JP3021928B2 - Document image processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document image processing system and a document image processing method for storing, transmitting, and recognizing a large amount of images of a document such as a voucher, which is shaded.

[0002]

2. Description of the Related Art In a document filing system currently in practical use, an image input by a scanner is binarized using a threshold value processing or a dithering method, and MH is converted.
Compressed by a method such as the MMR method or the like, and stored in a storage medium such as an optical disk.

Further, in an optical character recognition apparatus which is currently in practical use, an image input by a scanner is subjected to threshold processing, and the obtained pattern is subjected to character recognition. The character recognition device currently in practical use cannot recognize a character pattern overlapping a hatched pattern. For this reason,
In order to recognize the characters written on the hatch with the current optical character recognition device, the color of the hatch is limited to a specific color (dropout color), and only the dropout color is optically dropped. (For example, see Japanese Patent Application Laid-Open No. 56-2073).

[0004]

A first object of the present invention is to improve the compression efficiency of a shaded document image. Currently, binary image compression methods used in facsimile and document filing systems are MH method, M
These are methods such as the MR method, which encode and compress a sequence of run lengths of pixels of an image. An image obtained by inputting a shaded document by a scanner and binarizing the image by simple threshold processing or dithering has a low image compression efficiency because white points and black points frequently change. Also, a discrete cosine transform method used for compression of a multi-valued image (see “Image Engineering”, Corona Co., Ltd. (Satoshi Minami, Noboru Nakamura), pp. 144-175) is also used for images with many patterns having high spatial frequencies such as shading. Compression efficiency is low. For this reason, costs for storing a document in the document filing system and for transmitting an image in a facsimile are increased.

A second problem to be solved by the present invention is that
Recognition of characters written over the shaded area. Current character recognition devices cannot recognize character patterns that overlap a shaded pattern of a color other than the dropout color. Therefore, it is necessary to use separate scanners for the document filing system which needs to record the hatching and the optical character recognition device, which is a problem in constructing an integrated document system.

[0006]

According to the present invention, a high compression ratio is obtained by providing a means for removing and outputting a hatched pattern in a document. In addition, by providing means for extracting information relating to shading (shaded area and the like), shading can be effectively reproduced at the time of display or printing.

Another object of the present invention is to provide a system capable of recognizing characters written and printed on the hatching by removing the hatching pattern.

[0008]

When the hatching is removed from the binary image of the shaded document, the number of runs of the image is reduced. For this reason, the MH method,
The compression efficiency of the image by the MMR method or the like can be improved. Even in a multi-valued image, when shading is removed, the region of high frequency components is reduced, so that the compression efficiency of image compression by the discrete cosine transform method can be increased.

By inputting a document having a shade other than the dropout color and removing the shaded pattern from the obtained image, the same processing as in the case where the shade is removed by the conventional dropout color is performed. Images can be obtained. By using this image as an input to the character recognition device, it is possible to recognize a character pattern that is superimposed on the hatching.

[0010]

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

FIG. 1 shows the configuration of a first embodiment of the present invention. The present embodiment is a comprehensive document image processing system (for example, a document filing system) for efficiently storing (accumulating), retrieving, and outputting an image of a fixed-form document having shading.
It is.

First, a reference image of a reference document is input by a reference image input unit 101 which is an image input device such as an image scanner which reads a document and outputs a binary image. The reference image is an image of a non-written fixed form document such as a slip used in a financial institution, a public institution, a general company, or the like. The document format feature extraction unit 105 automatically extracts features for recognizing the format of a fixed document such as a frame line from the reference image. The document format feature storage unit 106 is a document format feature extraction unit 1
The features of the document extracted in 05 are stored. The reference image storage unit 107 stores a reference image.

An image of a document to be registered, that is, a registered image is registered image input unit which is an image input device such as an image scanner.
According to 102, a binary image is input. The registered image is an image of a standard document in which the reference document is manually written or the like. The document format recognition unit 108 refers to the contents of the document format feature storage unit 106, analyzes the registered image, and recognizes the format of the registered document. The document format feature storage unit 109 stores the recognition result of the document format recognized by the document format recognition unit 108. The shading removal processing unit 110 outputs a binary image obtained by removing shading from the registered image, that is, a hatched image. At this time, the reference image corresponding to the recognition result of the document format recognition unit 108 is retrieved from the reference image storage unit 107, and the density is effectively shaded by referring to the density and the number of lines in the retrieved reference image. Is removed. The shaded image storage unit 111 encodes and stores the shaded image. The shading-removed image is composed of only the frame lines and characters of the document, and the number of runs is greatly reduced. For this reason, the compression efficiency at the time of encoding by the MH method or the MMR method is high, and a larger number of documents can be stored as compared with a case where shading is not removed.

When retrieving a registered image, a retrieval condition of a document image is inputted by an image retrieval condition input unit 103 which is an input means such as a keyboard. The reference image search unit 112 searches the document format feature storage unit 109 for a document format of a document that matches the input search condition, and searches and outputs a binary image corresponding to the searched document format from the reference image storage unit 107. I do. The shaded image search unit 113 stores the shaded image of the document that matches the input search condition in the shaded image storage unit.
Search and output from 111. The strapping composite image processing unit 114 is the output of the reference image searching unit 112 and the shading removal image searching unit 1
Thirteen outputs are combined and output by ORing. By combining and outputting the reference image and the shading-removed image, an image reproducing the shading removed from the registered image can be used for display, printing, and the like. Numeral 115 denotes a magnetic disk or optical (magneto-optical) disk for storing a binary image, and a trapping composite image storage unit such as a facsimile for transmitting a binary image. Reference numeral 117 denotes a binary image. A strapping composite image printing unit such as a printer for printing, and a strapping composite image display unit 118 such as a CRT for displaying a binary image.

FIG. 2 illustrates the process of the shading removal process .
That is an image example. Reference numeral 201 schematically represents a reference image input to the reference image input unit 101, which is an image input device (FIG. 2).
(a)). The reference image 201 includes characters, frame lines, hatching, and the like, which are printed in advance. Reference numeral 202 schematically represents a registered image input to the registered image input unit 102, which is an image input device (FIG. 2B). The registered image 202 is an image of a standard document in which a reference document is written or the like. Reference numeral 203 schematically represents a hatched image which is an output of the hatched removal processing unit 110 (FIG. 2C). The shading-removed image 203 is composed of only the document frame lines and characters. Reference numeral 204 denotes an image in which the value of the pixel in the shaded area is 1 and the value of the pixel in the other area is 0 from the registered image 202, that is, the shaded area image is schematically shown ( FIG. 2D ).

FIG. 3 is a diagram showing a trapping removal processing unit 11 according to the present invention.
An example of the configuration of 0 is shown. A reference image search unit 313 searches the reference image storage unit 107 for a reference image according to the output of the document format recognition unit 108 in FIG. A processing parameter determination unit 314 determines and outputs parameters such as a matrix of a filter (described in FIG. 4) and a threshold value of a region to be subjected to shading removal processing from the retrieved reference image. The parameter may be determined in advance for each format, and the stored one may be output.
301 is an image input unit for inputting an image (202 shown in FIG. 2 (b)) is a binary image which has been dithered. If the document image on which the string is printed is binarized by the normal threshold processing, the string portion is crushed, and the characters printed on the string become unreadable. Using organized dither blurs character patterns. When a dither image based on error diffusion is used, a character pattern readable by human eyes can be obtained. However, the pixel pattern of the image due to error diffusion depends on the average density of the image, not on the rope point of the rope, and it becomes difficult to separate the rope area by texture analysis using a template or a Fourier measure. The method described here is a process particularly corresponding to the characteristics of such an error diffusion image. A blur processing unit 302, a vertical edge detection processing unit 304, and a horizontal edge detection processing unit 306 perform a linear spatial filtering process on a binary image and perform a filtering process for outputting a multi-valued image. The blur processing unit 302 has the same effect as approximately restoring the original multi-valued data from the dither image. Vertical edge detection processing unit 30
4. Each processing unit of the horizontal edge detection processing unit 306 has an effect equivalent to approximately restoring the original multi-valued data from the dither image and performing differentiation. Each processing result of the blur processing unit 302, the vertical edge detection processing unit 304, and the horizontal edge detection processing unit 306 is subjected to threshold processing by threshold processing units 303, 305, and 307. The OR processing units 308 and 309 output the OR of the outputs of the threshold processing units 303, 305 and 307. The thickening unit 310 performs an 8-neighbor thickening process on the binary image output from the OR processing unit 309. Reference numeral 311 denotes a logical product processing unit that calculates the logical product of the output of the thickening processing unit 310 and the original image, and reference numeral 312 denotes an image output unit that outputs a binary image. It is to be noted that the processing of the tether removal processing unit by using an LSI is accelerated, which leads to a reduction in hardware scale and the like, which is convenient.

FIG. 4 shows a matrix of filters used in the shading removal processing. The blur processing unit 302 performs a filter process using a matrix as shown in FIG. Through this processing, an area where black pixels are concentrated, such as characters and frame lines, can be extracted. Vertical edge detection processing unit 304
The horizontal edge detection processing unit 306 performs filter processing using matrices such as 403 and 402.
By this processing, it is possible to extract a region where the density of black pixels is rapidly changing, such as the edge of a character or a frame line. 402 is a matrix for detecting changes in the density of black pixels in the horizontal direction, and 403 is
9 is a matrix for detecting a change in the density of black pixels in the vertical direction.

FIG. 5 shows an example of the procedure of the shading removal processing in the present invention. In step 501, the blurring process of the input registered image 1 is executed to restore 2 densities, and the multi-valued image 2 is output. In step 502, threshold processing of image 2 is performed, and a binary image 3 is output. Step 5
At 03, the horizontal edge detection processing of the input image 1 is executed, and the multi-value image 4 is output. Image in step 504
4 is performed, and a binary image 5 is output. In step 505, the vertical edge detection processing of the input image 1 is performed, and the multivalued image 6 is output. In step 506, threshold processing of the image 4 is performed, and the binary image 7 is output.
In step 507, the logical sum processing of the image 5 and the image 7 is executed, and the binary image 8 is output. In step 508, the logical sum processing of the image 8 and the image 3 is executed, and the binary image 9 is output. In step 509, the thickening process of the image 9 is executed, and the binary image 10 is output. In step 510, the logical product of the image 1 and the image 10 is executed, and the binary image 11 is output. Image 11 is an image obtained by removing shading from image 1.

Although the trapping removal processing has been described using an example using a filter as shown in FIG. 4, (1) a method using a Fourier measure (R. Bajcsy, “ComputerIdentificationofVisu
alSurface ", ComputerGraphicsandImageProcessing, Vol.
2,1973, pp118-130 etc.), (2) Method using secondary measure (R
obertM.Haralicketal “TexturalFeaturesforImageClass
ification ", IEEETRANSACTIONSONSYSTEMS, MAN, ANDCYBERN
ETICS, VOL. SMC-3, No., 6NOVEMBER1973, pp. 610 to 621, etc.) can be used.

FIG. 6 shows the configuration of a second embodiment of the present invention. The present embodiment is an example of a relatively small-scale document filing system in which, for example, one type of reference image and a registered image are inputted, and the stitches are removed and combined. The document filing system shown as the second embodiment refers to a system whose main purpose is to store documents.

Reference numeral 601 denotes a reference image input unit which is an image input device such as an image scanner for reading a reference document and outputting a binary image. Reference numeral 602 denotes a reference image storage unit for storing a reference binary image. A registered image input unit that is an image input device such as an image scanner that reads a registered image in which a reference document is written or the like and outputs a binary image;
Reference numeral 604 denotes a hatching removal processing unit that removes hatching from a registered image; 605, a hatching removal image storage unit that stores a binary image from which hatching has been removed; and 606, a rope that combines two input binary images. Multiplied composite image synthesizing unit, 607: Tapped composite 2
This is a clockwork composite image output unit that outputs a value image.

In the second embodiment, since the reference image is clearly defined as compared with the first embodiment, the document format feature extraction unit 10 extracts and stores the features of the document format of the reference image.
5. The document format feature storage unit 106 is not provided. Also, the registered image does not have the document format recognition unit 108 and the document format feature storage unit 109, and the shading removal processing is directly performed on the registered image by the shading removal processing unit 604. Furthermore,
There is no need to search for registered images in consideration of the document format, because a fixed-form document or the like that is clearly defined is handled. The reference image search unit 112 and the shading-removed image search unit 113 as in the first embodiment are used.
Do not have. Upon output, the reference image storage unit 602
The reference image and the shading-removed image stored in the shading-removed image storage unit 605 may be directly synthesized by the shading composite image processing unit 606.

As shown in the first embodiment, the hatched composite image output unit 607 outputs the processing result of the hatched composite image processing unit 606 by means such as a printer, a facsimile, and a CRT. According to this embodiment, filing can be efficiently performed with a small amount of hardware when removing a ticking from a fixed document whose reference image is clearly defined.

FIG. 7 shows the configuration of a third embodiment of the present invention. This embodiment is an example of a document filing system in which image conditions such as a document format are input and combined with various stored reference images.

The image of the document to be registered, that is, the registered image is 2
A value image is input by a registered image input unit 701 which is an image input device such as an image scanner. Shading removal processing unit 7
03 outputs a binary image in which shading has been removed from the registered image, that is, a hatched image. The shading-removed image storage unit 705 encodes and stores the image from which shading has been removed by the shading removal processing unit 703. When searching for registered images, an image search condition input unit, which is an input means such as a keyboard, is used.
A search condition of a document image is input by using a button 702. The tug-removed image search unit 704 searches the hatched-removed image storage unit 705 for a hatched-removed image of a document that matches the input search condition, and outputs it. The document format recognition unit 706 analyzes the registered image with reference to the contents of the document format feature storage unit 707 stored in advance, and recognizes the format of the registered document. The reference image search unit 708 searches for and outputs a reference image corresponding to the input document format from the reference image storage unit 709 that stores the reference image in advance. The clockwork composite image processing unit 710 combines the output of the reference image search unit 708 and the output of the image search unit 704. Shaded composite image output unit 7
Reference numeral 11 outputs the processing result of the shaded composite image processing unit 710 by means such as a printer, a facsimile, and a CRT.

It should be noted that the third embodiment is characterized in that the format of a document is recognized and a shaded composite image is output when a registered image is searched, as compared with the first embodiment. That is,
In the first embodiment, first, the document format feature extraction unit 105 and the document format feature storage unit 106 determine the reference image to be input so that any registered image is not specified when the system is constructed. After extracting and storing the features of the document format of the image, and extracting and storing the features of the document format of the registered image in the document format recognition unit 108 and the document format feature storage unit 109 for the registered image, A configuration is employed in which the reference image search unit 112 searches for a document format that matches the search condition thus obtained. On the other hand, the third embodiment is based on the premise that although the registered image is not one type, it is some predetermined type of fixed-form document. The reference images of the several types of fixed-form documents are stored in the reference image storage unit 709, and the shading-removed images of the documents that match the search conditions input to the shading-removed image search unit 704 are stored. After searching from the section 705, the document format recognition section 706 recognizes the format of the registered image with reference to the contents of the document format feature storage section 707 stored in advance. Then, the reference image search unit 708 selects the images that match the reference images of several types of fixed-form documents stored in the reference image storage unit 709 in advance, and selects the desired images in the shaded composite image processing unit 710. This is to obtain a shaded composite image.

According to the present embodiment, if a number of types of reference images are stored in advance, it is possible to remove the tether from the registered image and combine them without specifying the reference image.

FIG. 8 shows the configuration of a fourth embodiment of the present invention. This embodiment is an example of a document image transmission system (e.g., facsimile) that transmits an image from which a string has been removed, which is different from the first to third embodiments.

The image of the document to be input, that is, the input image is 2
A value image is input by a document image input unit 801 which is an image input device. The shading removal processing unit 801 outputs a binary image obtained by removing shading from the input image, that is, a shaded image. The shaded image transmission unit 803 transmits the shaded image. The shaded area extraction unit 804 outputs a binary image having a value of 1 in the shaded area of the input image and a value of 0 in other areas, that is, a shaded area image, as indicated by 204 in FIG. . A shaded area image transmission unit 805 transmits a shaded area image. The shaded area generating unit 806 receives the shaded area image transmitted by the shaded area image transmission unit 805 as an input and the pixel value is 1
And outputs a hatched image, that is, a hatched reproduced image. The hatching combined image processing unit 807 combines the hatched removal image and the hatched reproduction image transmitted by the hatched removal image transmission unit 803. The strapping composite image output unit 808 outputs the processing result of the strapping composite image processing unit 807 using a printer, a CRT, or the like as described in the first embodiment.

In the fourth embodiment, a document image input unit 801, a shade removal processing unit 802, and a shaded region extraction unit 80
4 may be provided in the image transmitting apparatus, and the shading generating section 806, the shading synthetic image processing section 807, and the shading synthetic image processing section 808 may be provided in the remotely located image receiving apparatus.

According to the present embodiment, even if the image input device for inputting the document image is located at a remote place, the image can be collectively processed by transmitting and synthesizing the image from which the trapping-removed image is transmitted. The present invention is applicable to a document transmission process between a branch and a head office.

FIG. 9 shows the configuration of a fifth embodiment of the present invention. This embodiment is also a document image transmission system (for example,
Facsimile). Although the fourth embodiment shown in FIG. 8 does not have a reference image and extracts a shaded area, this embodiment stores a reference image in advance and transmits a document format. Reference numeral 901 denotes an image transmitting device, and 902 denotes an image receiving device placed remotely from the image transmitting device.

The image of the document to be input, that is, the input image is 2
The value image is input by the document image input unit 903 which is an image input device. The document format recognition unit 904 recognizes the format of the input image by referring to the contents of the document format feature storage unit 905 and analyzing the features of the binary image of the document. The shading removal processing unit 906 outputs a binary image obtained by removing shading from the input image, that is, a hatched image. The document format transmission unit 907 transmits the output of the document format recognition unit 904 from the image transmitting device 901 to the image receiving device 902. The shaded image transmission unit 908 transmits the shaded image from the image transmitting device 901 to the image receiving device 902. The reference image storage unit 909 stores a reference image that is a binary image. The reference image search unit 910 searches the reference image storage unit 909 for a binary image corresponding to the document format transmitted by the document format transmission unit 907, and outputs the binary image. Tethered composite image processing unit 911
Is the output of the reference image search unit 910 and the shaded image transmission unit 908
Synthesizes the shaded image transmitted by the above. A strapping composite image output unit 912 outputs the processing result of the strapping composite image processing unit 911 by a printer, a CRT, or the like.

In this embodiment, since the reference image has already been registered on the image receiving apparatus side, it is only necessary to transmit the document format, and the transmission time and the like can be shortened.

FIG. 10 shows the configuration of a sixth embodiment of the present invention. The present embodiment is an example of a document image processing system (for example, OCR or the like) that uses a string-removed image for character recognition.

A document image input unit 1001 as an image input device reads a document and outputs a binary image as an input image. The document to be read is a document in which characters are additionally written in a character frame designated for each format on a fixed-size sheet. The document image input unit 1001, which is the image input device, may be a device that does not drop out halftone dots, similarly to a scanner for a document filing system. The document format recognition unit 1002 refers to the contents of the document format feature storage unit 1003, analyzes features such as the frame of the input image, and recognizes the format of the input document. The character recognition field information storage unit 1005 stores the coordinates of a character frame to be subjected to character recognition corresponding to the document format. The shading removal processing unit 1004 includes the character recognition field information storage unit described above.
The above document format recognition unit of the character frame coordinates stored in 1005
A binary image obtained by removing the shading in the character frame from the input image by referring to the document format corresponding to the output document format of 1002, that is, a hatched image is output. Image smoothing unit 1006
Removes contour unevenness and noise from the hatched image. The character recognition unit 1007 refers to the character frame coordinates stored in the character recognition field information storage unit 1005, which corresponds to the document format of the output of the document format recognition unit 1002, and outputs the image in the image of the image smoothing unit 1006. Recognize characters.

According to the present embodiment, there is an effect that the line drawing from which the string has been removed can be subjected to image smoothing and used directly for character recognition.

FIG. 11 shows a seventh embodiment of the present invention. The present embodiment relates to an image input device (for example, an image scanner or the like) that detects a tether region and outputs a binary image that is not affected by the tether without removing the tether.

A document image input unit 1101 reads a document and outputs a multivalued image as an input image. Shaded area extraction unit 1102
Extracts a shaded area from the input image. The crushing prevention processing unit 1103 performs crushing prevention processing on the shaded area extracted by the shaded area extraction unit 1102 in order to prevent shading of the shaded area during binarization. As the crush prevention processing, density conversion such as gamma correction for reducing the density of the entire image is used. The binarization processing unit 1104 binarizes the image output from the crush prevention processing unit. The binarized image is output from the binary image output unit 1105.

According to the present embodiment, the shading prevention processing and the binarization processing can be performed on the shaded area and the image can be converted into a binary image without removing the stringing document from the ordinary binary image. There is an effect that the processing method can be applied as it is. Therefore, if this embodiment is applied to an image input device such as an image scanner, it can be used as it is as an image input device of a document image processing system such as an OCR and a document filing system.

FIG. 12 shows an eighth embodiment of the present invention. Also in this embodiment, a document image processing system (for example, a document filing that handles a color image) that processes a tethered area as a multi-valued image and processes an untapped area as a binary image to synthesize an image without removing the System).

A document to be registered is input as a multi-valued registered image by a document image input unit 1201 which is an image input device such as an image scanner. A shaded area extraction unit 1202 extracts a shaded area from the registered image. The area dividing unit 1203 obtains the coordinates of the shaded area and the unshaded area from the output of the shaded area extraction unit 1202. The binarization processing unit 1204 binarizes an unshaded area with reference to the area division unit 1203.
The area information storage unit 1205 stores area information, that is, area types and coordinates. The multi-level image encoding unit 1206 is divided into a region dividing unit 1203
And encode the shaded area with the multi-valued image. (If multi-valued encoding is used, even in the shaded area, the background, shaded pattern, and character pattern have differences in luminance, so they can be encoded as unique values. The multi-valued image storage unit 1207 stores the multi-valued image encoding unit 1
Store the output of 206. A binary image encoding unit 1208 encodes a binary image in an unshaded area, which is an output of the binarization processing unit 1204. The binary image storage unit 1209 is a binary image encoding unit 1208
The output of is stored.

When retrieving a registered image, a retrieval condition of a document image is inputted by a retrieval condition input unit 1210 which is an input means such as a keyboard. The area information search unit 1211 searches the area information storage unit 1205 for area information of a document that matches the input search condition, and outputs it. The multi-value image search unit 1212 searches the multi-value image storage unit 1207 for a coded multi-value image of a document that matches the input search condition, and outputs the coded multi-value image. The binary image search unit 1213 searches the binary image storage unit 1207 for an encoded binary image of a document that matches the input search condition, and outputs it. The multi-level image decoding unit 1214 decodes the output of the multi-level image search unit 1212 into a multi-level image. The binary image decoding unit 1215 decodes the output of the binary image search unit 1213 into a binary image. The image synthesis processing unit 1216 refers to the area information output from the area information search unit 1211 and outputs the output of the multi-value image decoding unit 1214 and the binary image decoding unit 1215.
Are combined to output a multi-valued image. A multi-value image output unit 1217 outputs a multi-value image, and outputs a processing result of the combined image processing unit 1216 by a printer, a facsimile, a CRT, or the like.

In this embodiment, a shaded area is extracted,
Although only the area is coded with multi-values, it is needless to say that a shaded document can be processed even if all the areas are coded with multi-values.

According to the present embodiment, since the shaded area is coded with multi-values, it is possible to process a clear image regardless of whether the shade is dark or light.

FIG. 13 shows a ninth embodiment of the present invention. The present embodiment is an example of a document image processing system (for example, a document filing system) that performs the removal of a trap according to conditions. The concept of the present embodiment can be applied to the above-described embodiments (first to sixth embodiments) in which shading is removed even if the object is different.

Reference numeral 1301 denotes a document image input unit which is an image input device such as an image scanner which reads a document and outputs a binary image. The shading removal processing condition input unit 1302 inputs a condition as to whether or not to execute shading processing. This condition is
Information such as the type of format and the image density can be input by either the system or the operator. Shading removal processing area judgment unit
A step 1305 determines whether or not a specific area (for example, a shaded area, a part of the shaded area, etc.) of the image input from the shade removal processing condition input unit 1302 satisfies the condition. (Of course, not only the specific area, but also the entire area of the document if the area is shaded.) The area dividing unit 1306 may determine the shaded area according to the output of the shaded area determination unit 1305. When the input image from the document image input unit 1301 is divided into regions and shading removal is performed according to the regions, the shading removal processing unit 1303 sends the other regions directly to the image synthesis processing unit 1307. Output a value image. The shading removal processing unit 1303 removes shading of a document in a specific area that satisfies the condition from the shading removal processing condition input unit 1302. The image synthesis processing unit 1307 obtains a logical sum of the binary image from the hatching removal processing unit 1303 that has performed shading removal on the specific area and the binary image of the other area to synthesize the image. Rope hanging removed image output unit 1304 is a binary image output means for outputting the binary image synthesized by the image synthesis processing unit 1307, you output printer, a facsimile, and the like CRT.

According to the present embodiment, the conditions for removing the tether can be changed, so that there is an effect that various applications can be handled.

FIG. 14 is a diagram for explaining a method of correcting an image from which a leash has been removed according to the present invention so that the image does not have a visually unnatural feeling. Reference numeral 1501 denotes an image before the trapping removal processing is performed. Reference numeral 1502 denotes an image after the trapping removal processing has been performed. Reference numeral 1503 denotes an image obtained as a result of writing an artificial strap on a portion where the strap is removed. The written artificial rope is a pattern having a high compression effect by encoding in the MH method or the like, such as a horizontal stripe. By this processing, an image having a high compression effect by encoding is obtained, which is visually similar to the image before removing the tackle.

In the document image processing system (including the document filing system) described in the above-described embodiments, the processing such as the removal of the trapping and the extraction of the shaded area reduces the image data of the drawing using the trapping. To do, or
It can be applied to various applications such as recognizing characters in screen-printed photographs.

Also, as an image input device such as a CCD sensor or an image scanner having a trapping removal processing section, an image from which trapping has been removed can be output, so that ordinary image processing can be applied as it is. Therefore, if the leash removal processing unit is constituted by an LSI or the like, it becomes more effective as an image input device.

[0052]

According to the present invention, an image of a shaded document can be compressed at a high compression rate and used for storage and transmission. Shading is reproduced during display and printing, and the original document can be faithfully reproduced.

Characters written and printed on non-dropout color shading, which were conventionally difficult, can be recognized.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a document image processing system according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an effect of a shading removal process.

FIG. 3 is a configuration diagram of a hatching removal processing unit.

FIG. 4 is a diagram illustrating a procedure of shading removal processing.

FIG. 5 is a diagram illustrating a matrix of filters used in a shading removal process.

FIG. 6 is a configuration diagram of a document filing system according to an embodiment of the present invention.

FIG. 7 is a configuration diagram of a document filing system according to an embodiment of the present invention.

FIG. 8 is a configuration diagram of a document image transmission system according to an embodiment of the present invention.

FIG. 9 is a configuration diagram of a document image transmission system according to an embodiment of the present invention.

FIG. 10 is a configuration diagram of a document image processing system as a character recognition device according to an embodiment of the present invention.

FIG. 11 is a configuration diagram of a document image processing system according to an embodiment of the present invention.

FIG. 12 is a configuration diagram of a document filing system according to an embodiment of the present invention.

FIG. 13 is a configuration diagram of a document image processing system according to an embodiment of the present invention.

FIG. 14 is an explanatory diagram for correcting an image that has been subjected to a trapping removal process.

[Explanation of symbols]

102: registered image input unit, 108: document format recognition unit, 1
10: Shading removal processing unit; 103: Image search condition input unit; 114: Twill composite image processing unit; 803: Twig removal image transmission unit; 1007: Character recognition unit.

──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Masaaki Fujinawa 1-280 Higashi-Koikekubo, Kokubunji-shi, Tokyo Inside the Hitachi, Ltd. Central Research Laboratory (72) Inventor Hiromichi Fujisawa 1-280 Higashi-Koikekubo, Kokubunji-shi, Tokyo Hitachi, Ltd. Central Research Laboratory (72) Inventor Hiroshi Samori 1-280 Higashi Koikekubo, Kokubunji-shi, Tokyo Inside Hitachi, Ltd. Central Research Laboratory (72) Inventor ▲ Taka ▼ Hashio 2880 Kozu, Kokuzu, Odawara-shi, Kanagawa Prefecture Hitachi Odawara Plant, Ltd. (72) Inventor Kazunori Kinoshita 2880 Kokuzu, Odawara City, Kanagawa Prefecture Inside the Odawara Plant, Hitachi, Ltd. (72) Inventor Kiyomichi Kurino 2880 Kokuzu, Kodawara City, Kanagawa Prefecture, Japan Inside the Odawara Plant, Hitachi, Ltd. 2880 Kozu, Odawara City, Kanagawa Prefecture (56) References JP-A-2-252082 (JP, A) JP-A-2-293976 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06T 1 / 00,5 / 20 G06K 9 / 36,9 / 40 G06F 17/30 H04N 1/387

Claims (1)

(57) [Claims] 1. A multi-valued image is output from a shaded document image by restoring the density by performing a blur process on a binary image of the input image, b) By performing threshold processing on the multi-valued image, the first 2
(C) performing horizontal and vertical edge detection processing of the multi-valued image, and (d) performing threshold processing on each of the multi-valued images subjected to the edge detection processing. (E) ORing the second and third binary images to output a fourth binary image, and (f) outputting the first and fourth binary images. (G) performing a thickening process on the fifth binary image to output a sixth binary image, and (h) performing the above (a). A) ANDing the binary image and the sixth binary image input in step (d), and outputting a shaded image.