JPH08317195A - Image forming device - Google Patents

Abstract

PURPOSE: To provide the image processing unit in which an upside/downside of an original image are automatically discriminated CONSTITUTION: The upside/downside of an original in which an under line is drawn in a character area and in access to the character area are discriminated. An image signal equivalent to an image area by one page of original read by a CCD 1 is stored in a frame memory 14 under the control of a CPU 13, a histogram circuit 15 obtains accumulated signals in the horizontal and vertical scanning directions, detect a position of the character string area detected from the accumulated value and a position of a line segment (under line) and discriminates the upside/downside of the image of the original based on position relation of the both. The original image is divided into plural areas and weighting is applied to the result of upside/downside discrimination based on the information denoting the area position in the original image to which a detected line segment belongs so as to enhance the accuracy of discriminating the upside/downside of the original.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image forming apparatus, and more particularly to an image forming apparatus capable of automatically recognizing the top and bottom of a document.

[0002]

2. Description of the Related Art Recent image forming apparatuses record two or four originals fed by an automatic document feeder on one recording sheet, or collect the image-formed recording sheets. Then, there are provided various automatic processing functions such as those having a bookbinding function of binding by staple.

In order to effectively use the image forming apparatus having such an automatic processing function, it is important that the originals of a plurality of originals fed by the automatic original feeder are aligned. Even if you make a copy without realizing that the originals are mixed upside down, if you notice this, you will have to discard the created copy. In that case, a lot of time and money will be wasted.

Therefore, when an upside down or sideways orientation of an original image is automatically determined and an upside down or horizontal orientation is found, an image forming apparatus for rotating the original image in a normal orientation is provided. Proposed. For example, in the image forming apparatus disclosed in Japanese Patent Laid-Open No. 4-229763, a character image sample is selected from image data obtained by scanning a document, and this sample is compared with a separately prepared reference character. The top or bottom or the rotation angle of the original image is discriminated and the original image is rotated in a normal direction if correction is necessary.

[0005]

However, a sample of a character image is selected from the image data obtained by scanning the above-mentioned document, and this sample is compared with a separately prepared reference character to determine the top or bottom of the document. With the means to determine the rotation angle,
Even if a sample of a character image is extracted and processed, a high-level image processing for reading the character and comparing it with the reference character is required. Therefore, the image forming apparatus automatically determines whether the document image is upside down or sideways. Was added, the cost of the device was significantly increased. The present invention is intended to solve the above problems.

[0006]

SUMMARY OF THE INVENTION The present invention is to solve the above-mentioned problems, and it is an image signal generating means for scanning an original to generate an image signal, and an image signal corresponding to an image area for one page of the original. And an image forming apparatus for processing an image signal, wherein the image processing means is an image corresponding to an image area for one page of a document stored in the storage means. The position of the character string area is detected from the cumulative value in the scanning direction of the signal, and the position of the line segment vertically separated from the character string area is detected, and the position information of the detected character string area and the position information of the line segment are detected. It is characterized in that the top and bottom of the document image are determined based on the.

The image area for one page of the original document is divided into a plurality of areas and weighted in advance, and the image processing means sets the divided image areas in a vertical direction from the character string area. Each time a line segment distant from is detected, the weighting value given to the image area to which the detected line segment belongs is accumulated, and the top and bottom of the original image are finally determined based on the accumulated value. Can be accurately determined.

Further, the weighting given in advance to the plurality of divided image areas is such that the weighting of the peripheral area is high,
It is advisable to set the weight of the central area low.

[0009]

The top and bottom of the document image is determined for the document in which the underline is drawn close to the character string in addition to the character string. Position information of the character string area detected from the cumulative value in the scanning direction of the image signal corresponding to the image area of one page of the document stored in the storage means, and position information of the line segment vertically separated from the character string area. Are compared with each other, and the top and bottom of the document image are determined from the positional relationship between the two.

[0010]

Embodiments of the present invention will be described below.
First, in addition to the character string as shown in FIG. 3, the manuscript is targeted at an underline, that is, a line segment drawn close to the character string.

FIG. 1 is a digital color chart to which the present invention is applied.
It is a block diagram of the image processing circuit of the copying machine. A document illuminated by a light source (not shown) is scanned by a scanner, and the reflected light is incident on the CCD line sensor 1 and converted into an electric signal. The output electric signal is amplified by the amplifier 2 and further converted into a digital image signal by the A / D converter 3. The image signal is subjected to correction of unevenness of light distribution of the light source, variation of sensitivity of CCD pixel, etc. in the shade correction circuit 4,
It is output as proper image data.

The D / A converter 5 connected to the amplifier 2 is
A digital signal output from the CPU 13 for instructing the gain setting is converted into an analog signal to set the gain of the amplifier 2, and the D / A connected to the A / D converter 3 is used.
The converter 6 converts the digital signal indicating the reference voltage output from the CPU 13 into an analog signal, and the A / D converter 3
It is for setting to. Also, the memory 7 is
This is a memory for storing the initial data for correcting the data, which is stored at the time of initial adjustment.

Appropriate image data corrected by the shearing correction circuit 4 is converted by the log conversion circuit 8 from the reflectance data up to that time to density data. Then, MTF correction is further performed in the MTF correction circuit 9,
Magnification processing is performed in the magnification circuit 10 to the designated magnification,
It is γ corrected by the γ correction circuit 11 and output to the printer 12.

The image data subjected to the shade correction in the shade correction circuit 4 is once stored in the frame memory 14 and is output to the histogram circuit 15 in accordance with the instruction of the CPU 13 for histogram calculation. Be seen. The handling of the calculated histogram will be described later. The image data stored in the frame memory 14 is subjected to rotation processing by the rotation processing circuit 18 as necessary and then output to the log conversion circuit 8.

In the ROM 16 connected to the CPU 13,
Predetermined weighting data and constants such as the number of divisions when dividing an image area are stored, and the RAM 17 stores data such as image data being processed and the result of top-bottom discrimination of a document. -Used for temporary storage of data.

FIG. 2 is a flow chart of the main routine showing the overall flow of the image data processing executed by the CPU 13. A document as shown in FIG. 3 is set on the document table to start scanning, and the document image is read by the CCD line sensor 1. The obtained image data is stored in the frame memory 14 after A / D conversion and shearing correction (step P1).

Next, the character string direction is detected from the image data (step P2), and the line segment (underline) perpendicular to the character string direction of the character string area in which the character string direction is detected is detected. It is detected (step P3). When detecting the line segment, the original image is divided into a plurality of image areas to detect the line segment. This point will be described in detail later.

An initial value of 1 is set in the divided area number counter n,
An initial value 0 is set to a line segment counter 1 which counts the number of processed line segments (step P4). When a line segment is detected, if a line segment exists at the same address (that is, at the same position) in an adjacent area in one divided document area, it is determined that the line segment is a continuous line segment and two or more line segments are detected. Area of 1
(Step P5). The reason for synthesizing the areas is to increase the detection accuracy when detecting the distance (interval) between the line segment and the character string region to be executed next. This will be described in detail later.

The distance (interval) between the line segment and the character areas above and below the line segment is detected (step P6), and the top and bottom of the document are discriminated from the detected distance (interval) (step P7). -Save the data (step P8). further,
The top-bottom discrimination result is weighted by information (area position information) indicating the position in the original image of the image area to which the detected line segment belongs (step P9), and the data is saved (step P10). By weighting the top-bottom discrimination result, the top-bottom discrimination accuracy of the document image can be improved.

Since the processing for one line segment has been completed, 1 is added to the line segment counter 1 (step P1).
1) The presence / absence of another line segment in the divided area is determined (step P12), and if there is another line segment, the processes of steps P5 to P11 are repeated. If there is no other line segment in the determination of step P12, the divided area number counter n is incremented (step P13), and the end of processing for all divided areas N (N> n) is determined (step P14), and processing is performed. If is not finished, step P5
Return to When the processing is completed, the top and bottom of the document are finally discriminated based on the top and bottom discrimination results (step P1).
5), the process ends.

FIG. 4 is a flow chart of FIG.
2 is a flowchart of a character string direction detection subroutine shown as 2. First, a cumulative value of image data in the main scanning direction is calculated to create a histogram (step P21),
It is determined whether the shape of the created histogram matches a predetermined pattern set in advance, and the determination result is temporarily stored (step P22).

Similarly, in the sub-scanning direction, the cumulative value of the image data is calculated to create a histogram (step P
23), it is determined whether or not the shape matches a preset predetermined pattern, and the determination result is temporarily stored (step P24).

The judgment result temporarily stored is checked (step P25). If either the shape of the histogram in the main scanning direction or the shape of the histogram in the sub scanning direction matches the predetermined pattern, step P26.
Proceed to.

Then, it is determined whether or not the shape of the histogram in the main scanning direction matches a predetermined pattern. When the shapes in the main scanning direction match, it is determined that the characters are arranged in the main scanning direction. (Step P27), and Step P
If the main scanning direction does not match in the judgment of 26, it is judged that the characters are arranged in the sub scanning direction (step P28),
Return to the main routine.

If the determination in step P25 is negative,
There are cases where both the shape of the histogram in the main scanning direction and the shape of the histogram in the sub-scanning direction match the prescribed pattern, and where both do not match the prescribed pattern. Since the arrangement direction cannot be identified, it is processed as a photographic image or other image, and the top and bottom of the document are not identified.

FIG. 5 shows a step P of the flow chart of FIG.
22 and P24 is a subroutine flow chart for determining whether or not the shape of the histogram indicated by 22 and P24 matches a predetermined pattern set in advance, in the main scanning direction and the sub scanning direction. The determination processing is the same processing except that the values of the parameters Pn and La are different.

First, the image data in the main scanning direction and the image data in the sub scanning direction of the original image are accumulated to obtain a histogram a in the main scanning direction and a histogram b in the sub scanning direction as shown in FIG. Shall be obtained. FIG. 7 shows a histogram a in the main scanning direction and a histogram b in the sub scanning direction shown in FIG.
Shows that (1) is entered.

In the flow chart of FIG. 5, as shown in FIG. 7, the histogram in the main scanning direction (sub-scanning direction) exceeds the preset threshold level (1).
The number P of black dots is counted (step P31). Here, the threshold level (1) refers to a level for determining the character string direction.

The number P of peaks is a preset number Pn
It is determined whether (for example, 2) has been exceeded (step P3
2) If it exceeds, it is further determined whether or not the width L of the peak is less than or equal to a preset predetermined width La (step P
33). The number of peaks P exceeds a predetermined number Pn,
If the width L of the histogram is less than or equal to the predetermined width La, it is determined that the shape of the histogram matches the preset predetermined pattern (step P34), and the determination data is stored in the memory ( Step P35), return to main routine.

If the result of the determination in step P32 is that the number P of peaks does not exceed the predetermined number Pn, and if the result of the determination in step P33 is that the width L of the peak is greater than or equal to the predetermined width La,
The shape of the histogram is a preset pattern
(Step P36), the main rule
Return to Chin.

FIG. 8 shows a step P of the flow chart of FIG.
3 is a flowchart of the line segment detection subroutine shown as 3. First, referring to the detection result of the character string direction in the original image (step P2 of the flow chart in FIG. 2), it is determined whether the character string direction in the original image is the main scanning direction or the sub-scanning direction (step P41). ), The original image is divided into N areas in the main scanning direction in the main scanning direction, the original image is divided into N areas in the sub scanning direction in the sub scanning direction, and the area number n is set. (Steps P42, P43). For example, when the character string direction is the main scanning direction, the original image is divided into 10 areas in the main scanning direction as shown in FIG.

Here, the reason why the original image is divided into a plurality of areas is to enhance the detection accuracy of the line segment in the original image. That is, when the original image is not divided as shown in FIG. 10A, the created histogram is the histogram peak value of the character string area and the histogram peak of the line segment as shown in FIG. 10B. Although there is almost no difference from the value, it becomes difficult to detect the line segment. However, when the original image is divided as shown in FIG. 7C, the histogram created in the divided areas is as shown in FIG. The histogram area of the character string area
The difference between the peak value and the histogram peak value of the line segment becomes clear, and the line segment can be easily detected. The number of divisions of the image area can be arbitrarily set depending on the size of the original image, the size of the characters, and the like.

An initial value 1 is set in the area number counter (step P44), and it is determined whether or not the content of the area number counter is smaller than the area division number N, that is, whether or not the processing is completed for all areas ( Step P45).
When the processing is completed for all areas, the process returns to the main routine. However, when N ≧ n, that is, when the processing is not completed, the nth area data is called (step P46) and the image data is displayed in the character string direction. The histogram is accumulated to create a histogram (step P47).

It is judged whether or not there is a histogram whose threshold level exceeds the preset threshold level (2) (step P48). Where the threshold level
(2) refers to the level at which a line segment is detected. If there is more than level (2), save the area number and line segment address (steps P49 and P50), increment the area number n (step P51), and return to step P45 to see all areas. The process is repeated for.

FIG. 11 shows an example of the histogram a created for each divided area of the original image shown in FIG.
Is an enlarged view of the histogram a of area number 2. In FIG. 12, line segments 1, 2, and 3 exceeding the threshold level (2) are detected, and the addresses of these line segments are A1, A2, and A3.

Although the processing for each divided area is completed as described above, line segments detected in one area may continue to adjacent areas. In this case, the line segment detection accuracy is improved. In order to do this, area synthesis is performed.

FIG. 13 is a flow chart of the area synthesis subroutine shown as step P5 in the flow chart of FIG. First, the line segment data of the nth area is read (step P61), it is determined whether or not there is a line segment in the area (step P62), and if there is no line segment, the area number n is incremented. (Step P63), the process returns to Step P61. If there is a line segment in the area, the address Al of the line segment is read (step P64).

The area number n is replaced with m (step P6
5) The line segment data of the adjacent m + 1th area is read (step P66), and it is determined whether or not there is a line segment in the address Al (step P67). If there is a line segment in the address Al, the area number m is incremented (step P
68), and returns to step P66. If there is no line segment in the address Al in the determination of step P67, the areas n to m that have been determined to have the line segment in the previous determination are combined (step P6).
9) Return to the main routine.

FIG. 14 is a flowchart of the subroutine for detecting the distance between the line segment shown as step P6 in the flowchart of FIG. 2 and the character string area. Here, FIG.
As shown in FIG. 5, a document in which the area (2) and the area (3) are combined because there is a line segment in the address Al of the second area (2) and the third area (3) An example will be explained.

First, an address Al having a line segment and image data in the vicinity thereof are accumulated to create a histogram (step P71). Next, the rising edge and falling edge addresses Bl and Cl of the histogram of the character string before and after the address Al are detected (step P72), and the process returns to the main routine. The distance between the line segment and the character string area can be known from the difference between the detected addresses Bl and Cl.

FIG. 16 is a flowchart of a subroutine for discriminating the top and bottom of the original image shown as step P7 in the flowchart of FIG. Here, as before,
As shown in FIG. 5, the original image in which the area (2) and the area (3) are combined because there is a line segment in the address Al of the second area (2) and the third area (3). Will be described as an example.

First, the distance between the upper character string area of the original image and the line segment, that is, the absolute value | Bl of the difference between the address Bl of the trailing edge of the histogram of the character string in front of the line segment and the line segment address Al. -Al | is the distance between the character string region and the lower line segment, that is, the absolute value | Al-Cl | of the difference between the line segment address Al and the rising edge address Cl of the histogram of the character string behind the line segment. Is determined (step P81), and if the absolute values of the differences are equal, it means that the upper and lower line segments are equidistant from the character string, so it is determined that the top and bottom of the original image cannot be determined. (Step P85), main rule
Return to Chin.

If the absolute values of the differences are not equal in the determination of step P81, the distance between the upper character string region and the line segment, that is, the address Bl of the trailing edge of the histogram of the character string in front of the line segment and the line segment. Absolute value of difference from address Al | Bl
-Al | is the distance between the character string region and the lower line segment, that is, the absolute value of the difference between the line segment address Al and the address Cl of the rising edge of the histogram of the character string behind the line segment | Al-
It is determined whether or not it is smaller than Cl | (step P8
2).

As a result of the above judgment, | Bl-Al | <| A
In the case of l-Cl |, it means that the lower line segment is close to the character string as shown in FIG. 15, so that the address Bl side of the trailing edge of the histogram of the character string is at the top of the original image ( Upper) (step P83), again | Bl
When -Al | <| Al-Cl | means that the upper line segment is close to the character string, the address Cl side of the rising edge of the histogram of the character string is the top (upper) of the original image. The determination is made (step P84).

After the top-bottom discrimination of the original image is completed based on the address information in the vicinity of the line segment, the top-bottom recognition information determined for the line segment (address Al) is weighted and used for the final top-bottom discrimination. To do. For weighting the top-bottom recognition information about the line segment, weighting information is given in advance to each area of the document image divided into a plurality of areas, and the weighting information is used.

In this embodiment, as shown in FIG. 18, focusing on the fact that an underline is often drawn in the headline in the text of the manuscript image, the area at the center of the manuscript image is low and Areas are given high weighting information. In addition to this, as shown in FIG. 19, for example, appropriate weighting effective for recognition of the top and bottom of the document image can be performed, for example, by assigning weighting information to the area around the document image surface to be high and to the central area to be low. is there.

FIG. 17 is a flowchart of the weighted subroutine shown as step P9 in the flowchart of FIG.
It is First, it is determined whether or not the target line segment extends over a plurality of areas (step P86), and if the line segment extends over a plurality of areas, the weighting information given to each area is averaged. The value is used as a weighting value (step P87). If it is a line segment of a single area, the weighting information given to the corresponding area is used as the weighting value (step P88).

By the above processing, the top and bottom of the original image are recognized from the relationship between all the line segments and the character strings, and the weighting value is determined. Therefore, the final top and bottom of the original image is determined from these information. .

FIG. 20 is a subroutine for the final top-bottom discrimination of the original image shown as step P15 in the flowchart of FIG.
It is a flowchart of chin. Here, the document image will be described with reference to the example shown in FIG. The number of line segments is counted by the line segment counter 1 (see step P11 of the flow chart in FIG. 2), and it is assumed that there are five line segments in total.

First, initial values are set in the counters k, b and c (step P101), and the weighting value W of the kth line segment is set.
k is read (step P102). Furthermore, the determination result of the top and bottom around the kth line segment is read out (step P103),
If the B side (explaining in FIG. 15 side of the falling edge of the character string histogram at the Bl address) is the top (upper),
The weighting value Wk is added to the counter b (step P10
4), if the C side (the address Cl side of the rising edge of the histogram of the character string in the case of FIG. 15) is the top (upper), the weighting value Wk is added to the counter c (step P105), Return to the main routine.

The presence or absence of the next unprocessed line segment is determined by the contents of the counter k (step P106). If there is an unprocessed line segment, the counter k is incremented (step P10).
7) The procedure returns to step P102 and is repeated until the processing for all line segments (five) is completed.

When the processing for all the line segments is completed, the contents of the counter b and the contents of the counter c are compared (step P108), and b ≠ c, that is, the total value of the weighting on the B side and the weighting on the C side. If the sum of weights is not equal, b
It is determined whether or not> c (step P109).

If b> c, that is, the sum of the weighting values on the B side is large in the judgment of step P109, the B side is moved up (up).
(Step P110), if b <c, that is, if the total value of the weighting on the C side is large, the C side is determined to be up (up) (step P111), and the process returns to the main routine.

Further, in the judgment of step P108, b =
If c is the same as the total weight of the B side and the total weight of the C side, it is determined that the final top-bottom determination of the original image is impossible (step P112), and the process is terminated.

In the original image shown in FIG. 21, line segments (5 lines)
Is determined to be the sky (up), the B side is also determined to be sky (up) even in the total value of weighting.

As described above, according to the present invention, the top and bottom of the original image can be discriminated from the distance between the character string area of the original image and the line segment. Therefore, when processing a plurality of originals, the first original is processed. The top of the original image based on the result of scanning
The ground (bottom) is determined, and when a document image in the opposite direction to the determined top (top) and ground (bottom) is found, the document image may be rotated and output. Or, the top (top) and bottom (bottom) of the original image are determined in advance, and as a result of reading the original,
When a document image in a direction opposite to the previously determined top (upper) ground (lower) is found, the document image may be rotated and output.

Further, in the image forming apparatus not provided with the function of rotating the document image, a warning is issued when the document image in the opposite direction to the determined top (upper) ground (lower) is found. Good to do.

[0058]

As described above, according to the present invention, the position of the character string area is detected from the cumulative value in the scanning direction of the image signal corresponding to the image area of one page of the original stored in the storage means. The vertical position of the original image is determined by detecting the position of the line segment vertically separated from the character string region and determining the top and bottom of the original image from the positional relationship between the detected character string region and the line segment. It does not require high-level image processing such as comparing the sample of the character image obtained by scanning the document with a reference character prepared separately like the method of discrimination, and the top and bottom of the document image can be increased with a slight cost increase. The determination can be performed easily and accurately.

[Brief description of drawings]

FIG. 1 is a block diagram of an image processing circuit of a digital color copying machine to which the present invention is applied.

FIG. 2 is a main routine flow chart showing the overall flow of data processing.

FIG. 3 is a diagram illustrating a document.

FIG. 4 is a flowchart of a character string direction detection subroutine.
To.

FIG. 5 is a subroutine flowchart for determining the shape of a histogram.

FIG. 6 is a diagram illustrating a histogram in a main scanning direction and a histogram in a sub scanning direction.

FIG. 7 is a diagram for explaining a relationship between a histogram and a threshold level (1).

FIG. 8: Flow chart of line detection subroutine.

FIG. 9 is a diagram illustrating an example in which a document image is divided into a plurality of areas in the main scanning direction.

FIG. 10 is a diagram illustrating detection of line segments when a document image is not divided and when it is divided.

11 is a diagram illustrating an example of a histogram created based on the document image of FIG.

FIG. 12 is a diagram illustrating the relationship between a histogram and a threshold level (2).

FIG. 13: Float chart of area synthesis subroutine

FIG. 14 is a subroutine for detecting the distance between a line segment and a character area.
Chin's flow chart.

FIG. 15 is a diagram illustrating a document image in which area (2) and area (3) are combined.

FIG. 16 is a flowchart of a subroutine that determines the top and bottom of a document image.

FIG. 17: Flow chart of weighted subroutine.

FIG. 18 is a diagram illustrating an example of weighting information added to a document image area.

FIG. 19 is a diagram illustrating another example of weighting information given to a document image area.

FIG. 20 is a flowchart of a subroutine for final upside-down determination of an original image.

FIG. 21 is a diagram illustrating an example of a document image.

[Explanation of symbols]

 1 CCD 4 Sheading correction circuit 7 Memory 8 log conversion circuit 9 MTF correction circuit 10 Magnification circuit 11 γ correction circuit 12 Printer 13 CPU 14 Frame memory 15 Histogram circuit 18 Rotation processing circuit

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kazuomi Sakata 2-3-3 Azuchi-cho, Chuo-ku, Osaka-shi, Osaka, Osaka International Building Minolta Co., Ltd. (72) Inventor Shoji Imaizumi Azuchi, Chuo-ku, Osaka-shi, Osaka 2-13-3 Machi Osaka International Building Minolta Co., Ltd.

Claims (3)

[Claims] 1. An image signal generating means for scanning an original to generate an image signal, a storage means for storing an image signal corresponding to an image area for one page of the original, and an image processing means for processing the image signal. In the image forming apparatus, the image processing means includes one original document stored in the storage means.
The position of the character string region is detected from the cumulative value in the scanning direction of the image signal corresponding to the image region of the image portion, and the position of the line segment vertically separated from the character string region is detected, and the detected character string region is detected. An image forming apparatus that determines the top and bottom of a document image based on the position information of the line segment and the position information of the line segment. 2. The image area for one page of the original is divided into a plurality of image areas in advance and weighted, and the image processing means separates the character string area from the character string area in each divided image area. Each time a line segment that is distant in the direction is detected, the weighting value given to the image area to which the detected line segment belongs is accumulated, and the top and bottom of the original image are finally determined based on the accumulated value. The image forming apparatus according to claim 1, wherein: 3. The weights given in advance to the plurality of divided image areas are set such that a peripheral area has a high weight and a central area has a low weight. 2. The image forming apparatus according to 2.