JP3384208B2 - Image reading device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image reading device for reading a book manuscript or the like by optical scanning.
The present invention relates to a technique for erasing an image of an end face of a document or a finger pressing a document by detecting a margin part existing on the left and right of the document, a margin part existing in the center of the document, and the like.

[0002]

2. Description of the Related Art Conventionally, in this type of image reading apparatus, a change point at which image data changes from black to white is obtained,
There is a processing technique for erasing an unnecessary image outside the image area by recognizing the image area and converting the image other than the image area into white data. Place the surface of a document such as a book face up on the platen, and read the image with the reading unit provided above the platen while holding both ends of the document with your fingers to prevent the page from floating. There is a device. in this way,
In an apparatus that reads an image while pressing a document with fingers, there is a possibility that it cannot be detected as a change point of image data because the color of the fingers is a skin color. Therefore, the image of the finger may not be deleted even with the above-described image processing technique.

An image reading apparatus has been proposed for the purpose of solving such a problem. One is to photograph the document with a color sensor and detect the skin color part of the finger from the color information to determine the photographing area and determine the position of the finger holding the document to erase the image of the finger. Device (see, for example, Japanese Patent Laid-Open No. 6-105091).

The other one is a device for detecting the photographing area by obtaining the number of pixels of the character area based on the difference in brightness between the background of the document or the area of the character and the area of the finger, and erasing the area of the finger. (For example, see Japanese Patent Laid-Open No. 6-78133). Another known device is one that takes an image of a document with a monochrome sensor, determines the area where the finger is placed and erases the image of the finger based on the brightness difference between the document background and the shadow of the finger in the captured image. Has been. Further, there is an image reading apparatus that interpolates image data line by line in the sub-scanning direction in order to correct shrinkage of a document image in the sub-scanning direction caused by a change in the height of a document such as a book. This reading device requires a buffer memory for the number of lines to be interpolated, and the correction amount is determined by the capacity of the buffer memory.

[0005]

However, in the image reading apparatus which detects the position of the finger based on the color information and erases the image of the finger as described above, the detection is erroneous due to the difference in the flesh color density and the tint. There was a possibility. Further, in an image reading apparatus that detects a shooting area only from the number of pixels of a character portion, a blank portion may occur at the center side from the position to be originally erased. There was a case where the position of the copy was erroneously determined and the necessary document portion was deleted. Furthermore, since the end face of a thick book has low luminance, the luminance of the finger cannot be discriminated from the luminance of the background of the original document at that portion, and therefore the photographing region may be erroneously determined. Further, in an image reading apparatus that interpolates image data line by line in the sub-scanning direction in order to correct shrinkage in the sub-scanning direction of a document image caused by a change in the height of a document such as a book, the number of lines to be interpolated Buffer memory is required, and the correction amount is determined by the capacity of the buffer memory. Therefore, if the distortion is large, the correction will be insufficient. Must be used, which increases the cost of the device.

The present invention solves the above-mentioned problems, and in an image reading apparatus for reading an original by optical scanning, the position of a finger pressing the original and the determination of the margin position of the original are not erroneously detected correctly. It is an object of the present invention to provide an image reading apparatus capable of determining a photographing area and accurately erasing margins on the left and right edges of a document. In addition to the margins on the left and right end faces of the document, the present invention also provides a margin at the center of a document such as a book or a margin between lines of a document in which characters are written vertically (these are collectively referred to as margins). The image reading device that can reduce the cost of the device by effectively using the buffer memory by interpolating the white data without using the buffer memory for the image corresponding to the blank part) is detected. It is intended to be provided.

[0007]

In order to achieve the above object, the invention according to claim 1 is an image reading apparatus for reading an original placed on an original table by optical scanning. An element, a storage unit for temporarily storing the brightness information of the image data read by the image sensor, and an image data read by the image sensor.
Position detecting means for detecting position information in the left-right direction corresponding to the horizontal direction, and brightness detecting means for detecting the background brightness of the document based on the histogram of the brightness information stored in the storage means.
Counting means for counting the number of pixels to be judged as a character portion based on the histogram of the brightness information stored in the storage means as the character frequency, and the position information of the document in the left-right direction ,
Background luminance, and the margin detection means for detecting a blank portion on the basis of the information of the character frequency to the left and right of the document, the right and left sides of the document
Is obtained by an image processing means for erasing the lateral direction of the outside of the image of the margin, which is detected by the margin detection means the center direction as a reference.

In the above arrangement, the image pickup device simultaneously scans and reads the document surface set on the document table and the image of the finger pressing the document. The storage unit detects the brightness information from the read image information for each line in the vertical direction of the document and temporarily stores the information. Position detecting means detects the position information in the lateral direction of the read image information. The brightness detecting means creates a brightness histogram from the temporarily stored brightness information of one line, and sets the background brightness of the original document and the brightness threshold value to be judged as the character portion from the histogram shape. The counting means is
The number of pixels below the set brightness threshold of the character portion is counted and stored as the character frequency. The margin portion detecting means comprehensively judges the horizontal position information, the background luminance information, and the character frequency information in line units obtained as described above, and determines the horizontal position which is considered to be the most blank in the manuscript. To detect. The image processing means erases the image information outside the margin position detected by the margin detection means. In this way, the image of the end face of the document or the image of the finger can be erased.

According to a second aspect of the present invention, in an image reading apparatus for reading a document such as a book set on a document table by optical scanning, an image pickup device for reading the document and the height of the document are detected. Height detection means, storage means for temporarily storing brightness information of image data read by the image pickup device, and image read by the image pickup device
Position detecting means for detecting position information in the horizontal direction corresponding to the data, Hisutogu luminance information stored in the storage means
A luminance detecting means for detecting a background luminance of the document on the basis of the ram, a counting means for counting the number of pixels of the character portion based on the histogram of the luminance information stored in the storage means, the document height, the left-right direction Margin information detecting means for detecting a margin portion of the document based on each of the position information , background luminance, and the number of pixels of the character portion, and the image of the margin portion detected by the margin portion detecting means is used as white data. For white de
Data generation means and the contraction of the document in the sub-scanning direction caused by the height change of the document are stored in the same storage means .
It is provided with a correcting means for correcting by repeatedly reading out the image information of the row, and a blank portion correcting means for replacing the image data of the blank portion detected by the blank portion detecting means with white data. .

In the above structure, the storage means temporarily stores the brightness information of the image data read by the image pickup element, the position detection means detects the left and right positions of the original read by the image pickup element, and the brightness detection means, The background luminance of the original is detected based on the histogram of the luminance information stored in the storage means, and the counting means detects the hysteresis of the luminance information stored in the storage means.
The number of pixels in the character portion is counted based on the toggle . The margin part detecting means is for the height of the document, the horizontal position, the background brightness,
And the margin portion of the document (including the margin portions on the left and right end faces of the document described above) based on the respective information of the number of pixels of the character portion, and the white data generating unit whitens the image of the detected margin portion. Data. The correction unit corrects the contraction of the document in the sub-scanning direction caused by the height change of the document by repeatedly reading the image information stored in the storage unit. The blank area correction unit corrects the shrinkage of the blank area in the sub-scanning direction by repeatedly outputting white data without passing through the storage unit. In this way, the white data can be interpolated in the image of the blank space such as the space between lines of the document without using the storage means such as the buffer memory, so that the capacity of the buffer memory can be reduced.

[0011]

DETAILED DESCRIPTION OF THE INVENTION An image reading apparatus according to an embodiment of the present invention will be described below with reference to the drawings. Figure 1
Shows the overall configuration of the image reading apparatus, and FIG. 2 shows a state in which an original is placed on the original table of the image reading apparatus. A book document 10 such as a book or a file, which is a subject, is placed facing upward on the document table 1, and a line sensor for reading the spread surface of the book document 10 by optical scanning is provided above the book document 10. The imaging device 2 is provided. The original placing surface of the original table 1 is colored darker than the background color of a general original, and when the book original 10 is read with the original placing surface as the background, the original surface and the original placing surface are distinguished from each other. You can do it. In this apparatus, an illumination unit 3 arranged above the document table 1 and illuminating the book document 10 on the document table 1, an operation panel 4 for setting image reading conditions and the like, and a document table 1 at the back. And a control unit (see FIG. 20) which is disposed on the side and controls the photographing operation such as the preliminary scanning operation and the main scanning operation of the image pickup device 2 of the book document 10 which is inclined and reflects the side surface of the book document 10.
Is provided. The image data photographed by the image pickup device 2 is subjected to various processes by the control unit and output to a desired output device (printer, computer, etc.).

Here, the names of the respective parts of the book document 10 placed on the document table 1 are defined (see FIG. 2). The entire left and right pages of the opened book document 10 are referred to as a document surface 10a, the right outer surface inclined with respect to the document surface 10a is a right surface 10f, and the left outer surface is a left surface 10f '.

FIG. 3 and FIG. 4 show schematic configurations of the present apparatus as viewed from the front and the sides, respectively. The image pickup apparatus 2 includes a CCD line sensor 7 in which a plurality of image pickup elements are arranged on a line from the front side to the back side (main scanning direction), and a document surface 10a.
And an optical system having an image pickup lens 6 for projecting the image on the line sensor. The line sensor 7 moves in the sub-scanning direction (the arrow direction in FIG. 3) orthogonal to the main scanning direction on the image forming surface on which the image of the document surface 10a is formed,
The image on the document surface 10a is read. Further, the imaging lens 6 is
The lens drive unit 33 (FIG. 20) is provided so as to be movable in the optical axis direction. The lens drive unit 33 (FIG. 20) moves according to the height of the book original 10 obtained by height detection described later, and is always in focus on the line sensor 7. The image of the document surface 10a is formed by.

The height detecting mirror 5 extends in the sub-scanning direction on the inner side of the document table 1 and is 45 ° with respect to the document mounting surface of the document table 1.
It is installed so as to be inclined at an angle of, and reflects the side surface of the book document 10 placed on the document table 1. The side surface image of the book document 10 reflected on the mirror 5 is projected by the lens 6 together with the document surface 10a. The line sensor 7 has a sufficient length to read the projected original surface 10a and the side surface image projected on the mirror 5, and simultaneously scans the original surface 10a and the side surface image by scanning movement. Although the line sensor is used in this embodiment, an area sensor may be used instead of the line sensor.

When the document surface 10a of the book document 10 is opened and placed on the document table 1, it bends in the height direction. Therefore, the image on the document surface 10a is shrunk in the sub-scanning direction, and the length of the image in the main scanning direction is long in the portion where the height of the document is high and short in the portion where the height of the document is low. As a result, image distortion occurs. Further, since the distance from the image pickup device 2 to the document surface 10a also changes depending on the height of the document, it is necessary to adjust the focus according to the height. In order to solve such a problem, the height of the book document 10 at each position in the sub-scanning direction is detected, and the image distortion is corrected and an image is formed on the line sensor 7 according to this height. Adjust the image focus.

FIG. 5 is a diagram showing the principle of the height detection processing of this embodiment. By placing the book original 10 at a predetermined position, an image 11 of the side surface of the book original is reflected on the mirror 5, and an image 11 of the side surface of the book original 10 reflected on the mirror 5 is displayed.
Is read by the line sensor 7 to obtain the height distribution of the book document 10.

FIG. 6 shows a state of image data read by the image pickup apparatus 2 having the above structure. In the figure, a is the image of the document surface 10a, b is the image of the document table 1, c is the image of the background portion reflected on the mirror 5, d is the image of the side surface of the book document 10 reflected on the mirror 5, and e is the document. The alignment reference of is shown. The image a on the document surface and the image d on the document side surface are read as being curved in the main scanning direction due to the height change of the document. The document surface and the document side surface are generally white and are read as white. On the other hand, the image of the document table 1 and the background portion c reflected on the mirror 5 which are colored darker than the background of the document have a small amount of reflected light and are read in black.

FIG. 7 shows the output of one line in the main scanning direction read by the line sensor 7. In this example, the case where the image at the position indicated by the dotted line in FIG. 6 is read by the line sensor 7 is shown. The horizontal axis represents the address of the image sensor of the line sensor 7, and the vertical axis represents the output of each image element (image brightness). In the figure, is an image c of the background portion reflected on the mirror 5, is an image d of the side surface of the document reflected on the mirror 5,
Indicates an area on the image sensor of each image a of the document surface 10a and image b of the document table 1. Dth is a predetermined threshold value for discriminating whether the image is a document image or another image. n1 is the threshold Dt
The minimum address value of the image sensor whose output exceeds h, that is,
It is a value indicating the position where the upper edge 10b of the document surface is imaged in the image 11 of the document side surface. n2 is an address value of the image sensor corresponding to the document alignment reference and is a fixed value. (N2-n1) is the number of pixels corresponding to the document height used in the height detection processing. (N2- of each line
The height distribution data of the document (see FIG. 10) is obtained from the value of n1). From this height distribution data, the image pickup lens 6 is driven in the vertical direction so as to eliminate the image distortion correction coefficient for correcting the image distortion caused by the height change of the document and the defocus caused by the height change of the document. And automatic focus (AF) control data for calculating

Next, the detection of the background brightness of the original and the number of pixels of the character portion will be described. FIG. 8 shows a state of an image when the book document 10 is imaged by the CCD line sensor 7 of the imaging device 2. To prevent the pages of the book document 10 from being turned over, the operator presses the part of the book document 10 on which no characters are written (margin part) with fingers. The image data for at least one line in the main scanning direction captured by the CCD line sensor 7 is temporarily stored in the storage unit (buffer memory 29 in FIG. 20). A luminance histogram is created from the temporarily stored image data of one line.

FIG. 9 shows a luminance histogram prepared from the image data of one line of the dotted line shown in FIG. The horizontal axis represents the output of each image element (image brightness), and the vertical axis represents the number of pixels at that brightness. As shown in the figure, in the histogram of one line, the number of peak pixels (Np) is calculated on the higher brightness side, and the number of pixels is half the number [1 / p].
2) · Np] of the brightness values (LB, LB ') corresponding to
The lower luminance value (LB) is the original background luminance of the line. A brightness value obtained by subtracting a predetermined value (L) from the background brightness is set as a brightness threshold value (LC) to be determined as a character portion.

The method of obtaining the original background luminance and the character portion judgment threshold value is not limited to the above method, and the number of pixels is counted from the higher luminance, and when the predetermined number of pixels is counted. The brightness value may be the base brightness.

FIG. 10 shows original height distribution data in the sub-scanning direction. FIG. 11 shows a background luminance distribution obtained as a result of obtaining a background luminance LB by creating a luminance histogram for all lines in the sub-scanning direction of the document image shown in FIG.
As can be seen from FIG. 10, since the height of the central portion of the book document 10 when the pages are widened is low, the distance from the illumination unit 3 and the image pickup device 2 is large, and the inclination is relative to the illumination unit 3. Therefore, the background luminance becomes a low value. Further, the left and right end portions of the book original 10 also have a low background luminance for the same reason as the central portion. Tx and Tx 'indicate the highest positions of the left and right pages of the book document 10, and x1 and x1' are
The boundary positions between the document surface 10a and the left and right side surfaces 10f and 10f 'are shown.

FIG. 12 shows the above-mentioned character portion luminance threshold value L for all lines in the sub-scanning direction of the original image shown in FIG.
FIG. 6 is a character frequency distribution chart obtained by counting the number of pixels having a luminance lower than C. In the information part in which the content of the manuscript is written, there are many parts with low brightness such as characters and photographs, and the frequency of characters increases. Since the content of the document is not written in the central part of the document, the breaks of paragraphs, and the left and right margins, the character frequency has a small value. Further, when the fingers are not placed on the edge of the document, the background luminance of the edge is low, so the threshold value of the character portion is also low, and as a result,
The character frequency has a small value (broken line in part A of FIG. 12). When the finger is placed on the edge of the document, if the brightness of the finger is lower than the character threshold of the edge, it is counted as the character frequency (solid line in part A of FIG. 12), but higher than the character threshold. For example, it is not counted as a character frequency. That is, the frequency of characters varies greatly due to the difference in brightness between the end portion and the finger portion. The solid line of the portion A is a portion that is erroneously detected because the brightness of the finger is lower than the character portion threshold value at the end portion. Tm is a predetermined value of the number of character pixels for determining whether or not it is a blank area.

FIG. 13, FIG. 14, and FIG. 15 show the membership function of the probability (PB) of the blank space to be erased with respect to the original background luminance, the character frequency, and the left and right position values. These functions are stored in storage means (memory 2) described later.
It is stored in 8). The definition of the margin here is the original side 1
It is the left and right parts within 0a where there are no characters or photographs. The membership function value is closer to 1 when the probability of a blank to be erased is higher, and closer to 0 when it is lower.

In FIG. 13, the original background luminance of the original is a predetermined threshold value T.
If it is lower than L, the left and right side surfaces 10f, 1 outside the document surface 10a
It is determined to be 0f ′, and the probability of being a blank is 0.
As the document background luminance becomes larger than the threshold value TL, the probability of being a blank becomes closer to 1.

Further, in FIG. 14, the number of character pixels is a predetermined value T.
If it is larger than m, there is a high possibility that there are characters, photographs, etc., so the probability of being a blank is low, and the number of character pixels is a predetermined value Tm.
If the number is smaller, it is considered that the probability of being a blank is higher. The solid line in the figure shows the membership function in which the probability of being a blank decreases from 1 to 0 in proportion to the increase in the number of character pixels from 0 to Tm, and the broken line shows the membership function from 0 to Tm /
A membership function in which the probability of being a blank is 1 up to 2 and the probability of being a blank is 0 when the number of character pixels is Tm / 2 or more is shown.

Further, in FIG. 15, the closer to the center of the book document 10, the closer to the information part, there are often breaks of paragraphs that should not be erased and margins of the vertically written document. It is considered that the probability of being a margin is higher, and the margins are more likely to be located on the left and right sides from the highest position Tx, Tx ′ of the original page. Therefore, the solid line in the figure indicates the position T where the left and right positions are the highest in the document.
The margin is proportional to the position of 2 × Tx from x (assuming that Tx is the center position of each page, the position of 2 × Tx is the page edge, and the probability of being a blank space at this 2 × Tx is 1.). A membership function in which a certain probability increases from 0 to 1 is shown, and a broken line indicates a blank space when the position is 0 to Tx × 3/2, and a blank space when the position is Tx × 3/2 or more. A membership function with a probability of 1 is shown. Although the membership functions shown by the solid lines are used in this embodiment, these membership functions can be appropriately set including other forms.

FIG. 16 shows the background luminance distribution of FIG.
12 shows the output value (PBL) multiplied by the membership function of FIG. 17, and FIG. 17 shows the output value (PBM) obtained by multiplying the character frequency distribution of FIG. 12 by the membership function of FIG. Portion B in the figure is a blank area when fingers are not placed. Further, FIG. 18 shows the output value (PBX) of the membership function of FIG. 15 with respect to the sub-scanning position of the document. Each of them shows the distribution of the margin probability (PBL, PBM, PBX). By multiplying these three margin probabilities, as shown in FIG. Probability of blank space at each position (PB
) Can be asked. The position with the highest probability of being a margin is the margin position (Ex, Ex
It is determined to be '), and the image outside thereof is erased. When the fingers are placed, the boundaries x2 and x2 ′ between the inside where the fingers are placed and the blank portion are the erasing positions, and when the fingers are not placed, the boundaries x1 between the front surface and the end surface of the document. x1 '
Is the erasing position, and the image on the outside is erased.

FIG. 20 shows a block configuration of a circuit in the control section. In this embodiment, the line sensor 7 precedes the actual scanning operation of image reading (main scanning operation), and the height of the document surface 10a at each position in the sub-scanning direction,
A preliminary scan operation is executed to detect the background brightness and the number of character pixels. The output (image data) of the line sensor 7 obtained by the preliminary scan is A / D converted by the A / D converter 21 in order from the image pickup device at address 1 for each line.
After D conversion, it is input to the comparator 22. Comparator 22
In the above, the above-mentioned threshold value Dth is preset by the CPU 23. When pixel data having a level exceeding the threshold value Dth is input to the comparator 22, the count value of the counter 24 is stored in the memory 28. Counter 24
Is for counting in synchronization with the dot clock given to the line sensor 7, and the count value of the counter 24 indicates the address of the image data compared by the comparator 22. The CPU 23 recognizes the minimum value of the count values fetched in the memory 28 for each line as n1, and stores n1 for each line in the memory 28. The height of each line of the book document 10a is obtained from the difference between this n1 and the reference position address n2.

The line sensor 7 based on the preliminary scan
Is output from the A / D converter 21 and is input to the buffer memory 29. The image data stored in the buffer memory 29 is sent to the CPU 23,
The CPU 23 creates the brightness histogram described with reference to FIG. 9 for each line based on the sent image data,
From this luminance histogram, the background luminance LB of the original and the character portion threshold value LC are calculated. The calculated background luminance LB is stored in the memory 28.

The CPU 23 counts, for each line, the pixels whose luminance value is below the character portion threshold value LC from the luminance histogram, and obtains the number m of character pixels included in one line. The character pixel number m is stored in the memory 28, and the character frequency distribution is obtained. Further, the CPU 23 detects the position X in the sub-scanning direction of the document currently being photographed by the CCD line sensor 7. The position X in the sub-scanning direction is stored in the memory 28. The CPU 23 has functions of a brightness detecting unit, a character pixel number counting unit, and a position detecting unit.

The image data of the line sensor 7 obtained by the main scanning operation is A / D converted by the A / D converter 21 in order from the image pickup device at address 1 for each line, and then the image of several lines is obtained. The data is sequentially written in the buffer memory 29 capable of storing data. Since the written image data includes the image data of the side surface of the book document 10 described above, the image data of the side surface is deleted, and only the image data of the document surface 10a is sequentially read by the image processing circuit 30, The printer 31 receives appropriate correction processing.
Output to and printed.

Here, in the image processing circuit 30, the count value n1 of each line obtained by the preliminary scan
(Actually, based on the height data (n2-n1)),
The image distortion in the main scanning direction and the sub scanning direction is corrected.
Further, during the main scan operation, the CPU 23 outputs a control signal to the lens driving device 32 based on the count value n1, moves the lens according to the reading position of the line sensor 7, and moves the lens above the line sensor 7. Therefore, the image on the document surface 10a is always focused. Furthermore, CPU
Reference numeral 23 determines the blank areas to be erased located on the left and right sides of the document by processing the above-described three pieces of information on the background luminance, the character frequency, and the sub-scanning position, and the image processing circuit based on the blank area data. By controlling 30, the image is erased. The blank space detection unit is configured by the CPU 23.
Further, the CPU 23 outputs a control signal to the sensor moving unit 25 and the lamp control unit 27 to control the scan movement of the line sensor 7 and the lighting of the lamp of the illumination unit 3.

FIG. 21 is a flow chart showing the procedure of the reading operation controlled by the CPU 23 of the image reading apparatus constructed as described above. When the start of the reading operation is input from the operation panel 4, the CPU 23 turns on the lamp of the illumination unit 3 via the lamp control unit 27 to illuminate the book document 10 (# 1). Then, the sensor moving unit 25 is instructed to start the preliminary scan (# 2),
The sampling of the count value n1 for each reading line is executed. In this operation, while moving the CCD line sensor 7 from one end in the sub-scanning direction, the side surface of the document reflected on the mirror 5 and the document surface 10a of the document 10 are photographed, and the count value n1 is stored in the memory 28 (# 3). The read image data is input to the buffer memory 29. The image data stored in the buffer memory 29 is stored in the CPU 2
3, the CPU 23 creates the brightness histogram described with reference to FIG. 9 for each line based on the sent image data (# 4). From this luminance histogram, the background luminance LB of the original and the character portion threshold value LC are calculated. The calculated background luminance LB is stored in the memory 28 (# 5). Further, the CPU 23 counts the number of data whose luminance is lower than the calculated character portion threshold value LC (# 6). This operation is repeated in a constant cycle until the preliminary scan is completed for all lines, and when the preliminary scan is completed (YES in # 7), height data is obtained from the sampled count value n1 (# 8).

Next, the membership function shown in FIGS. 13 to 15 is applied to the background luminance LB, the number of character pixels m, and the sub-scanning position X obtained in the preliminary scan,
Probability of margins PBL and PB shown in FIGS.
M and PBX are obtained, and the margin position is determined from the probability PB that is the total margin shown in FIG. The membership function is stored in the CPU 23, and the calculation up to the determination of the margin position is also performed by the CPU 23 (# 9). When the above operation is completed, next, the sensor moving unit 25 is instructed to start the main scan, the line sensor 7 is scanned and moved in the direction opposite to the preliminary scan, and the main scan for photographing the original 10 is performed. (# 10). During the main scan operation, the CPU 23 outputs a control signal to the lens driving device 32 based on the height data obtained in the process of # 8 to execute focus adjustment of the lens 6 (# 11), and In order to correct the distortion or the like of the pixel data obtained by the scanning operation, the correction amount is set in the image processing circuit 30 based on the height data. At this time, the CPU 23 controls not to output the line sensor 7 with respect to the margin position obtained in # 9, thereby deleting the image determined as the margin position (# 12). When the main scan operation is completed for all lines (YES in # 13), the lamp is turned off (# 14), and the image reading operation is completed.

Next, another embodiment of the present invention will be described with reference to FIGS. In the above-described embodiment, the left and right margins of the document are detected and the outside thereof is deleted. However, in the following embodiments, a membership function different from that used in the above-described embodiment is used. In addition to the left and right margins and sides of the original, it detects the margins in the original, such as the binding in the center of the original such as a book and the margin between the lines of the original in which characters are written vertically, With respect to the detected portion, processing for replacing the image data with white data is performed.

More specifically, the embodiment described above (eg, FIG. 1)
In 5), it is described that the probability of a blank in the sub-scanning direction is set to be high only on the left and right sides of the document, and the detected blank portion is erased from the image data.
Another membership function is used to detect a binding part at the center of a document such as a book, a margin part such as a line space of a document in which characters are written vertically, and the sub-scanning caused by the height of the document. The image distortion in the direction is corrected. Regarding this image distortion correction, there is an image reading apparatus that interpolates image data in line units in order to correct shrinkage in the sub-scanning direction of an image caused by a change in the height of a document. In this type of reading device, the image data read in the sub-scanning direction is output line by line to an image output device or the like, and when the image data is interpolated, the read image data is repeatedly output. Further, while the image data is repeatedly output, the original is continuously read, and the image data read during this time is temporarily stored in the buffer memory provided in the apparatus, It is output after the repeated output of data is completed. Therefore, a buffer memory having a capacity for the number of lines for interpolating image data in order to correct the contraction of one original in the sub-scanning direction is required. In other words, the correctable amount of shrinkage in the sub-scanning direction is determined by the capacity of the buffer memory. In such an image reading measure, even when the shrinkage of the blank portion of the document without characters, photographs, etc. is corrected, even if one line is entirely white, that is, line data without image information, this is temporarily stored in the buffer memory. Since the stored data is stored, when correcting a document with a lot of blank space, the capacity may be exceeded only by storing the line data having no image information in the buffer, and the shrinkage correction in the sub-scanning direction cannot be performed sufficiently. There is a problem that ends up. The embodiment described below solves such a problem.

FIG. 22 is a graph showing the relationship (function) between the inclination angle θ of the document obtained from the change in the height of the book document 10 and the probability Pθ of the blank portion in the read image of the book document 10. . If the inclination angle is high, there is a high possibility that the binding portion is at the center or the left and right side surfaces 10f and 10f ', and there is a high possibility that the content is not written.

FIG. 23 shows the background luminance LB of the book original 10.
9 is a graph showing the relationship between the probability PLB that is a blank area and the probability PLB. If the background brightness LB of the original is low, it is considered that the book is in the central binding area of the book original 10 and the left and right side surfaces 10f and 10f '. The probability PLB that is a part is increased.
Here, the membership function of FIG. 23 has the characteristics reverse to those of the membership function of FIG. FIG. 24 is a graph showing the relationship between the character frequency m in the sub-scanning line of the book document 10 and the probability PM of the blank portion. In this case, if the character frequency m is small, it can be determined that there is a high possibility of a blank portion. FIG. 25 is a graph showing the relationship between the position X in the sub-scanning direction and the probability Px of a blank area. There is a high possibility that the content is written closer to the center of each of the left and right pages, and there is a high possibility that the content is not written at the binding portion, which is the central portion of the document, or the left and right end portions. The functions shown in FIGS. 22 to 25 are stored in the CPU 23.

FIG. 26 shows a distribution of the probability P.theta. Which is a blank portion obtained by applying the function of FIG. 22 to the height distribution of FIG. FIG. 27 shows the distribution of the probability PL, which is the margin portion, obtained by applying the function of FIG. 23 to the background luminance distribution of FIG.
FIG. 28 shows a distribution of the probability PM which is a blank part obtained by applying the function of FIG. 24 to the character pixel number distribution of FIG.

Probabilities Pθ, PL, PM of FIGS. 26 to 28
When the total probability P of the blank portion at each position of the document is obtained by combining the probabilities Px of FIG. The broken line shown in FIG. 29 is used as a threshold value for dividing a necessary part / margin part in the document. Since it is necessary to faithfully reproduce the image in the portion having a low probability of being a blank portion, the image data is sequentially written in the buffer memory 29, and the image data is corrected in accordance with the distortion correction amount in the sub-scanning direction obtained from the height of the document. The image distortion is corrected by repeatedly reading. On the other hand, since it is not necessary to faithfully reproduce the image in the portion having a high probability of being the blank portion, the writing of the image data to the buffer memory 29 is stopped, and the white data is repeatedly read to correct the image distortion. The white data is prepared in the CPU 23 in advance.

FIG. 30 is a block diagram for correcting image distortion in the image reading apparatus of this embodiment. The output signal of the line sensor 7 is digitally converted by the A / D converter 21 and then input to the serial / parallel converter 35, and the serial output signal is converted into a parallel output signal having a predetermined number of pixels as one block. To be done. The image data converted into the parallel output signal is written in the buffer memory 29 according to the instructions of the CPU 23 and the address controller 37. The written image data is CP
According to an instruction from U23 and the address controller 37, a predetermined number of pixels are read from the buffer memory 29 at the same time. The read image data is converted from a parallel output signal to a serial output signal by the parallel / serial converter 36 and sent to the image processing apparatus at the subsequent stage. Next, the buffer memory 29 will be described. Cereal/
The image data sent from the parallel converter 35 is read by the line sensor 7 in the buffer memory 29.
It is stored in line units. The buffer memory 29 shown in FIG.
The conceptual diagram of is shown. The buffer memory 29 has a capacity capable of storing n rows read by the line sensor 7 in units of one line. The first line of the image data is stored in the first line of the memory chip of the buffer memory 29, and
The line proceeds to the second line, the third line to the third line, and when the last line is stored, it returns to the first column and n +
The first line is stored in the first line. Buffer memory 29
The reading of the image data from is performed in the unit of one line as in the case of storing the image data. When correcting image distortion in the sub-scanning direction, the CPU 23 issues an instruction to the address controller 37 to repeatedly read the same line in the buffer memory 29. Thereby, the image in the sub-scanning direction can be expanded, and the image distorted by the inclination of the document surface can be corrected to be flat. The amount of image distortion that can be corrected by the configuration of this embodiment is determined by the total number of lines that can be expanded, and this number of lines is determined by the buffer memory 29.
It depends on the storage capacity that can be written to. The output switching means 39 selectively outputs either the white data input from the white data generating section 38 or the image data read from the buffer memory 29 and input by a command from the CPU 23. is there.

In this embodiment, for an image having a low probability of being a necessary information portion, writing to the buffer memory is stopped, white data from the white data generating section 38 is selected, and
Outputting corrects image distortion. Therefore, the buffer memory 29 can be effectively used, and the distortion correction amount in the sub-scanning direction can be increased.

The present invention can be variously modified without being limited to the configurations of the above embodiments. For example, the background luminance information of a document is mainly for detecting the boundary between the surface and the end face of the document. However, instead of using the base luminance, it is also possible to detect the boundary between the front surface and the end surface of the document by using the inflection point of the document height. As shown in FIG. 31, the height distribution on the surface of the document changes smoothly on the surface of the document, but discontinuity occurs at the boundary with the end face of the document. Since this boundary portion substantially coincides with the place where the background luminance distribution in FIG. 11 falls below the threshold value TL, by detecting this discontinuous portion, the probability distribution of the probability of the blank space corresponding to FIG. 16 can be obtained. . FIG. 31 shows how the probability distribution of the probability of being a blank is obtained by using the inflection point of the document height. In FIGS. 9A and 9B, the probability of being a blank is set to 1 outside the discontinuity of the document height. Further, as a method for measuring the document height, a method of irradiating a light beam as disclosed in JP-A-3-117965 and JP-A-5-316 are disclosed.
Although a method using the edge of the document surface as disclosed in Japanese Patent No. 302 is conceivable, the method is not particularly limited to these methods.

[0045]

As described above, according to the present invention, in an image reading apparatus for reading an original by optical scanning, the margins on the left and right of the original are determined based on the information of the left and right positions of the original, the background luminance, and the character frequency. By detecting it, it is possible to accurately detect the margin portion, and it is possible to prevent the occurrence of a determination error that has occurred in a device that detects an unnecessary portion of the document end face by using the conventional skin color information, etc. At the position, it is possible to erase unnecessary images outside the margins on the left and right of the document. Furthermore, according to the present invention, in addition to the above-described detection of the margins on the left and right of the document, the binding at the center of the document such as a book and the margin between the lines of the document in which characters are written vertically are detected. As for the image corresponding to the blank area, by interpolating the white data without using the buffer memory, the image can be corrected by effectively using the buffer memory.

[Brief description of drawings]

FIG. 1 is an overall configuration diagram of an image reading apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing how a book document is placed on an image reading apparatus.

FIG. 3 is a schematic configuration diagram of the image reading device as seen from the front.

FIG. 4 is a schematic configuration diagram of the image reading device viewed from the side.

FIG. 5 is a diagram showing the principle of height detection according to the present embodiment.

FIG. 6 is a diagram showing a state of image data read by an image reading device.

FIG. 7 is a diagram showing an output example of one line in the main scanning direction read by a sensor.

FIG. 8 is a diagram showing a state of image data when fingers are placed on a document.

FIG. 9 is a diagram showing a histogram for one line in the main scanning direction read by a sensor.

FIG. 10 is a diagram showing height distribution data obtained by height detecting means.

FIG. 11 is a diagram showing a background luminance distribution obtained from a histogram.

FIG. 12 is a diagram showing a character pixel number distribution obtained from a histogram.

FIG. 13 is a diagram showing a membership function that represents the probability of a blank space to be erased with respect to the original background luminance.

FIG. 14 is a diagram showing a membership function representing a probability that a blank space should be erased with respect to a character frequency.

FIG. 15 is a diagram showing a membership function representing the probability of a blank space to be erased with respect to a sub-scanning position.

FIG. 16 is a diagram showing an output value obtained by multiplying a background luminance distribution by a membership function.

FIG. 17 is a diagram showing an output value obtained by multiplying a character frequency distribution by a membership function.

FIG. 18 is a diagram showing an output value obtained by multiplying a sub-scanning position by a membership function.

FIG. 19 is a diagram showing an output value obtained by multiplying the output values of FIGS.

FIG. 20 is a block diagram of a control circuit in the image reading apparatus of the present invention.

FIG. 21 is a diagram showing a control flowchart of the present invention.

FIG. 22 is a diagram showing a membership function used in an image reading apparatus according to another embodiment of the present invention, which represents a probability of a blank portion in a document image with respect to a document inclination angle.

FIG. 23 is a diagram showing a membership function representing the probability of a blank portion in a document image with respect to document background luminance.

FIG. 24 is a diagram showing a membership function that represents the probability of being a blank portion in a document image with respect to the character frequency.

FIG. 25 is a diagram showing a membership function that represents the probability of a blank portion in a document image with respect to a position in the sub-scanning direction.

FIG. 26 is a diagram showing output values obtained by multiplying document height distribution data by a membership function.

FIG. 27 is a diagram showing an output value obtained by multiplying original background luminance distribution data by a membership function.

FIG. 28 is a diagram showing an output value obtained by multiplying character pixel number distribution data by a membership function.

FIG. 29 is a diagram showing output values obtained by multiplying the output values of FIGS. 26, 27, and 28.

FIG. 30 is a block diagram of temporarily storing image data for image distortion correction.

FIG. 31 is a conceptual diagram of a buffer memory.

FIG. 32 is a diagram showing how a probability distribution that is a margin is obtained using an inflection point of the document height.

[Explanation of symbols]

1 Document Plate 2 Image Pickup Device 7 CCD Line Sensor 10 Book Document 23 CPU (Brightness Detecting Unit, Character Pixel Counting Unit, Position Detecting Unit) 29 Buffer Memory 30 Image Processing Circuit

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H04N 1/38 H04N 1/10

Claims (2)

(57) [Claims] 1. An image reading apparatus for reading a document placed on a document table by optical scanning, and an image sensor for reading the document, and a memory for temporarily storing luminance information of image data read by the image sensor. Means, position detection means for detecting position information in the left-right direction corresponding to the image data read by the image pickup device, and the background luminance of the document based on the histogram of the luminance information stored in the storage means. Luminance detecting means for detecting, counting means for counting the number of pixels which should be judged as a character portion as a character frequency based on the histogram of the luminance information stored in the storage means, the position information in the left-right direction of the document, the background luminance , And the margin part detecting means for detecting the margin parts on the left and right sides of the document based on each information of the character frequency and the center of the document in the left-right direction. An image reading apparatus, comprising: an image processing unit that erases an image outside the margin portion in the left-right direction detected by the margin portion detecting unit. 2. An image reading apparatus for reading a document such as a book set on a document table by optical scanning, an image sensor for reading the document, a height detecting means for detecting the height of the document, and storage means for temporarily storing the luminance information of the image data read by the imaging device, position detecting means for detecting position information of <br/> lateral direction corresponding to the image data read by the image pickup device, the storage Brightness detecting means for detecting the background brightness of the document based on the histogram of the brightness information stored in the means; counting means for counting the number of pixels of the character part based on the histogram of the brightness information stored in the storage means; the height of the document, the position information in the horizontal direction, and the margin detection means for detecting a blank portion of the document based on the background luminance, and the information of the number of pixels of the character portion, the excess And white data generating means for the image of the detected margin by section detecting means white data, the contraction in the sub-scanning direction of the document caused by the height variation of the document, the same row stored in the storage means and correction means for correcting by repeatedly reading the image information, the image of the detected margin by the margin portion detecting means
An image reading apparatus comprising: a blank area correction unit that replaces data with white data .