus 1 having such a configuration, the operation will be described below. The image signal m1 input from the scanner 11 is supplied to the layout analysis unit 12, and the layout analysis unit 12 receives the supplied image signal m1. By processing with a known layout analysis technique, object arrangement information m2 and page attribute information m3 are output. The image component converting unit 13 outputs a component image m4 using the supplied image signal m1 and object arrangement information m2.

  The image dictionary unit 14 outputs an image dictionary m5 and a dictionary index m6 using the known bitmap dictionary conversion technique and the page attribute information m3 for the component image m4. The image encoding unit 15 receives these signals, generates code data m7, and stores it in the storage unit 16.

  FIG. 2 is a block diagram illustrating an example of the configuration of the layout analysis unit of the image compression apparatus. As shown in FIG. 2, the layout analysis unit 12 includes a reduction processing unit 21 that receives the image signal m <b> 1, a connected pixel search unit 22 to which the reduced image signal m <b> 11 is supplied from the reduction processing unit 21, and a connection pixel search unit 22. The region coordinate conversion unit 23 that receives the region information m12 that is the upper left coordinate, the lower left coordinate, the upper right coordinate, and the lower right coordinate of the region, the page attribute determination unit 24 that similarly receives the region information m12, and the page attribute determination unit 24 are connected. A table 25 is provided.

  In the layout analysis unit 12 having such a configuration, upon receiving the image signal m1, the reduction processing unit 21 reduces the image at a predetermined reduction rate and outputs a reduced image signal m11. The connected pixel search unit 22 searches in eight directions using a known chain algorithm, and outputs area information m12 that is the upper left coordinates, lower left coordinates, upper right coordinates, and lower right coordinates of the areas where the pixels are connected. Here, since the coordinates of the area information m12 are in the coordinate system reduced by the reduction processing unit 21, the area coordinate conversion unit 23 converts the coordinate system to the same coordinate system as the image signal m1 and outputs it as object arrangement information m2. On the other hand, the page attribute determination unit 24 totals the area information m12 in the page, compares it with the page determination data m13 read from the table 25, and outputs page attribute information m3.

  Here, an operation example of the layout analysis unit 12 excluding the operation of the page attribute determination unit 24 is shown in FIG. In the reduced image m11 obtained by reducing the image signal m1, it can be seen that characters and the like are connected pixels. When a connected region is calculated for this image using a known chain algorithm, a plurality of regions for each cluster as indicated by the dotted line in the region information m12 are calculated. Then, the area coordinate conversion unit 23 receives the area information m12 and calculates object arrangement information m2 in which the coordinate system is returned to before reduction.

  Next, FIG. 4 is a block diagram illustrating an example of the configuration of the page attribute determination unit of the image compression apparatus. The page attribute determination unit 24 includes a region inclination calculation unit 31 to which the region information m12 is supplied, a region distance calculation unit 32 to which the region information m12 is supplied, and a region distance to which the inclination information m21 is supplied from the region inclination calculation unit 31. It has the judgment part 33 to which distance information m22 is supplied from the calculation part 32.

  In the page attribute determination unit 24 having such a configuration, the region inclination calculation unit 31 uses the region information m12 to determine the inclination for each region = MAX (| upper left X coordinate−lower left X coordinate |, | upper left Y coordinate− The upper right Y coordinate |) is calculated and obtained, and the average of the inclinations of all the areas in the page is calculated as the inclination information m21.

Further, the area distance calculation unit 32 obtains the center coordinates of each area, obtains an area pair having the nearest center coordinates for each center coordinate (P1 and P2 below are obtained,
Distance for each area = MIN (| X coordinate of P1−X coordinate of P2 |, Y coordinate of | P1−Y coordinate of P2 |)
Is obtained as the distance information m22.

  The determination unit 33 receives the inclination information m21 and the distance information m22, and reads page determination data m13 as shown in FIG. The determination unit 33 determines the document type based on the result of comparing the inclination information m21 and the distance information m22 with a predetermined threshold value and the page determination data m13, and outputs it as page attribute information m3.

  Next, FIG. 6 is a block diagram illustrating an example of the configuration of the image component converting unit of the image compression apparatus. As shown in FIG. 6, the image componentization unit 13 receives the vertical direction pixel count unit 41, the horizontal direction pixel count unit 42, and the vertical direction pixel count unit 41 to which the image signal m <b> 1 and the object arrangement information m <b> 2 are respectively supplied. From the comparator 43 to which the vertical projection m31 and the horizontal projection m32 from the horizontal pixel count unit 42 are supplied, and from the vertical projection m31 and the horizontal pixel count unit 42 from the vertical pixel count unit 41. A horizontal projection m32 and a selection unit 44 to which the control signal m33 from the comparator 43 is supplied, and a pixel division unit 45 to which the selection signal m34 and the image signal m1 from the selection unit 44 are supplied.

  In the image componentizing unit 13 having such a configuration, the vertical pixel counting unit 41 divides the image signal m1 by the coordinate unit of the object arrangement information m2, and the projection of the pixel count on the vertical axis is projected in the vertical direction. Output as m31. The horizontal pixel count unit 42 also divides the image signal m1 by the coordinate unit of the object arrangement information m2, and outputs the projection of the pixel count on the horizontal axis to the comparator 43 as the horizontal projection m32. The comparator 43 outputs to the selection unit 44 a control signal m33 that allows the selection unit 44 to select the one with the larger variance of the projection values. The selection unit 44 supplies the pixel division unit 45 with a selection signal m34 for selecting the one with the larger variance of the projection values. The pixel dividing unit 45 divides the image signal m1 using the selected projection value and outputs a component image m4.

  These operations will be described with reference to FIG. FIG. 7 is an explanatory diagram illustrating an example of the operation of the image component converting unit of the image compression apparatus. The comparator 43 compares the projection m31 in the vertical direction and the projection m32 in the horizontal direction, and performs a horizontal projection with a large variance. The pixel dividing unit 45 performs threshold processing on the projection to calculate horizontal division coordinates (dotted line), and outputs the result as a component image in units indicated by ◯ in FIG.

  Next, FIG. 8 is a block diagram showing an example of the configuration of the image dictionary unit of the image compression apparatus. The image dictionary unit 14 includes a document dictionary 53, a presentation dictionary 52, a map dictionary 51, and a non-document dictionary 50 corresponding to the page attribute information m3 shown in FIG. 5, and is further supplied with page attribute information m3. A selection unit 55 that selects each dictionary and a dictionary selected by the selection unit 55 are connected, and a matching unit 54 that supplies a component image m4 and outputs an image dictionary signal m5 and a dictionary index m6 is provided.

  In the image dictionary unit 14 having such a configuration, the selection unit 55 selects and outputs an appropriate dictionary among the document dictionary 53, the presentation dictionary 52, the map dictionary 51, and the non-document dictionary 50 according to the page attribute information m3. The matching unit 54 determines whether or not the component image m4 exists in the dictionary selected by the selection unit 55 by combining the known position shift and matching processing, and outputs the dictionary index m6 if it exists. The dictionary index m6 is the corresponding dictionary name, dictionary index, and position information of the component image on the image.

  If there is no corresponding bitmap pattern in the dictionary, the matching unit 54 registers the component image m4 in the currently selected dictionary and assigns an index. The matching unit 54 outputs each dictionary data as the image dictionary m5 when the above matching processing is completed for all the documents input from the scanner 11. Then, the image encoding unit 15 compresses the image dictionary m5 with a known image compression technique (for example, run length), and outputs it as code data m7 together with the dictionary index m6.

  Next, the effect of the characteristic dictionary according to the present invention will be described with reference to FIG. In general, when a plurality of documents are scanned, there may be a mixture of documents of different document types such as a general document, a document for presentation as shown in FIG. 9, and a map shown in FIG. In these documents, there is a high possibility that the same font design and size are used for each document type, but there is a high possibility that the design and size will be different for different document types.

  Generally, these dictionary data are collectively registered in one dictionary, and search and matching are performed. However, in the embodiment according to the present invention described above, dictionary data is dictionaryd as a plurality of dictionaries for each document type, and when using the dictionary, an appropriate dictionary is selected and switched according to the page attribute information, and search and matching are performed. I do. Thereby, since an inappropriate dictionary that does not correspond to the page attribute information is not searched and matched, dictionary matching can be performed efficiently and at high speed. That is, when matching document images, efficient and high-speed dictionary matching can be performed by selecting a document dictionary and not using a presentation dictionary, a map dictionary, or a non-document dictionary.

  If the matching unit 54 loosens the matching accuracy, a dictionary bitmap compressed file is created irreversibly. However, it is difficult to control image quality and matching accuracy with a system that manages various font designs in a single dictionary. . However, if the dictionary is switched for each document type as in the method of the present invention, the probability of matching within the same font design group is high, so the matching accuracy can be relaxed while suppressing degradation in image quality. A dictionary bitmap compressed file having a high compression rate can be provided.

  Note that the layout analysis method, page information calculated therefrom, dictionary formation, compression method, and dictionary switching method are not limited to the above-described embodiments, and the scope of implementation of the present invention is within the scope of those skilled in the art. Is in the range that can be expected.

(Modification of the first embodiment)
Next, a modification of the first embodiment of the present invention will be described in detail with reference to the drawings. FIG. 10 is a block diagram illustrating a configuration of an image compression apparatus that is a modification of the first embodiment.

  The modification of the first embodiment is an image compression apparatus and an image compression method for determining page attribute information m3 from only the region inclination (without considering the region distance), and generating and selecting a dictionary based on the determined page attribute information m3. I will provide a. In the modified example of the first embodiment, the same processing unit as that of the first embodiment is basically assigned the same number, and the layout analysis unit 12 ′, its page attribute information m3 ′, and a dictionary based on it. Since the image dictionary conversion unit 14 ′ to be performed is the same except that it is slightly different, only the changed block will be described.

  The configuration of the layout analysis unit 12 ′ is as shown in FIG. 11. The difference from the first embodiment is the configuration of the table 25 ′, page determination data m13, and page attribute determination unit 24 ′ of the page attribute determination unit 24 ′. The generated page attribute information m3 ′ is different.

  The configuration of the page attribute determination unit 24 'is shown in FIG. The structural difference between the page attribute determination unit 24 ′ according to the modification of the first embodiment and the page attribute determination unit 24 according to the first embodiment is that the page attribute determination unit 24 ′ according to the modification according to the first embodiment has a region. Without using the distance calculation unit 32, only the region inclination calculation unit 31 ′ outputs the inclination angle m21 ′ and the inclination variance m22 ′.

  The inclination angle m21 'defines an inclination as shown in FIG. 13, for example, and is simply calculated by the following equation.

if ((upper left Y coordinate−upper right Y coordinate)> 0) {
θ = atan ((lower right Y coordinate−lower left Y coordinate) / (lower right X coordinate−lower left X coordinate))
}
else {
θ = atan ((upper right Y coordinate−upper left Y coordinate) / (upper right X coordinate−upper left X coordinate))
}
The reduction processing unit 21 outputs the average value of these θ as the inclination angle m21 ′, and outputs the same variance as the inclination variance m22 ′.

  FIG. 14 is an explanatory diagram of an example of page determination data of the image compression apparatus. As in the first embodiment, the determination unit 33 uses the page determination data m13 illustrated in FIG. 14 as an angle category based on the magnitude of dispersion obtained by comparing the range of the inclination angle m21 ′ and the inclination dispersion m22 ′ with a threshold value. The page attribute information m3 ′ is output after being classified into four categories of standard, plus angle, minus angle, and special. This is because, in the case of a map document as exemplified in the first embodiment, there is a category that has a high possibility that the region has an angle but is not an angle due to skew.

  FIG. 15 is a block diagram illustrating an example of the configuration of the image dictionary unit. The image dictionary conversion unit 14 ′ according to the modification of the first embodiment is characterized by having a dictionary corresponding to the angle category of FIG. 14, and includes a standard dictionary 56, a plus angle dictionary 57, and a minus angle dictionary. 58, a special dictionary 59, a selection unit 55 to which these dictionaries are connected and page attribute information m3 is supplied, and a matching unit 54 to which the dictionary selected by the selection unit 55 is connected.

  In general, when a plurality of document images are input, there is a possibility that the direction of the document is slightly inclined depending on the input condition, and this angle is called skew. Further, not only skew but also the orientation of the document (upside down, 90 degrees difference) may be input together.

  Normally, such conditions are common to the entire page of the image, but if lexicalization is simply performed, even if the same character is input correctly, it is discriminated as a separate character due to skew and lexicographically. End up. Therefore, when creating a dictionary, the character bitmap to be searched increases, and the search takes time.

  By adopting the configuration of the modified example of the first embodiment, it is possible to perform dictionary processing at high speed against distortion due to the direction dependence of the entire document. In addition, since matching target data is prepared for each angle, a highly compressed dictionary bitmap compressed file in which image quality deterioration is suppressed can be provided as in the first embodiment even if the dictionary matching accuracy is relaxed.

  For skew, it is possible to process the angle uniformly after correcting the angle, but even if the angle is corrected by the correction, there is a possibility that the bitmap may be slightly corrupted. The effect is obtained.

  Further, as a further modification of this embodiment, the layout analysis unit 12 ′ uses a configuration in which a type is specified for each area, not for each page, and a dictionary is switched for each type for each area. Lexical control is possible. For example, there is a possibility that the font design to be used is different between the document area and the graph or table area. Therefore, the dictionary matching process can be performed at a higher speed by switching the dictionary for each type of area.

(Second Embodiment)
In the second embodiment, the character processing unit 18 is used in place of the layout analysis unit, and the object arrangement information m2 ′ and the character information m3 ″ are output. In the image dictionary unit 14, a dictionary bitmap compressed file is created at high speed by applying a dictionary according to the character characteristics.

  FIG. 16 is a block diagram showing an example of the configuration of the image compression apparatus according to the second embodiment of the present invention. In the image compression apparatus according to the second embodiment of the present invention, as shown in FIG. 16, the same processing units as those in the first embodiment are basically assigned the same numbers, and are the same except that the character processing unit 18 is different. Therefore, only the changed block will be described.

  The character processing unit 18 has the configuration shown in FIG. That is, the character processing unit 18 includes a character recognition unit 46 using known OCR (Optical Character Recognition), and a language determination unit 47 to which character code information m51 is supplied from the character recognition unit 46. The character processing unit 18 receives the image signal m1 and outputs object arrangement information m2 'as character unit information as in Patent Document 3. When the language determination unit 47 receives the code information m51 from the character processing unit 18, the language determination unit 47 determines whether the character being processed is English or Japanese and other language type information and outputs the character information m3 ″.

  Since the image component converting unit 13 receives the object arrangement information m2 'as character unit information from the character processing unit 18, it can output a component image m4 that is more accurate than in the first embodiment. In addition, the image dictionary unit 14 prepares and switches three types of dictionaries for whether the character information is English or Japanese. As described above, according to the second embodiment, it is possible to perform a compression process using a high-speed dictionary bit map by performing a matching operation according to a language type. Moreover, in 2nd Embodiment, although it switched using the language classification as character information, character information is not limited to this embodiment.

(Third embodiment)
The third embodiment particularly provides an image compression technique for compressing a bitmap by converting it into a dictionary.

  FIG. 18 is a block diagram showing an example of the configuration of the image compression apparatus according to the third embodiment of the present invention. In the image compression apparatus according to the third embodiment of the present invention, as shown in FIG. 18, basically the same processing units as those in the first embodiment are assigned the same numbers, and the image dictionary conversion unit uses a bitmap dictionary. 14 ″ is the same except that it is different. Accordingly, the third embodiment can perform high-speed compression by updating the index search order at the time of compression according to the occurrence frequency of the bitmap.

  As shown in FIG. 19, the image dictionary unit 14 ″ -1 receives the bitmap dictionary 61 in which the existing bitmaps are stored with a plurality of indexes and the page attribute information m3, and searches the bitmap dictionary 61. Index reordering unit 62 that reads out the bitmap dictionary data m41 by changing the above, and a data reading unit that supplies the read number m42 to the search index reordering unit 62 and supplies the candidate bitmap m43 from the search index reordering unit 62 63, and a matching unit 54 that is supplied with the candidate bitmap m43 from the data reading unit 63 and performs a matching process with the component image m4.

  In the image dictionary unit 14 ″ -1 having such a configuration, the search order of the bitmap dictionary 61 is changed according to the given page attribute information m3, and the search order of the bitmap dictionary 61 is changed. The bitmap dictionary data m41 corresponding to the read number m42 to be read is read, and the bitmap itself and the number index are supplied as the candidate bitmap m43 to the data readout unit 63. The matching unit 54 knows the candidate bitmap m43 and the component image m4 as known. The position shift and the matching process are combined to determine whether or not the component image m4 matches the candidate bitmap m43, and if it matches, the number index is output as the dictionary index m6. If the pattern is not in the dictionary, the part image It is registered in the bitmap dictionary 61 in accordance with a new number index and page attribute information m3 to 4.

  When the matching process for all the originals input from the scanner 11 is completed, the image dictionary unit 14 ″ -1 sorts the bitmap dictionary in the order of the number index and outputs it as an image dictionary m5. The image dictionary m5 is compressed by a known image compression technique (for example, run length), and is output to the storage unit 16 as code data m7 together with the dictionary index m6.

  Next, a characteristic dictionary search order switching effect in the third embodiment will be described with reference to FIG. For example, assuming that the processing for the four-page image has been completed up to the third page, a number index is assigned each time a new character appears, and the page attribute information is assigned.

  Here, since the fourth page is “Presentation”, the search index is rearranged, and a dictionary bitmap whose presentation is the page attribute is brought to the head of the search. It is not necessary to actually rearrange the data, and only a correspondence table of number indexes to be read is created for the search index (read number m42).

  When scanning multiple pages and creating one file, the same font (size, design) is likely to be used for the same document type. Therefore, dictionary matching is performed at high speed by preferentially searching the page attribute being processed for the search order of data for dictionary matching in accordance with the page attribute, so a dictionary bitmap compressed file is provided at high speed be able to.

  In addition, when the dictionary matching is not complete matching, the code data m7 is irreversibly compressed. However, when there are a plurality of candidates in the matching algorithm, matching accuracy is further improved by using the page attribute information together. Therefore, it is possible to provide a dictionary bitmap compressed file with high image quality and high compression.

  In the third embodiment, page numbers are also managed, and when stored in the bitmap dictionary 61 after the page search is completed, the page numbers are re-stored in the order of page numbers (third page), but the page numbers are ignored. It is also possible to combine documents into documents and presentations into presentations and page attributes.

  Note that the layout analysis method, page information calculated therefrom, dictionary formation, indexing method, search method switching method, and the like are not limited to the above-described embodiments.

(Fourth embodiment)
The fourth embodiment is characterized in that a bitmap count unit is provided in the search index rearrangement unit of the bitmap dictionary of the third embodiment.

  FIG. 21 shows an image dictionary unit 14 ″ -2 which is an example of the configuration of the fourth embodiment. The configuration is basically the same as the image dictionary unit 14 ″ -1 of the third embodiment, and the same numbers are assigned. However, a bitmap count unit 64 is newly added.

  In such an image lexicon 14 ″ -2, the matching unit 54 combines the candidate bitmap m43 and the component image m4 with a known position shift and matching process to determine whether the component image m4 matches the candidate bitmap m43. If there is no corresponding bitmap pattern in the dictionary, the bitmap image is combined with the new number index and page attribute information m3 in the component image m4. The image dictionary converting unit 14 ″ -2 sorts the bitmap dictionary in the order of the number index and outputs it as an image dictionary m5 when the matching process for all the documents input from the scanner 11 is completed.

  Here, a bitmap count unit 64 characteristic of the fourth embodiment will be described with reference to FIG. The bitmap count unit 64 has a configuration as shown in FIG. 22 (a). The page attribute information m3, the number index m46 and its index when each dictionary bitmap shown in FIG. 17 (b) is generated. A number index table 66 that stores bitmap statistical information m53 that is a count value that is an appearance frequency, a number index counter 65 that increments a count value indicated by a number index m46 in the bitmap statistical information m53, and a count value An appearance frequency rearrangement information generation unit 67 for generating appearance frequency rearrangement information m45 that is access rearrangement information in accordance with the appearance frequency within the currently processed page attribute when the value exceeds a predetermined threshold. Yes.

  Next, how the count value is incremented to update the appearance frequency rearrangement information will be described with specific numerical values given in FIG. When the threshold value for generating the rearrangement is “799” and the number index “6” of the “presentation” category is input as the page attribute information m3, the counter value “800” incremented by the number index counter 65 is the bit. It is updated as map statistical information m53 and held in the number index table 66. Here, the appearance frequency rearrangement information generation unit 67 changes the appearance frequency rearrangement information m45 of the “presentation” category number index “6” from “2” to “1” based on the updated counter value “800”. By changing and changing the appearance frequency rearrangement information m45 of the “presentation” category number index “5” from “1” to “2”, rearrangement is performed within each page attribute.

  As can be seen from the operation example of the entire image dictionary unit 14 ″ -2, the number index m46 input here is information specified as a result of matching, and thus the updated appearance frequency rearrangement information m45 is The count value m52 is applied from the next part image that exceeds the threshold value.

  As described above, in the fourth embodiment, the search index rearranging unit 62 basically performs rearrangement according to the page attribute information m3 as in the first embodiment, and at the same time, the appearance frequency rearrangement information in the page attribute. By using m45 to change the dictionary search order in the page attribute, matching processing can be performed at higher speed.

  In the fourth embodiment, the page attribute information m3 is also used to rearrange the page attributes for speeding up the rearrangement, but the page attribute information is not used in order to simplify the processing. It is also possible to generate rearrangement information only with frequency information of individual dictionary bitmaps.

  Also, the dictionary bitmap statistical processing method, update method, and the like are not limited to the above-described embodiment. In addition, the statistical processing directly uses the information of the lexicographic bitmap, but as described in Patent Document 3, image parts are formed based on the OCR result, and the same character code such as “A” is determined based on the OCR result. It is possible to group dictionary bitmaps and rearrange them according to the frequency of occurrence of the groups. As a result, a search can be performed preferentially in the entire approximate bitmap group, and a dictionary bitmap compressed file can be provided at high speed.

(Draft 1)
An analysis unit that analyzes the input image and outputs object placement information and page attribute information;
Based on the object arrangement information, the image is converted into a component, and a component converting unit that outputs a component image;
A plurality of image dictionaries corresponding to the page attribute information from the analysis unit are selected from the already existing image dictionaries, and one image dictionary is selected from the selected plurality of image dictionaries according to a predetermined order and selected. An image dictionary and the component image from the component conversion unit are compared to determine a corresponding image dictionary, and a dictionary conversion unit that outputs the determined image dictionary and a dictionary index of the component image;
An encoding unit for encoding the image dictionary and the dictionary index;
An update unit that rearranges and updates the predetermined order of the dictionary unit based on the page attribute information;
An image compression apparatus comprising:

(Draft 2)
2. The image compression apparatus according to claim 1, wherein the page attribute information referred to by the updating unit is document type information including at least one of a document and a map.

(Draft 3)
The image compression apparatus according to claim 1, wherein the page attribute information referred to by the updating unit is document direction information including at least one of a skew angle and a top and bottom.

(Draft bill 4)
Analyzes the input image and outputs object placement information and page attribute information.
Based on the object arrangement information, the image is converted into a component and a component image is output.
Select a plurality of image dictionaries corresponding to the page attribute information from the existing image dictionaries,
Select one image dictionary according to a predetermined order from the selected plurality of image dictionaries,
Compare the selected image dictionary with the component image to determine the corresponding image dictionary,
Determine a dictionary index of the component image determined by the image dictionary;
Encode the determined image dictionary and the dictionary index of the component image,
An image compression method, wherein the predetermined order is rearranged and updated based on the page attribute information.

JP 2006-254329 A JP2007-174008 JP 2005-301663 A

Claims (7)

An analysis unit that analyzes the input image and outputs object placement information and page attribute information;
Based on the object arrangement information, the image is converted into a component, and a component converting unit that outputs a component image;
An image dictionary corresponding to the page attribute information from the analysis unit is selected from among existing image dictionaries, and the corresponding image dictionary is determined by comparing the selected image dictionary with the component image from the componentizing unit. A dictionary forming unit for outputting the determined image dictionary and a dictionary index of the component image;
An encoding unit for encoding the image dictionary and the dictionary index;
An image compression apparatus comprising:   The image compression apparatus according to claim 1, wherein the input image includes a plurality of pages, and the analysis unit outputs page attribute information for each page.   2. The image compression apparatus according to claim 1, wherein the page attribute information is document type information including at least one of a document and a map.   2. The image compression apparatus according to claim 1, wherein the page attribute information is document direction information including at least one of a skew angle and a top and bottom.   The image compression apparatus according to claim 1, wherein the lexicographic unit selects an image dictionary corresponding to the object attribute information from the existing image dictionary. A character recognition unit that character-recognizes the component image output by the componentization unit and outputs character information;
2. The image compression apparatus according to claim 1, wherein the lexicographic unit selects an image dictionary corresponding to character information of the character recognition unit from existing image dictionaries. Analyzes the input image and outputs object placement information and page attribute information.
Based on the object arrangement information, the image is converted into a component, and a component image is output.
An image dictionary corresponding to the page attribute information is selected from existing image dictionaries, the selected image dictionary is compared with the component image to determine a corresponding image dictionary, and a dictionary index of the component image is output. ,
An image compression method, wherein the corresponding image dictionary and the dictionary index are encoded.

Images