JP2008199155A - Image reading apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve detection accuracy of a code or the like based on image data obtained by reading out an inclined document. <P>SOLUTION: A line sensor 55 reads out an image from the reverse side of a document 1 and a CCD image sensor 78 reads out an image from the front side of the document 1 on the document carrying direction downstream side of the line sensor 55. A skew quantity detection part 90 detects the skew quantity of the document 1 on the basis of second image data read out by the line sensor 55 and outputs the obtained skew quantity detection result to an image feature detection part 100. The image feature detection part 100 detects codes included in first image data read out by the CCD image sensor 78. Then, the image feature detection part 100 reads out a template of a code with a corresponding skew angle on the basis of the inputted skew quantity detection result and compares the read template with first code image data out of the first image data to detect the code. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an image reading apparatus that reads an image of a document.

  Conventionally, a reading device that reads an image formed on a document while conveying the document is known. In addition, a first reading unit that reads an image on one side (front side) of the original and a second reading unit that reads an image on the other side (back side) are provided in the conveyance path of the original, and the original is conveyed once. There is also a double-sided reading device that reads images formed on both sides of the original.

  Further, for example, identification information that is difficult to be visually recognized by human eyes is added to an image formed on a document, and the identification information (for example, a code) is read together with the image on the document by a reading device. A technique has been proposed in which whether or not an image can be copied is determined based on the above (see Patent Document 1).

  By the way, in the above-described double-sided reading apparatus, when the original is conveyed in an inclined state, the image on the front surface and the image on the back surface to be read are also inclined. Also, as in Patent Document 1, when identification information is added to an original together with an image, the identification information to be read also tilts, making it difficult to determine the content of the identification information. Therefore, a technique for detecting the skew amount on the front side of the document based on the image data on the front surface by the first reading unit and detecting the skew amount on the back side of the document based on the image data on the back side by the second reading unit. Has been proposed (see, for example, Patent Document 2).

JP-A-4-367082 JP 2000-349980 A

  The present invention has been made against the background of the above-described technology, and its object is to improve the detection accuracy of codes and the like based on image data obtained by reading a tilted document with a simpler configuration. There is.

  For this purpose, an image reading apparatus to which the present invention is applied includes a first sensor that outputs first image data obtained by reading an image on a first surface of a conveyed document, A second sensor that outputs the second image data obtained by reading the image on the second side of the document upstream of the sensor in the document transport direction, and detects the tilt of the document based on the second image data And a code for detecting the first code included in the first image data and detecting the second code included in the second image data based on the inclination detection result by the inclination detection unit. And a detector.

  In such an image reading apparatus, the code detection unit uses the inclination detection result with the polarity reversed when detecting the first code, and uses the inclination detection result as it is when detecting the second code. Can be a feature. The code detection unit further includes a selection unit that selectively outputs the first image data or the second image data. The selection unit outputs the second image data after outputting the first image data in the same document. Image data can be output. Further, the first image data can be output with the first code added, or the second image data can be output with the second code added.

  From another point of view, an image reading apparatus to which the present invention is applied includes a first reading unit that reads an image on the first side of the conveyed document and an image on the second side of the conveyed document. Second reading means for reading, inclination detecting means for detecting the inclination of the document based on the second image data obtained by reading the second surface by the second reading means, and inclination detection result by the inclination detecting means And feature detection means for detecting the feature of the first image data obtained by reading the first surface by the first reading means.

  In such an image reading apparatus, the feature detection unit can detect a code included in the first image data as a feature of the first image data. In this case, it further includes storage means for storing a plurality of templates with different inclination angles of the code, and the feature detection means reads the corresponding template from the storage means based on the inclination detection result by the inclination detection means, A code can be detected from a comparison result between the image data and the read template. Further, the feature detection means can be characterized in that the sign of the inclination detection result by the inclination detection means is reversed. Furthermore, the feature detection unit can further detect the feature of the second image data using the tilt detection result by the tilt detection unit.

According to the first aspect of the present invention, it is possible to improve the code detection accuracy based on the image data obtained by reading the inclined original.
According to the second aspect of the invention, it is possible to more accurately grasp the inclination of the front and back surfaces of the document.
According to the third aspect of the present invention, it is possible to perform processing for image data one by one.
According to the fourth aspect of the present invention, for example, in the subsequent device, it is possible to perform appropriate processing on the image data of each surface.
According to the fifth aspect of the present invention, it is possible to improve the accuracy of feature detection based on image data obtained by reading an inclined document.
According to the sixth aspect of the present invention, it is possible to improve the code detection accuracy based on the image data.
According to the seventh aspect of the present invention, since it is not necessary to correct the tilt angle with respect to the image data itself, for example, a necessary memory capacity can be reduced.
According to the eighth aspect of the present invention, it is possible to more accurately grasp the inclination of the document.
According to the ninth aspect of the present invention, it is possible to detect characteristics of image data on both sides of a document.

The best mode for carrying out the present invention (hereinafter referred to as “embodiment”) will be described below with reference to the accompanying drawings.
FIG. 1 is a diagram showing an overall configuration of an image reading apparatus to which the present embodiment is applied. The image reading apparatus includes a document feeding device 10 that sequentially conveys the document 1 from a stacked document bundle, and a scanner device 70 that reads an image by scanning.

  The document feeder 10 includes a document storage unit 11 on which a bundle of documents composed of a plurality of documents 1 is stacked, and a paper discharge storage unit 12 that is provided below the document storage unit 11 and loads a document 1 that has been read. In the present embodiment, the original 1 loaded on the original accommodating portion 11 is set so that the first surface (hereinafter referred to as the front surface) faces upward. The document feeder 10 also includes a feeding roll 13 that takes out the document 1 from the document storage unit 11 and conveys it. Further, on the downstream side of the feeding roll 13 in the document conveyance direction, a mechanism 14 for separating sheets one by one is provided. The separating mechanism 14 includes a feed roll 14a that conveys the document 1 supplied by the feeding roll 13 further downstream, and a retard roll 14b that separates the documents 1 supplied by the feeding roll 13 one by one. A registration roll 16, a conveyance roll 17, a platen roll 18, an out roll 19, and a discharge roll 20 are provided in the conveyance path 15 where the document 1 is conveyed in order from the upstream side in the document conveyance direction. The registration roll 16 rotates and temporarily stops, then restarts at the same timing, and supplies the document 1 while performing registration adjustment on a document reading system described later. The conveyance roll 17 further conveys the conveyed document 1 toward the platen roll 18. The platen roll 18 assists the conveyance of the document 1 being read by the scanner device 70. The out-roll 19 conveys the document 1 read by the scanner device 70 further downstream. The discharge roll 20 further conveys the read original 1 and discharges it to the paper discharge storage unit 12.

Further, as shown in FIG. 1, a CIS (Contact Image Sensor) 50 is provided between the registration roll 16 and the conveyance roll 17 of the document feeder 10. The CIS 50 includes a light source and a lens (not shown), and a line sensor 55 that functions as a second sensor or a second reading unit.
Since the CIS 50 is provided inside the conveyance path 15, it can read an image on the second side (hereinafter referred to as the back side) of the document 1. As the line sensor 55, a CCD, a CMOS sensor, a contact type sensor, or the like can be used, and an image with an actual width (for example, A4 longitudinal width 297 mm) can be read.

  The scanner device 70 supports the document feeding device 10 described above so that it can be opened and closed, supports the document feeding device 10 with a device frame 71, and reads an image on the surface of the document 1 conveyed by the document feeding device 10. ing. Since the scanner device 70 is provided outside the conveyance path 15, it can read the first surface of the document 1. The scanner device 70 also reads a device frame 71 forming a casing, a first platen glass 72A on which a document 1 to be read is placed in a stationary state, and a document 1 conveyed by the document feeder 10. 2nd platen glass 72B which has the opening part of this light.

  Further, the scanner device 70 is stationary under the second platen glass 72B or scans the entire first platen glass 72A to read an image, and the light obtained from the full rate carriage 73 is input to the imaging unit. A half-rate carriage 75 for supplying is provided. The full rate carriage 73 is provided with an illumination lamp 74 for irradiating light on the document 1 and a first mirror 76A for receiving reflected light obtained from the document 1. Further, the half-rate carriage 75 is provided with a second mirror 76B and a third mirror 76C that provide the light obtained from the first mirror 76A to the imaging unit. The scanner device 70 also includes a CCD (Charge Coupled Device) image sensor 78 and an imaging lens 77. Among these, the imaging lens 77 optically reduces the optical image obtained from the third mirror 76C. The CCD image sensor 78 functioning as the first sensor or the first reading means photoelectrically converts the optical image formed by the imaging lens 77. That is, the scanner device 70 forms an image on the CCD image sensor 78 using a so-called reduction optical system.

  The scanner device 70 further includes a control unit 79 and an image processing unit 80. The control unit 79 controls the operation of each unit in the image reading operation. The image processing unit 80 processes the read image data. These functions are realized by a CPU or the like controlled by a program. The control unit 79 controls the operations of the document feeder 10 and the scanner device 70 described above. The control unit 79 controls the motor of the document feeder 10, various roll operations, the operation of the feed clutch, and the like, and the image data capturing operation by the line sensor 55. Further, the control unit 79 controls an image data capturing operation and the like by the CCD image sensor 78 in the scanner device 70. The image processing unit 80 receives the image data of the document 1 read by the CCD image sensor 78 and the line sensor 55 and outputs the image data to a subsequent device that is an output target. Examples of the subsequent device to be output include a host system such as a printer or a personal computer (PC).

  With the above-described configuration, the document reading apparatus is configured to fix and read the document 1 on the first platen glass 72 </ b> A and to transport the document 1 using the document feeder 10. It is possible to operate in a reading and reading mode. Further, this document reading apparatus is configured to use the single-sided reading mode for reading an image formed on the surface of the document 1 using the CCD image sensor 78 in the conveyance reading mode, and the document 1 using the CCD image sensor 78 and the line sensor 55. It is possible to operate in a double-sided simultaneous reading mode for reading images formed on both sides.

First, image reading of the document 1 in the fixed reading mode will be described.
In the fixed reading mode, the document 1 is set on the first platen glass 72A with the surface to be read facing down. When reading is started, the full rate carriage 73 and the half rate carriage 75 move in the scanning direction (arrow direction) at a ratio of 2: 1. At this time, the light of the illumination lamp 74 of the full rate carriage 73 is applied to the surface to be read of the document 1. Then, the reflected light from the document 1 is reflected in the order of the first mirror 76 A, the second mirror 76 B, and the third mirror 76 C and guided to the imaging lens 77. The light guided to the imaging lens 77 forms an image on the light receiving surface of the CCD image sensor 78. The CCD image sensor 78 is a one-dimensional sensor and processes one line at the same time. The full-rate carriage 73 and the half-rate carriage 75 are moved in this line direction (scanning sub-scanning direction), and the next line of the document 1 is read. By executing this over the entire document 1, the reading of one page of the document is completed.

Next, image reading of the document 1 in the conveyance reading mode will be described.
First, image reading of the document 1 in the single-sided reading mode will be described.
When the document 1 transported through the transport path 15 from the document storage unit 11 passes over the second platen glass 72B, the full rate carriage 73 and the half rate carriage 75 are stopped at the solid line positions shown in FIG. I'm left. Then, the reflected light of the first line of the document 1 that has passed through the platen roll 18 of the document feeder 10 is imaged by the imaging lens 77 via the first mirror 76A, the second mirror 76B, and the third mirror 76C. After one line in the main scanning direction is simultaneously processed by the CCD image sensor 78, one line in the next main scanning direction of the document 1 conveyed by the document feeder 10 is read. After the leading edge of the document 1 reaches the reading position of the second platen glass 72B, the trailing edge of the document 1 passes through the reading position of the second platen glass 72B, thereby completing one page of document reading in the sub-scanning direction. To do.

Next, image reading of the document 1 in the double-sided simultaneous reading mode will be described.
In the double-sided simultaneous reading mode, the full-rate carriage 73 and the half-rate carriage 75 are stopped at the solid line positions shown in FIG. 1, and the surface of the document 1 is read by the CCD image sensor 78 through the second platen glass 72B. The back side of the document 1 is read by the sensor 55. At this time, the surface of the document 1 is read by the CCD image sensor 78 in the same process as in the single-sided reading mode described above. On the other hand, the back side of the document 1 is read by the CIS 50 as follows. In the CIS 50, first, a light source irradiates the back surface of the document 1 with light. The lens collects reflected light of the light irradiated on the back surface of the document 1. The line sensor 55 reads the light collected by the lens. As described above, after one line in the main scanning direction is simultaneously processed by the line sensor 55, one line in the next main scanning direction of the document 1 conveyed by the document feeder 10 is read. After the leading edge of the document 1 reaches the reading position of the CIS 50, the trailing edge of the document 1 passes through the reading position of the CIS 50, thereby completing one page of document reading in the sub-scanning direction. Note that “simultaneous” in double-sided simultaneous reading does not completely coincide with time, but is defined here as meaning about the same time of document conveyance.

  In the present embodiment, as shown in FIG. 1, the reading position of the back surface of the document 1 by the line sensor 55 is shifted from the reading position of the front surface of the document 1 by the CCD image sensor 78 to the upstream side in the document transport direction. ing. Therefore, in the double-sided simultaneous reading mode, the document 1 supplied one by one from the document storage unit 11 at a predetermined timing is read first by the line sensor 55 and then read by the CCD image sensor 78. become. In this manner, the images of the plurality of documents 1 stacked on the document storage unit 11 are read in the order of the back surface and the front surface of the document 1. When only one document is set in the document storage unit 11, the image is read once for each of the front and back surfaces.

  By the way, in recent years, for the purpose of preventing forgery and document management, when printing on a paper document, a technique of printing a code image together with a document image such as a text or image to be printed has been used. I'm starting. For example, if there is a request to prohibit copying of a paper medium original document or to add ID information to the original document, a predetermined code image is embedded in the original document together with the document image. When a document processed in this way is read by a scanner device such as a scanner, a code image is also read simultaneously with a document image. For example, when a specific code having a copy prohibition instruction is detected from the read code image, the read data of the original is not output or the content of the document image is determined in the process of the subsequent stage. It is possible to take measures such as printing in black so that there is no problem.

  The image reading apparatus in the present embodiment can also read a document on which such a code image is printed. For example, the CCD image sensor 78 reads the surface of the document 1 and thereby includes first image data including first document image data that is read data of a document image and first code image data that is read data of a code image. To get. Further, for example, the line sensor 55 reads the back surface of the document 1, whereby the second image including the second document image data that is the read data of the document image and the second code image data that is the read data of the code image. Get the data. The code image is formed in a color that is difficult for human eyes to distinguish, such as yellow.

Next, an example of a code image embedded in the document 1 will be described. The code image is formed by an array of bit patterns expressed by a plurality of minute line bitmaps having different inclination angles.
FIG. 2 is a diagram for explaining a code image. FIG. 2 (a) shows a bit pattern A represented by a slash ("/") and a bit represented by a backslash ("\") on a 12 pixel by 24 pixel grid with 1 pixel as a unit. Pattern B is displayed side by side. Bit pattern A represents bit value 1 and bit pattern B represents bit value 0. Therefore, 1-bit information (0 or 1) can be expressed with one bit pattern as a minimum unit. Arbitrary information can be embedded in the document 1 by forming a code by combining the bit patterns as described above.
In the present embodiment, a code image is formed by arranging a plurality of codes in which bit patterns as described above are arranged over the entire surface of the document 1. A copy-prohibited code image is configured by appropriately combining these bit patterns.

  By the way, when a document is read in the above-described transport reading mode, a so-called skew (skew) may be generated in which the document is transported while being tilted. Skew is most likely to occur when a document is drawn into the conveyance path from a storage unit on which a bundle of documents is stacked. Therefore, in general, an automatic document feeder is provided with a mechanism for correcting skew in the conveyance path (hereinafter referred to as skew correction mechanism). Taking the document feeder 10 in the present embodiment as an example, a skew may occur when the document 1 is drawn from the feed roll 13 into the conveyance path 15. Therefore, a registration roll 16 that also functions as a skew correction mechanism is provided to correct the generated skew. When the leading edge of the document 1 comes into contact with the registration roll 16 at the timing when the rotation of the registration roll 16 is once stopped, the document 1 is formed in a loop shape. As a result, the edge on the leading end side in the conveyance direction of the document 1 is aligned so as to be substantially perpendicular to the conveyance path 15. Therefore, the skew of the original 1 is eliminated.

However, the skew may not be completely eliminated even by the skew correction mechanism. When the original is read without the skew being eliminated in this way, a document image that is not inclined on the original to be read is output in an inclined state after being read. Furthermore, when a code image as described above is embedded in a document, the occurrence of skew causes a decrease in detection accuracy when a bit pattern is detected from code image data.
FIG. 2B shows a state in which the code (bit pattern) is output tilted as a result of image reading while the document 1 is skewed. As described above, if the code image is read at an angle different from the original angle (the document is not tilted), an erroneous pattern may be recognized when the bit pattern is detected from the code image. In some cases, the bit pattern cannot be detected from the code image.

<Embodiment 1>
In the first embodiment, the following configuration is used to accurately detect the bit pattern embedded in the document 1 even when the document 1 is read in a skewed state.
The first embodiment is based on the premise that code images having the same contents are printed on the front surface and the back surface of the document 1. Therefore, it is sufficient to detect the bit pattern from the code image on either side of the document 1.

FIG. 3 shows a configuration of the image processing unit 80 in the present embodiment. The image processing unit 80 includes a skew amount detection unit 90 and an image feature detection unit 100.
A skew amount detection unit 90 that functions as an inclination detection unit or an inclination detection unit detects the skew amount of the document 1 based on the second image data input from the line sensor 55. The skew amount detection unit 90 can output the skew amount of the document 1 as distance and angle data. In the present embodiment, the skew amount detection unit 90 detects the tilt angle (skew amount) of the document 1 based on the second image data, and passes the angle data to the image feature detection unit 100. Therefore, in this embodiment, the skew amount is handled as angle data, and the following description will be given.

  By the way, it has been empirically obtained that an original document skewed in the conveyance path is conveyed while maintaining the skew amount, that is, the inclination angle. For this reason, in the present embodiment, the skew amount is not detected for each of the first image data corresponding to the front side of the document and the second image data corresponding to the back side of the document. Is detected, and the obtained skew amount is applied to the front surface and the back surface of the document 1, respectively. In the present embodiment, as described above, the same original 1 is read by the line sensor 55 before the CCD image sensor 78. For this reason, in this embodiment, the skew amount detection unit 90 is configured to detect the skew amount based on the second image data by the line sensor 55.

  The image feature detection unit 100 functioning as a feature detection unit is configured to select the first of the first image data read by the CCD image sensor 78 based on the skew amount of the document 1 obtained from the skew amount detection unit 90. Bit pattern detection from code image data. At this time, the skew amount received by the image feature detection unit 100 from the skew amount detection unit 90 is an inclination angle obtained from the back side of the document 1. In the present embodiment, the skew amount of the document surface is obtained based on the skew amount obtained from the second image data that is the reading result of the back side of the document, and the process for the first image data that is the reading result of the document surface is performed. Is going.

  Here, when the original viewed from the front side is tilted (rotated) in the + (positive) direction with the top and bottom of the original fixed, when viewed from the back side, the negative (−) direction. (Hereinafter, the counterclockwise rotation is defined as + (positive) and the clockwise rotation is defined as-(negative)). The same applies to the relationship between the line sensor 55 that reads the back surface and the CCD image sensor 78 that reads the front surface of the document 1 conveyed on the conveyance path 15. That is, since the document 1 is conveyed on the conveyance path 15 with the top and bottom, that is, the front end and the rear end fixed, the back side read image of the document 1 read by the line sensor 55 is skewed in the positive direction. Therefore, the surface reading image of the document 1 read by the CCD image sensor 78 is skewed in the negative direction. From this, the skew amount of the back side read image of the document 1 and the skew amount of the front side read image have the same absolute value, but have a relationship in which the positive and negative signs are reversed with respect to the inclination direction. Therefore, the image feature detection unit 100 obtains the surface skew amount by reversing the sign of the skew amount (angle) received from the skew amount detection unit 90.

FIG. 4 is a diagram for explaining the configuration of the image feature detection unit 100 and the data stored in the image feature detection unit 100.
FIG. 4A shows the configuration of the image feature detection unit 100. As illustrated in FIG. 4A, the image feature detection unit 100 includes a binarization processing unit 101, a database 102, and a comparison unit 103.

  The binarization processing unit 101 performs binarization processing on first code image data composed of multi-value data read by the CCD image sensor 78. The binarization processing unit 101 performs a process of converting the code image data represented by continuous tone so that the code image data is represented by a binary value with a predetermined threshold. Then, the binarization processing unit 101 passes the binarized first code image data to the comparison unit 103.

The database 102 as storage means stores, for example, a total of six matching templates (hereinafter referred to as templates) shown in FIGS. 4B to 4D.
The template is a pattern that is referred to when a bit pattern is detected from the binarized code image data. In FIGS. 4B to 4D, the pattern is expressed by pixels filled with diagonal lines on a grid of 8 pixels × 8 pixels with 1 pixel as a unit. A template used for detection of the bit pattern A is referred to as template A, and a template used for detection of the bit pattern B is referred to as template B. These templates are configured by rotating the bit pattern shown in FIG. 2A by a predetermined angle and further giving a width of 2 pixels in the horizontal direction (may be regarded as the vertical direction). In FIG. 4B, a template A (hereinafter referred to as template A (0 degree)) in which the bit pattern A is inclined by 0 degree (no inclination) and has a width of 2 pixels in the horizontal direction is shown. Templates B inclined by 0 degrees (no inclination) and having a width of 2 pixels in the horizontal direction (hereinafter referred to as template B (0 degrees)) are shown side by side.
FIG. 4C shows a template A (hereinafter referred to as template A (+8 degrees)) in which the bit pattern A is inclined by +8 degrees and has a width of 2 pixels in the horizontal direction, and the bit pattern B is inclined by +8 degrees. A template B having a width of 2 pixels in the horizontal direction (hereinafter referred to as template B (+8 degrees)) is shown side by side.
FIG. 4D shows a template A (hereinafter referred to as template A (−8 degrees)) in which the bit pattern A is inclined by −8 degrees and has a width of 2 pixels in the horizontal direction, and the bit pattern B is − Templates B inclined by 8 degrees and having a width of 2 pixels in the horizontal direction (hereinafter referred to as template B (−8 degrees)) are shown side by side.

The comparison unit 103 that functions as a code detection unit calls a template of a corresponding inclination angle from the database 102 based on the skew amount sent from the skew amount detection unit 90. For example, when the skew amount of the document 1 is 0 degree or more and less than ± 4 degrees, the template A (0 degree) and the template B (0 degree) are called. For example, if the skew amount of the document 1 is +4 degrees or more and less than +10 degrees, the template A (+8 degrees) and the template B (+8 degrees) are called. Further, for example, if the skew amount of the document 1 is −4 degrees or more and less than −10 degrees, the template A (−8 degrees) and the template B (−8 degrees) are called. In this way, a predetermined template determined in advance according to the skew amount is called from the database 102. Then, the comparison unit 103 refers to the called template and detects a bit pattern (hereinafter referred to as pattern matching) that matches the template from the binarized code image data. When the bit pattern is detected, the comparison unit 103 passes the bit value obtained from the detected bit pattern to the output destination device at the subsequent stage.
In this embodiment, the case where three types of templates having different inclination angles are prepared for the bit pattern A and the bit pattern B has been described. However, the templates stored in the database 102 are not limited to these. It is not limited. The database 102 can store templates corresponding to more inclination angles including positive and negative in the inclination direction.

FIG. 5 is a flowchart for explaining a pattern matching procedure performed by the image feature detection unit 100.
First, the binarization processing unit 101 acquires code image data (first code image data) read from the surface of the document 1 by the CCD image sensor 78 (step 101). Then, the binarization processing unit 101 performs a process of converting the multi-value data into the binary data for the first code image data acquired from the CCD image sensor 78 (step 102), and binarizes the first code image data. One code image data is passed to the comparison unit 103. Further, the comparison unit 103 acquires the skew amount of the document 1 detected by the skew amount detection unit 90 based on the second image data obtained by the line sensor 55 reading from the back side of the document 1 (step 103). Next, the comparator 103 inverts the skew amount of 1 of the acquired document (step 104), and reads out a corresponding template from the database 102 based on the skew amount that has been reversed in the positive and negative directions (step 105). Then, the comparison unit 103 performs pattern matching on the binarized first code image data based on the read template.

Next, a pattern matching procedure for one minute line bitmap (hereinafter referred to as a mark) which is a detection target in the code image data will be described. The comparison unit 103 first extracts a mark from the first code image data (step 106). Next, the comparison unit 103 performs pattern matching based on the template A on the extracted mark, and determines whether or not the mark matches the template A (step 107). As a result, when the mark matches the template A, the comparison unit 103 determines that the mark is the bit pattern A and proceeds to step 109. On the other hand, if the mark does not match template A in step 107, the process proceeds to the next step 108.
The comparison unit 103 compares the mark that does not match the template A in step 107 with the template B this time, and determines whether or not the mark matches the template B (step 108). If it is determined that the mark matches the template B, the comparison unit 103 determines that the mark is the bit pattern B, and advances the process to step 109. On the other hand, if the mark does not match the template B, the comparison unit 103 determines that the mark is not applicable, that is, neither the bit pattern A nor the bit pattern B, and advances the process to step 109.
And the comparison part 103 outputs the detection result obtained in step 107 and step 108 (step 109). At this time, as a detection result, any one of bit pattern A, bit pattern B, and not applicable is output. Then, the comparison unit 103 determines whether or not the pattern matching has been completed for all the marks obtained from the first code image data (step 110). If the pattern matching has not been completed, the process proceeds to step 106. Returning to step 106 to step 109, the next mark is processed. On the other hand, if pattern matching has been executed for all the marks obtained from the first code image data, the series of processes is terminated.

Here, the specific processing of pattern matching in steps 107 and 108 will be described by taking the case where the skew angle of the document 1 is 0 degree and the case where it is +5 degrees as an example.
Here, a case where the skew angle of the document 1 is 0 degree will be described first.
FIG. 6A shows two marks obtained by performing binarization processing on the first code image data read from the document 1 by the binarization processing unit 101. As described above, the mark obtained from the first code image is compared with a predetermined template and the target mark to determine whether the mark is bit pattern A, bit pattern B, or not applicable. Do. In this case, since the skew amount of the document 1 is 0 degree, the templates referred to in pattern matching are the template A (0 degree) and the template B (0 degree) shown in FIG. .
FIG. 6B shows a state in which the two marks shown in FIG. 6A and the template shown in FIG. 4B are overlaid for comparison. In FIG. 6B, the pixels filled with diagonal lines indicate the template pattern. A pixel painted in dark gray indicates that the mark matches the template.
The condition for determining that the mark matches the bit pattern in the comparison unit 103 is that there are a predetermined number or more (for example, three or more) of pixels constituting the mark to be compared on the pixels constituting the template, and the template There are no other pixels that make up the mark.
Therefore, referring to FIG. 6B, for the left mark, all the pixels constituting the mark are present on the template A (0 degree). Therefore, the left mark is determined to be the bit pattern A. In addition, all the pixels constituting the mark on the right mark are present on the template B (0 degree). Therefore, it can be determined that the mark on the right side is the bit pattern B.

  By the way, as described above, the template A (0 degree) and the template B (0 degree) shown in FIG. 4B are the same as the bit pattern A and the bit pattern B shown in FIG. It may be understood that the width is 2 pixels. By using such a template as a reference pattern for pattern matching, even if a code image is read from an unskewed original 1 with a shift of 0.5 pixels in any of the vertical and horizontal directions, the bit pattern can be accurately read. It becomes possible to detect.

Next, a case where the skew angle of the document 1 is +5 degrees will be described. In this case, the image is inclined by 2 pixels vertically with respect to 24 pixels horizontally.
FIG. 7A is a diagram illustrating a relationship between the marks on the document 1 conveyed in a skewed state and reading by the CCD image sensor 78. Here, the gray grid indicates the skewed document 1 and marks formed on the document 1 are represented by black fill. A black grid indicates a reading pitch by the scanner, that is, the CCD image sensor 78.
FIG. 7B shows a mark (first code image) obtained by reading the skewed document 1 with the CCD image sensor 78. In this example, since the first code image data output from the CCD image sensor 78 is multi-valued data, pixels with low gradation are displayed darkly and pixels with high gradation are displayed lightly.
FIG. 7C illustrates two marks obtained by performing binarization processing by the binarization processing unit 101.

The subsequent pattern matching procedure is the same as the case where there is no skew amount described in FIG. However, since the first code image is read with the document 1 skewed by +5 degrees, the template A (+8) shown in FIG. 4C is applied to the mark obtained from the first code image. Degree) and template B (+8 degrees) are used for pattern matching.
FIG. 7-2 (d) shows a state where the two marks shown in FIG. 7-2 (c) and the template shown in FIG. 4 (c) are overlaid for comparison.
Referring to FIG. 7D, for the mark on the left side, all the pixels constituting the mark are present on the template A. Therefore, the left mark is determined to be the bit pattern A. In addition, all pixels constituting the mark on the right mark are present on the template B. Therefore, it can be determined that the mark on the right side is the bit pattern B.

  Here, as a comparative example, a case will be described in which the bit pattern cannot be detected because pattern matching is performed without referring to an appropriate template even though the document 1 is read with skew. Here, while the skew amount of the document 1 is +5 degrees, the template A (0 degree) and the template B (0 degree) are applied to the mark obtained from the read first code image data. Assume that pattern matching is performed as a reference pattern.

FIG. 8A shows two marks obtained by the binarization processing performed by the binarization processing unit 101. FIG. 8B shows a state in which the two marks shown in FIG. 8A and the template shown in FIG. 4B are overlaid for comparison. In FIG. 8B, pixels filled with diagonal lines indicate a template pattern. A pixel painted in dark gray indicates that the mark matches the template. Furthermore, the pixels filled in with black represent the pixels that constitute the mark that is out of the template.
Here, referring to FIG. 8B, for the mark on the left side, all the pixels constituting the mark are present on the template A. Therefore, the left mark is determined to be the bit pattern A. However, with respect to the mark on the right side, the pixels constituting the mark exist in a grid other than the pixels constituting the template (pixels filled with diagonal lines). In such a case, an error occurs and it is determined that the mark on the right side is not the bit pattern B.
The mark on the right side in error is a bit pattern as a result of pattern matching not being performed in consideration of the skew amount of the document 1 although the bit pattern B is read while the document 1 is skewed. It means that it will not be detected.

Next, in this embodiment, the temporal output of the sensors (line sensor 55, CCD image sensor 78), skew amount detection unit 90, image feature detection unit 100, and front and back image output of the document 1 read by the sensor. The important relationship will be described.
In the following description, the distance between the front surface reading position of the document 1 by the CCD image sensor 78 and the back surface reading position of the document 1 by the line sensor 55 (hereinafter referred to as front / back reading gap) is 50 mm. Further, the document 1 to be read is A4 size, and is a long edge feed (LEF) of the document 1. Furthermore, the distance (gap between documents) between the rear end in the transport direction of the nth document and the front end in the transport direction of the (n + 1) th document is 210 mm.

FIG. 9 shows first image data (first document image data, first code image data) by the CCD image sensor 78, and second image data (second document image data, second code data) by the line sensor 55. It is a timing chart regarding the process of each component with respect to (code image data).
The first document 1 sent out from the document container 11 first passes through the back surface reading position of the document 1 by the line sensor 55. At this time, the line sensor 55 reads the back image of the first document 1 and acquires second image data. Here, the skew amount detection unit 90 detects the skew amount from the second image data obtained by the line sensor 55. In the present embodiment, the detection of the skew amount is completed by the time when the CCD image sensor 78 starts reading the first surface.

Next, the first document 1 passes through the surface reading position by the CCD image sensor 78. At this time, the CCD image sensor 78 reads the surface image of the first document 1. The image feature detection unit 100 detects a bit pattern from the first code image data obtained by the CCD image sensor 78. As described above, since the skew amount is obtained before the reading of the code image on the first sheet is started, the image feature detection unit 100 can detect the bit pattern from the leading edge of the document 1. it can. Here, the detection result of the bit pattern obtained from the back side of the document 1 is sent to the output destination device at the subsequent stage. Further, the second document image data obtained by the line sensor 55 and the first document image data obtained by the CCD image sensor 78 are sent in real time to an output destination device.
Thereafter, the same processing is performed on the second, third, and fourth documents 1 sent out from the document container 11.

<Embodiment 2>
In the first embodiment, it is assumed that a code image having the same content is printed on the front surface and the back surface of the document 1. However, in practice, different code images may be printed on the front surface and the back surface of the document 1.
Therefore, the second embodiment is applied when code images having the same contents or different contents are printed on the front surface and the back surface of the document 1 from which an image is to be read. That is, in the present embodiment, each bit pattern is detected from the code image data formed on each of the front surface and the back surface of the document 1. In the present embodiment, the same components as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  FIG. 10 shows a configuration of the image processing unit 80 in the present embodiment. The image processing unit 80 in the present embodiment is configured by further adding a storage unit 130 and a selection unit 140 to the skew amount detection unit 90 and the image feature detection unit 100 described in the first embodiment.

  In the present embodiment, the image feature detection unit 100 detects a bit pattern based on the code image data (second code image data) on the back side of the document 1 obtained by the line sensor 55, and the CCD image sensor. The bit pattern is detected based on the code image data (first code image data) on the surface of the document 1 obtained by 78.

  The storage unit 130 temporarily stores one sheet of image data on the back side of the document 1 read by the line sensor 55, that is, second image data (second document image data, second code image data). . When a request is received from the selection unit 140, the second image data is output.

  The selection unit 140 passes the first code image data read by the CCD image sensor 78 to the image feature detection unit 100 in real time. The selection unit 140 passes the second code image data read from the storage unit 130 to the image feature detection unit 100 at a timing when the first code image data is not sent. As described above, the selection unit 140 selectively outputs the first code image data from the CCD image sensor 78 or the second code image data from the line sensor 55 to the image feature detection unit 100.

  As described in the first embodiment, the image feature detection unit 100 includes the binarization processing unit 101, the database 102, and the comparison unit 103. However, since the first code image data and the second code image data are processed in the second embodiment, the function of the comparison unit 103 is slightly different from that of the first embodiment. When the pattern matching is performed on the first code image data, the comparison unit 103 uses a value obtained by reversing the sign of the skew amount received from the skew amount detection unit 90 as in the first embodiment (FIG. 5). Step 104 shown in FIG. Then, a predetermined template is called from the database 102 and pattern matching is performed. On the other hand, when the comparison unit 103 performs pattern matching on the second code image data, the value received from the skew amount detection unit 90 is used as it is as the skew amount, and a predetermined template is called without executing step 104 above. After that, pattern matching is performed.

Next, in the present embodiment, a time for outputting images of the front and back surfaces of the document 1 read by the sensor (line sensor 55, CCD image sensor 78), skew amount detection unit 90, image feature detection unit 100, and sensors. The general relationship will be described.
Note that the reading conditions (front / back reading gap, gap between originals, original size) of the original 1 in the following description are the same as those in the first embodiment.

11 shows first image data (first sentence image data, first code image data) by the CCD image sensor 78, and second image data (second document image data, second code data by the line sensor 55). It is a timing chart regarding the process of each component with respect to (code image data).
The first document 1 sent out from the document container 11 first passes through the back surface reading position of the document 1 by the line sensor 55. At this time, the line sensor 55 reads the back side image of the first document 1 and acquires second image data. Further, the skew amount detection unit 90 detects the skew amount from the second image data obtained by the line sensor 55. In the present embodiment, the detection of the skew amount is completed by the start of reading of the first surface by the CCD image sensor 78. The second image data corresponding to the first document 1 is stored in the storage unit 130.

  Next, the first document 1 passes through the surface reading position by the CCD image sensor 78. At this time, the CCD image sensor 78 reads the surface image of the first document 1 and acquires first image data. The obtained first image data is sent to a subsequent output destination in real time. First, the first code image data corresponding to the first document 1 is transferred to the image feature detection unit 100. At this time, since the skew amount is already known, the image feature detection unit 100 detects the bit pattern, and the detection result of the surface bitmap pattern corresponding to the obtained first document 1, that is, the first result. Is passed to the output device at the subsequent stage. On the other hand, the first document image data corresponding to the first original 1 is sent to the output destination device at the subsequent stage.

When the detection process of the first code image data corresponding to the first document 1 is completed, the selection unit 140 selects the second code 1 corresponding to the first document 1 held in the storage unit 130. Read image data. Of the read second image data, the second document image data is sent to the output destination device at the subsequent stage. On the other hand, the second code image data is transferred to the image feature detection unit 100, and bit pattern detection is executed. The bit pattern detection result on the back surface corresponding to the first document 1 obtained here, that is, the second code, is passed to the output destination device at the subsequent stage.
Thereafter, the same process is repeated for the second, third, and fourth documents 1 sent out from the document storage section 11.

In this embodiment, the first code based on the first code image data obtained by reading the surface of the document 1 is added to the first document image data obtained by reading the surface of the document 1. It is preferable to add and output. If such an output is performed, for example, in a subsequent device, it becomes possible to restrict copy output to the first document image data based on the content of the first code.
Similarly, the second document image data obtained by reading the back surface of the document 1 is added with a second code based on the second code image obtained by reading the back surface of the document 1 and output. It is preferable to do. If such an output is performed, for example, it becomes possible to restrict copy output or the like for the second document image data based on the content of the second code in the subsequent device.

<Embodiment 3>
In the second embodiment, the first code image data obtained by reading the front surface of the document 1 and the second code image data obtained by reading the back surface of the same document 1 are sequentially supplied to the image feature detection unit 100. The bit pattern was detected for each input. In contrast, the present embodiment is different from the second embodiment in that an image feature detection unit is provided for each of the first code image data and the second code image data. The second embodiment is also different from the second embodiment in that a selection unit is not required as a result of providing two image feature detection units. In the present embodiment, the same components as those in the second embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  FIG. 12 shows a configuration of the image processing unit 80 in the present embodiment. The image processing unit 80 in the present embodiment is configured by further adding a second image feature detection unit 120 to the skew amount detection unit 90 and the image feature detection unit 100 described in the first and second embodiments.

  In the present embodiment, the image feature detection unit 100 detects the bit pattern on the surface of the document based on the first code image data read by the CCD image sensor 78. Further, the second image feature detection unit 120 detects the bit pattern on the back side of the document based on the first code image data read by the line sensor 55.

  The second image feature detection unit 120 that functions as a feature detection unit basically has the same configuration as the image feature detection unit 100. That is, as described with reference to FIG. 4A in the first embodiment, the second image feature detection unit 120 includes the binarization processing unit 101, the database 102, and the comparison unit 103. However, since the second image feature detection unit 120 processes the second code image data corresponding to the back side of the document 1, the function of the comparison unit 103 provided in the second image feature detection unit 120 is the image feature detection unit. It differs from that provided in 100. More specifically, the comparison unit 103 of the second image feature detection unit 120 uses the skew amount value received from the skew amount detection unit 90 as it is when performing pattern matching on the second code image data. After calling a predetermined template, pattern matching is performed.

Next, the sensor (line sensor 55, CCD image sensor 78), skew amount detection unit 90, image feature detection unit 100, second image feature detection unit 120, and the surface of the document 1 read by the sensor in the present embodiment. A temporal relationship regarding the image output on the back surface will be described.
Note that the reading conditions (front / back reading gap, gap between originals, original size) of the original 1 in the following description are the same as those in the first embodiment.

FIG. 13 is a timing chart of processing of each component unit for read image (document image, code image) data obtained by the CCD image sensor 78 and the line sensor 55 in the present embodiment.
The first original 1 sent out from the original accommodating portion 11 first passes through the back side reading position of the original by the line sensor 55. At this time, the line sensor 55 reads the back image of the first document and acquires second image data. Further, the skew amount detection unit 90 detects the skew amount from the second image data obtained by the line sensor 55. The skew amount detection of the document 1 by the skew amount detection unit 90 is completed by the time when reading of the front surface of the first sheet is started.

  Next, the first document 1 passes through the surface reading position by the CCD image sensor 78. At this time, the CCD image sensor 78 reads the surface image of the first document 1 and acquires first image data. The obtained first image data is sent to a subsequent output destination in real time. Of these, the first document image data corresponding to the first document 1 is sent to the subsequent device. On the other hand, the first code image data corresponding to the first document 1 is passed to the image feature detection unit 100. At this time, since the skew amount of the first sheet of the document 1 is known by the skew amount detection unit 90, the image feature detection unit 100 detects the bit pattern from the first code image data. The detection result of the surface bit pattern corresponding to the first original 1 obtained here is passed to the output destination device at the subsequent stage.

On the other hand, in the present embodiment, the second image data corresponding to the first document 1 read by the line sensor 55 is also sent to the subsequent output destination in real time. First, the second document image data corresponding to the first document 1 is transmitted to the subsequent device. On the other hand, the second code image data corresponding to the first document 1 is sent to the second image feature detection unit 120. At this time, the second image feature detection unit 120 starts detecting the bit pattern from the second code image data as soon as the skew amount by the skew amount detection unit 90 is known. In other words, the bit pattern is not detected from the leading edge of the back surface of the document 1, and the bit pattern is detected from the point of time shifted by the time required for detecting the skew amount. In the present embodiment, since a plurality of codes are arranged over the entire surface of the document 1, even if a bit pattern is detected from the middle of the document 1 as in the present embodiment, a predetermined code image cannot be read. Is possible. The obtained bit pattern detection result on the back side of the first document 1 is passed to the output destination device at the subsequent stage.
Thereafter, the same process is repeated for the second, third, and fourth documents 1 sent out from the document storage section 11.

<Embodiment 4>
In the third embodiment, the bit pattern cannot be detected from the front end of the document 1 with respect to the back surface of the document 1, and the bit pattern detection is started in the middle of the detection. On the other hand, in the present embodiment, the second image data read by the line sensor 55 is subjected to delay processing, so that the bit pattern can be detected from the leading edge of the back side of the document 1 as well. Is. In the present embodiment, the same components as those in the third embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  FIG. 14 shows a configuration of the image processing unit 80 in the present embodiment. The image processing unit 80 of the present embodiment is configured by further adding an image delay unit 150 to the skew amount detection unit 90, the image feature detection unit 100, and the second image feature detection unit 120 described in the third embodiment. .

  The image delay unit 150 receives the second image data of the document 1 read by the line sensor 55. The image delay unit 150 temporarily holds the second image data for a time required for the skew amount detection unit 90 to detect the skew amount of the document 1 based on the second image data. Then, after the skew amount of the document 1 is determined by the skew amount detection unit 90, the image delay unit 150 outputs the held second image data to a subsequent output destination.

Next, the surface of the document 1 read by the sensors (line sensor 55, CCD image sensor 78), skew amount detection unit 90, image feature detection unit 100, second image feature detection unit 120, image delay unit 150 and the sensor The temporal relationship regarding the back side image output will be described.
In the following description, the reading conditions of the document 1 (front and back reading gap and document gap) are the same as those in the first embodiment.

FIG. 15 is a timing chart of processing of each component unit for read image (document image, code image) data obtained by the CCD image sensor 78 and the line sensor 55 in the present embodiment.
The first document 1 sent out from the document container 11 first passes through the back surface reading position of the document 1 by the line sensor 55. At this time, the line sensor 55 reads the back image of the first document and acquires second image data. Further, the skew amount detection unit 90 detects the skew amount from the second image data obtained by the line sensor 55. Here, the second image data corresponding to the first document 1 read by the line sensor 55 is sent to the image delay unit 150. The image delay unit 150 outputs the second image data corresponding to the sent first document 1 while being delayed for a predetermined time. This fixed time is the time required for the skew amount detection unit 90 to complete the detection of the skew amount of the document 1. The detection of the skew amount is completed until the CCD image sensor 78 starts reading the surface of the first document 1.

  Next, the first document 1 passes through the surface reading position by the CCD image sensor 78. At this time, the CCD image sensor 78 reads the surface image of the first document 1 and acquires first image data. The first image data read by the CCD image sensor 78 is sent to a subsequent output destination in real time. First, the first code image corresponding to the first document 1 is transferred to the image feature detection unit 100. At this time, since the skew amount of the document 1 is already known, the image feature detection unit 100 detects the bit pattern from the first code image data. On the other hand, the first document image data corresponding to the first original 1 is sent to the output destination device at the subsequent stage.

When the skew amount detection by the skew amount detector 90 is completed and the second image data corresponding to the first document 1 is output from the image delay unit 150, the first document 1 is supported. The second document image data is transmitted to the subsequent device. The second code image data corresponding to the first document 1 is passed to the second image feature detection unit 120. The second image feature detection unit 120 performs bit pattern detection on the second code image data sequentially sent from the image delay unit 150. The bit pattern detection result obtained here is passed to the subsequent device.
Thereafter, the same process is repeated for the second, third, and fourth documents 1 sent out from the document storage section 11.

  In the first to fourth embodiments, when the document 1 is stacked on the document storage unit 11, the document 1 is set with the front side facing up. However, the present invention is not limited to this, and the back side is set with the front side facing upward. It doesn't matter. Further, the line sensor 55 for reading the back surface of the document 1 is provided upstream of the conveyance path from the reading position by the CCD image sensor 78, but the present invention is not limited to this, and is provided downstream. Also good. Further, the back side reading is not limited to the CIS, and a reduction optical system such as a CCD sensor may be used. Conversely, the front side reading may be performed by the CIS. The method is not limited as long as an image reading system capable of reading images on the front and back sides of the document 1 is provided.

Note that an existing technique can be used for the skew amount detection method in the present embodiment. Any method may be used as long as the skew amount can be detected.
Further, by providing a front / back reading gap of 50 mm, the skew amount is detected between the gaps except for the third embodiment, and the bit pattern can be detected from the leading edge of the document 1. Therefore, the front / back reading gap can be further shortened by reducing the detection time for detecting the skew amount. Further, by shortening the time required for detecting the skew amount, it is possible to further reduce the area of the document 1 where the bit pattern cannot be detected in the third embodiment.

  In the present embodiment, a template with an inclined bit pattern is prepared in advance, and pattern matching is performed on the code image read from the document 1, but the pattern matching method is not limited to this. Absent. For example, a pattern stored as a template can be a bitmap itself, not a tilted bitmap. In this case, pattern matching may be performed by performing an inclination process on the code image data based on the skew amount obtained from the skew amount detection unit 90.

  Furthermore, although the case where the code image is configured by a bit pattern has been described as an example in the present embodiment, the code image to which the present embodiment is applied is not limited to this. For example, a QR code or a barcode can be used as a code image, and a text image can be read by OCR or the like.

It is a figure which shows the whole structure of the image reading apparatus with which this Embodiment is applied. (A) is a diagram illustrating a code image, and (b) is a diagram illustrating a code image (bit pattern) that has been read in a skewed state. 2 is a diagram illustrating a configuration of an image processing unit according to Embodiment 1. FIG. (A) is a figure which shows the structure of an image feature detection part, (b)-(d) is a figure which respectively shows the data (template) which an image feature detection part hold | maintains. It is a flowchart explaining the flow of the pattern matching which an image feature detection part performs. When the skew angle of the original is 0 degree, (a) shows two marks obtained from the code image read from the original, and (b) shows two marks and a template for 0 degree superimposed. It is a figure which shows a state, respectively. (A) is a relationship between a mark on a document conveyed in a skewed state and reading by a sensor, and (b) is a mark (first code image) obtained by reading a skewed document by a sensor. FIG. When the skew angle of the original is +5 degrees, (c) is the two marks obtained from the code image read from the original, and (d) is the overlap of the two marks and the +5 degree template. It is a figure which shows a state, respectively. When the skew angle of the original is +5 degrees, (a) shows two marks obtained from the code image read from the original, and (b) shows two marks and a 0 degree template superimposed. It is a figure which shows a state, respectively. 3 is a timing chart regarding processing of each component in the first embodiment. 6 is a diagram illustrating a configuration of an image processing unit in Embodiment 2. FIG. 6 is a timing chart regarding processing of each component in Embodiment 2. 10 is a diagram illustrating a configuration of an image processing unit in Embodiment 3. FIG. 10 is a timing chart regarding processing of each component in Embodiment 3. FIG. 10 is a diagram illustrating a configuration of an image processing unit in a fourth embodiment. 12 is a timing chart regarding processing of each component in Embodiment 4.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Document feeder, 55 ... Line sensor, 70 ... Scanner apparatus, 78 ... CCD image sensor, 80 ... Image processing part

Claims (9)

A first sensor that outputs first image data obtained by reading an image of the first side of the conveyed document;
A second sensor for outputting second image data obtained by reading an image on the second surface of the document on the upstream side in the transport direction of the document from the first sensor;
An inclination detector that detects the inclination of the document based on the second image data;
A code detection unit that detects a first code included in the first image data and detects a second code included in the second image data based on a tilt detection result by the tilt detection unit; An image reading apparatus.   When detecting the first code, the code detection unit uses the inclination detection result with the sign reversed, and when detecting the second code, the code detection unit uses the inclination detection result as it is. The image reading apparatus according to claim 1. A selection unit that selectively outputs the first image data or the second image data to the code detection unit;
The image reading apparatus according to claim 1, wherein the selection unit outputs the second image data after outputting the first image data in the same document.   2. The image according to claim 1, wherein the first code is added to the first image data for output, or the second image data is added to the second code data for output. Reader. First reading means for reading an image of the first side of the conveyed document;
Second reading means for reading an image on the second side of the document being conveyed;
An inclination detecting means for detecting an inclination of the document based on second image data obtained by reading the second surface by the second reading means;
An image reading apparatus comprising: a feature detection unit that detects a feature of first image data obtained by reading the first surface by the first reading unit, using a tilt detection result by the tilt detection unit.   The image reading apparatus according to claim 5, wherein the feature detection unit detects a code included in the first image data as a feature of the first image data. Storage means for storing a plurality of templates with different inclination angles of the code;
The feature detection unit reads a corresponding template from the storage unit based on the tilt detection result by the tilt detection unit, and detects the code from a comparison result between the first image data and the read template. The image reading apparatus according to claim 6.   The image reading apparatus according to claim 5, wherein the feature detection unit reverses the sign of the tilt detection result by the tilt detection unit.   The image reading apparatus according to claim 5, wherein the feature detection unit further detects a feature of the second image data using the tilt detection result by the tilt detection unit.

Images