JP5838763B2 - Image processing apparatus and image processing program - Google Patents

Description

  The invention disclosed in this specification relates to a technique for classifying a document image into one of a plurality of document categories based on read image data including the document image.

  Conventionally, based on read image data of a document image read by an image reading unit such as CIS (Contact Image Sensor), the document image is classified into one of document categories such as a color document and a monochrome document. There is an image processing apparatus (see Patent Document 1). The image processing apparatus divides the read image into a plurality of blocks, and based on whether or not the number of pixels classified into color pixels or monochrome pixels among the pixels included in each block is greater than or equal to a threshold value, Each block is classified as a color block or a black and white block. The image processing apparatus classifies the document image as a color document or a monochrome document based on the number of color blocks and monochrome blocks.

  In addition, there is an image processing apparatus that classifies the original image into a single-color (plain color) original or a multi-color original based on the read image data of the original read by the image reading means (see Patent Document 2). The image processing apparatus divides the read image into a plurality of blocks, and based on whether or not the number of pixels classified into single-color (blank) pixels among the pixels included in each block is equal to or greater than a threshold value. Each block is classified as a single color block or a multicolor block. Then, the image processing apparatus classifies the document image as a single color document or a multicolor document based on the number of plain blocks.

JP 2003-1116011 A Japanese Patent Laid-Open No. 2008-21981

  By the way, since the reading range of the image reading unit is wider than the document size, the edge portion of the document is also read, and read image data including the edge image of the document may be generated. When the document category is classified by the above-described block division for the read image including the edge image of the document, the document image is classified into an incorrect document category by the color information of the pixels constituting the edge image. There is a risk that. Specifically, even when an actual document image is a color document, the document image may be classified as a monochrome document by determining that the block including the edge image of the document is a monochrome block. Even when the actual document image is a blank document, the document image may not be a blank document because the block including the edge image of the document is determined to be a black and white block.

  The present specification discloses a technique for suppressing a document image from being classified into an incorrect document category due to the influence of the edge image of the document.

  An image processing apparatus disclosed in this specification includes a data acquisition unit that acquires data of a read image including an edge image of a document, and a control unit, and the control unit includes a target range in the read image. A pixel classification process for classifying each pixel belonging to one of a plurality of pixel categories based on color information of the pixel, a block division process for dividing the target range into a plurality of blocks, and each of the plurality of blocks. A block classification process for classifying the document image into any one of a plurality of block categories based on the number of pixels classified into at least one of the plurality of pixel categories in the pixel classification process; Based on the number of blocks classified into at least one of the plurality of block categories in the determination process, a plurality of blocks A document classification process for classifying the document image into any one of the document categories, the target range does not include at least one edge image of the document image, and the block includes the document image. The length in the direction parallel to the one edge image is longer than the length in the direction orthogonal to the one edge image.

  In the image processing apparatus, in the block classification process, the control unit classifies each of the plurality of blocks into one of a block category including at least two of a color block, a monochrome block, and a gray block, and the document classification In the processing, the document image may be classified into any one of document categories including at least two of a color document, a monochrome document, and a gray document.

  In the image processing apparatus, the control unit classifies each of the plurality of blocks into one of a block category including a blank block and a non-blank block in the block classification process, and the document image in the document classification process. May be classified into either a blank document or a document category including a non-blank document.

In the image processing device, in the block classification process, the control unit classifies each of the plurality of blocks into a block category including a blank block and a non-blank block, and the plurality of blocks. Performing a second classification process for classifying each into one of a block category including at least two of a color block, a black and white block, and a gray block;
In the document classification process, the document image is classified into one of a document category including a blank document and a non-blank document based on the result of the first classification process, and the document is classified based on the result of the second classification process. The image may be classified into any one of document categories including at least two of a color document, a monochrome document, and a gray document.

  In the image processing apparatus, the control unit detects the resolution of the read image, and increases the block size to increase the number of pixels included in the block by increasing the size of the block as the detection value of the resolution increases. A configuration for executing the changing process may also be used.

  The image processing apparatus includes a reading device that reads the document while relatively moving with respect to the document, and the data acquisition unit acquires data of a read image including an edge image of the document from the reading device, The target range may not include at least one of a pair of edge images orthogonal to the relative movement direction of the reading device and the document among the document images.

  In the image processing apparatus, the target range may include an edge image having a relatively low image density and not including an edge image having a relatively high image density among the pair of edge images. .

  In the above-described image processing apparatus, the reading device has a light source disposed at a position where light is irradiated obliquely from behind a relatively moving document, and a position that receives light irradiated by the light source and reflected from the document. The target range does not include an edge image read before the reading device, but includes an edge image read later, out of the pair of edge images. But you can.

  The present invention can be realized in various modes such as the image processing apparatus, the image processing method, a computer program for realizing the functions of these methods or apparatuses, and a recording medium on which the computer program is recorded.

  According to the present invention, the target range used for classification of the document category does not include at least one edge image among the document images. For this reason, compared with a configuration in which both the one edge image and the other edge image are included in the target range, it is possible to suppress the influence of the document edge image on the classification of the document category. Further, the block has a shape in which the length in the direction parallel to the one edge image is longer than the length in the direction orthogonal to the one edge image. In such a block, for example, since the ratio of the edge image area to one block is smaller than that of a square block, the influence of the edge image of the document on the classification of the block category can be suppressed. it can. Thereby, it is possible to suppress the document image from being classified into an incorrect document category due to the influence of the edge image of the document.

1 is a schematic internal configuration diagram of an image reading apparatus according to an embodiment. Block diagram schematically showing the electrical configuration of the image reading apparatus Flow chart showing the reading process Flow chart showing block analysis processing Schematic diagram showing the block of this embodiment Schematic diagram showing the block of the comparative example Flow chart showing document determination processing

  An embodiment will be described with reference to FIGS. 1 to 7. In the following description, it is assumed that the left side of FIG. 1 is the front side of the image reading device 1, the front side of the paper is the right side of the image reading device 1, and the upper side of the paper is the upper side of the image reading device 1. Note that the image reading apparatus 1 is an example of an image processing apparatus.

(Mechanical configuration of image reading apparatus)
As shown in FIG. 1, the image reading apparatus 1 includes a document tray 2, a main body unit 3, and a discharge tray 4. The image reading apparatus 1 reads an image on the document M being conveyed by the reading device 24 while conveying the document M placed on the document tray 2 into the main body 3, and then reads the document M on the discharge tray. 4 is a sheet feed scanner which discharges to 4. The document M is not limited to paper but may be plastic.

  The document tray 2 is provided on the rear side of the main body 3 with the front side inclined downward, and one or a plurality of documents M are placed thereon. A conveyance path 22 extending from the front end of the document tray 2 to the rear end of the discharge tray 4 is provided in the main body 3. In addition, a pickup roller 20, a separation pad 21, a feed roller 23, a reading device 24, a discharge roller 25, a front sensor 26, and a rear sensor 27 are provided around the transport path 22.

  The pickup roller 20 is disposed on the front side of the document tray 2 and is driven by a motor (not shown), and draws one or a plurality of documents M placed on the document tray 2 into the main body 3 by frictional force. . The separation pad 21 is disposed to face the pickup roller 20 and separates a plurality of documents M one by one by frictional force. As a result, the originals M are conveyed one by one into the main body 3.

  The feed roller 23 is provided on the downstream side of the conveyance path 22 relative to the pickup roller 20 and the like, is driven by a motor (not shown), and conveys the document M existing on the conveyance path 22 forward. The reading device 24 is provided on the downstream side of the conveyance path 22 with respect to the feed roller 23, and reads an image of the document M conveyed by the feed roller 23.

  The reading device 24 is disposed below the conveyance path 22 and reads an image on the other side of the document M, in other words, the lower side. The reading device 24 has a configuration having a CIS (Contact Image Sensor). Specifically, the reading device 24 has a configuration in which a light source 31, a light guide member 32, an imaging unit 33, and a platen glass 34 are mounted on a carriage 35. The light source 31 has a configuration in which a plurality of RGB light emitting elements are arranged side by side in the left-right direction, and irradiates the light L1 with respect to the transport path 22 obliquely from behind.

  The imaging unit 33 is configured by arranging a plurality of imaging elements (not shown) in the left-right direction. The platen glass 34 is disposed along the transport path 22. The light source 31 irradiates the original M in the transport path 22 with the light L 1 through the platen glass 34, and the imaging unit 32 receives the reflected light L 2 from the original M through the light guide member 32. The reading device 24 is not limited to the CIS, and may be configured to have, for example, a CCD (Charge Coupled Drive Image Sensor).

  The discharge roller 25 is provided on the downstream side of the conveyance path 22 with respect to the reading device 24, and sends the document M on which an image has been read by the reading device 24 to the outside of the main body 3. The discharge tray 4 is provided on the front side of the main body 3, and the original M sent out to the outside of the main body 3 is stacked on the discharge tray 4. The conveyance path 22, the pickup roller 20, the feed roller 23, and the discharge roller 25 constitute a conveyance mechanism 29.

  The front sensor 26 is provided on the front end side of the document tray 2, detects the presence of the document M arranged on the document tray 2, and outputs a detection signal SG1 corresponding to the detection result. The rear sensor 27 detects the presence of the document M being transported on the upstream side of the transport path 22 relative to the reading device 24, and outputs a detection signal SG2 corresponding to the detection result.

(Electrical configuration of image reading apparatus)
As shown in FIG. 2, the image reading apparatus 1 includes a control board 40. The control board 40 includes a control unit 41, a device control unit 42, an analog front end (hereinafter referred to as AFE) 43, and a drive unit 44. The front sensor 26, the rear sensor 27, the transport mechanism 29, and the like are connected to these via a bus 45. A display unit 46, an operation unit 47, and the like are connected. The operation unit 46 includes a plurality of buttons, and allows a user to input various instructions and settings. The display unit 47 includes a display, a lamp, and the like, and can display various setting screens and operation states of the apparatus.

  The control unit 41 is an example of a control unit, and includes a central processing unit (hereinafter referred to as CPU) 41A, a memory 41B, and an image processing circuit 41C. Various programs for controlling the operation of the image reading apparatus 1 are stored in the memory 41B, and the CPU 41A controls each unit of the image reading apparatus 1 according to the program read from the memory 41B. The memory 41B has a RAM and a ROM. The medium for storing the various programs may be a non-volatile memory such as a CD-ROM, a hard disk device, or a flash memory (registered trademark) in addition to a RAM or the like.

  The device control unit 42 is connected to the reading device 24, and transmits a signal for controlling turning on / off of the light source 31 and reading by the imaging unit 33 to the reading device 24 based on a command from the CPU 41 </ b> A. When the reading device 24 receives a signal from the device control unit 42, the reading device 24 turns on the light source 31 and the reflected light L <b> 2 reflected from the document M conveyed on the conveyance path 22 is received by the imaging unit 33. In addition, the reading device 24 outputs first read image data, which is an analog signal corresponding to the amount of light received by each imaging element of the imaging unit 33, to the AFE 43.

  The AFE 43 performs A / D conversion on the first read image data, which is an analog signal output from the reading device 24, and converts it into second read image data, which is RGB color system gradation data. The second read image data is stored in the memory 41B via the bus 45. The AFE 43 is an example of a data acquisition unit.

  The image processing circuit 41C is, for example, a hardware circuit dedicated to image processing, and performs YCbCr conversion processing described later on the second read image data stored in the memory 41B. The third read image data after the processing is stored in the memory 41B. The drive unit 44 controls the transport mechanism 29 based on a command from the CPU 41A and transports the document M.

(Reading process)
When the user places the document M on the document tray 21 and performs an input operation of a scan instruction while setting a desired resolution with the operation unit 47, the CPU 41A is based on the detection signal SG1 from the front sensor 26. If it is determined that there is a document M on the tray 2, the program is read and the reading process shown in FIG. 3 is executed. The program for executing the reading process is an example of an image processing program. The image reading apparatus 1 classifies the read document image G1 into one of a plurality of document categories while reading the image of the document M by executing this reading process.

  Here, the document category is a category classified by the expression color of the document image G1, and includes, for example, a color document, a single color document, and a blank document. A color manuscript means that the expression color of the manuscript is color, for example, a manuscript in which the proportion of the chromatic color portion is a predetermined proportion or more. Monochromatic manuscript means that the original color is a single color. For example, the proportion of chromatic parts such as black and white and gray scale manuscripts is less than the prescribed ratio, and the proportion of achromatic parts is the prescribed percentage. This is a single-color original document. A blank document refers to a document whose proportion of chromatic and achromatic parts is less than a predetermined ratio, such as a blank document.

  Specifically, the CPU 41A executes a blank sheet removal function and a document color determination function in the reading process. The blank paper removal function is a function that classifies the document image G1 into a blank document in which no image is formed and a non-blank document in which an image is formed, and removes read data of the blank document. The document color discrimination function classifies the document image G1 into, for example, a color document, a gray document, and a monochrome document, performs a gray conversion process on the read data of the gray document, and performs a monochrome conversion process on the read data of the monochrome document. It is a function to perform.

  The CPU 41A instructs the drive unit 44 to cause the transport mechanism 29 to start transporting the document M on the document tray 21 (S1), and waits until the front end of the document M is detected (S2: NO). The CPU 41A detects the front end of the document M based on the level inversion of the detection signal SG2 from the rear sensor 27. When the CPU 41A detects the front edge of the document M (S2: YES), the CPU 41A instructs the device control unit 42 to start a reading operation that causes the reading device 24 to read an image of the document M (S3). At this time, the reading device 24 turns on the RGB light sources in a time-sharing manner, and reads the image on each surface in the color format. Then, the CPU 41A stores the second read image data sequentially output from the AFE 43 in the memory 41B.

  The CPU 41A instructs the image processing circuit 41C to start YCbCr conversion processing (S4). In this YCbCr conversion process, the image processing circuit 41C reads out the second read image data from the memory 41B, and converts the second read image data from the RGB color system data to the luminance data Y and the two color difference data. The data is converted into third read image data of the YCbCr color system having Cb and Cr, and stored in the memory 41B. In this embodiment, the value range of the luminance data Y is 0 to 255, and the larger the value, the higher the luminance. The value range of each color difference data is −128 to +127, and the absolute value is large. It is assumed that there are more red or blue components.

(1) Block analysis process CPU41A performs the block analysis process shown in FIG. 4 after the start of reading operation (S5). 5 and 6 show the original image G1, the read image G2 read by the reading device 24, the analysis range H1 of the block analysis process, and the blocks B1 and B2. It is assumed that the read image G2 is read by the reading device 24 from the upper side to the lower side in the drawing. That is, the solid line at the periphery of the document image G1 indicates the edge image of the document M, the edge image on the upper side of the paper is the front edge image E1 of the document M, and the edge image on the lower side of the paper is the document M. The left and right edge images of the document M are the left and right edge images E3 and E4 of the document M. [m1]

  The read image G2 is an image of the entire reading range of the reading device 24, in other words, the entire image based on the third read image data, and includes all four edge images of the document image G1. The analysis range H1 is a target range of the block analysis process, and does not include the front end image E1 and the rear end image E2 of the document image G1, but includes the left and right end images E3 and E4. A block B1 in FIG. 5 is a block used in the present embodiment, and has a long rectangle in the left-right direction. A block B2 in FIG. 6 is a block of a comparative example and forms a square.

In the block analysis process, the CPU 41A first determines the size of the block B1 (S11). Specifically, the CPU 41A detects the resolution set by the operation unit 47, increases the block size as the resolution detection value is higher, and increases the number of pixels included in the block. Execute the change process. An example of the relationship between the resolution detection value and the block size is as follows.
For 150 dpi, left / right width = 57 pixels, front / rear width = 42 pixels
In the case of 200 dpi, the left-right width = 76 pixels, the front-rear width = 56 pixels
In the case of 300 dpi, the left-right width = 114 pixels, the front-rear width = 84 pixels
In this way, the size of the block B1 is increased as the resolution detection value is higher, and the number of pixels included in the block B1 is increased in proportion. For this reason, it is possible to suppress the influence of the edge image of the original M due to the difference in resolution.

  When determining the size of the block B1, the CPU 41A determines the analysis margin D (S12). As shown in FIG. 5, the analysis margin D is a margin in the front-rear direction of the analysis range H1 with respect to the read image G2, and more specifically, the number of pixels between the front end images of the document image G1 and the read image G2. , And the number of pixels between the rear end images. In the present embodiment, the left and right margins of the analysis range H1 with respect to the read image G2 are set to a minimum value in the specifications of the image reading apparatus 1. The reason for this is to ensure the wide left and right width of the analysis range H1 as much as possible in order to deal with various sizes of the original M by the automatic original mode in which the size of the original M is not designated in advance.

Specifically, the CPU 41A executes an analysis margin changing process in which the analysis margin D is increased as the resolution detection value is higher, and the number of pixels corresponding to the analysis margin is increased. For example, the relationship between the resolution detection value and the analysis margin is as follows.
In the case of 150 dpi, the analysis margin D = 153 pixels
In the case of 200 dpi, the analysis margin D = 204 pixels
In the case of 300 dpi, the analysis margin D = 306 pixels
Thus, the analysis margin D is increased as the resolution detection value is higher, and the number of pixels corresponding to the analysis margin is increased in proportion. For this reason, it can suppress that the area ratio of the analysis range H1 with respect to the read image G2 varies by the difference in resolution.

  When the analysis margin D is determined, the CPU 41A starts execution of a block division process for dividing the image within the analysis range H1 into a plurality of blocks B1 based on the third read image data as shown in FIG. 5 (S13). ). After starting the block division process, the CPU 41A initializes the target block string number K to 1 (S14), and selects the foremost block string (K = 1) as the target block string. Then, the CPU 41A executes a pixel classification process for classifying each pixel included in each block B1 of the target block row into one of a plurality of pixel categories based on the luminance data Y and the two color difference data Cb and Cr. To do. The luminance data Y and the two color difference data Cb and Cr are examples of color information.

Specifically, the CPU 41A creates histograms for the color difference data Cb and Cr and the luminance data Y for the pixels included in each block B1, and executes the first pixel classification process for the blank page removal function (S15). . In the first pixel classification process, the CPU 41A classifies each pixel as a blank pixel (white pixel) or a non-blank pixel using the following pixel determination condition. An example of the pixel determination condition is as follows.
Blank pixel condition: Y value = 225 to 255 range Non-blank pixel condition: Y value = 0 to 224 range

Further, the CPU 41A executes a second pixel classification process for the document color discrimination function (S16). In the second pixel classification process, the CPU 41A classifies each pixel as one of a chromatic color pixel, a gray pixel, a black pixel, and a blank pixel (white pixel) using the following pixel determination condition. An example of the pixel determination condition is as follows.
Chromatic pixel condition: Cb value outside the range of -10 to +10 or Cr value outside the range of -10 to +10 Gray pixel condition: Cb value within the range of -10 to +10 and Cr value = -10 Within +10 range and Y value = 64 to 224 Black pixel condition: Cb value = within −10 to +10 range, Cr value = within −10 to +10 range, and Y value = 0 -63 range Blank pixel condition: Cb value = within a range of −10 to +10, Cr value = within a range of −10 to +10, and Y value = 225 to 255 For the classification of pixel categories In addition to the method of this embodiment, various known classification methods can be used.

  Next, the CPU 41A executes block classification processing for classifying each block B1 belonging to the target block column into one of a plurality of block categories based on the result of the pixel classification processing. Specifically, the CPU 41A executes the first block classification process for the blank paper removal function (S17). In the first block classification process, the CPU 41A classifies each block B1 as either a blank block (monochromatic block) or a non-blank block (multicolor block) using the following block determination condition. The block determination condition includes, for example, that the ratio of the number of pixels belonging to a specific pixel category to the number of all or a part of each block B1 is within a specified range or outside a specified range. If all or part of the number of pixels is a fixed value, the CPU 41A indirectly determines whether the number of pixels in a specific pixel category is equal to or less than a predetermined threshold, so that the block B1 It may be determined whether or not a block determination condition is satisfied.

An example of the block determination condition is as follows.
Blank block condition: Ratio of the number of blank pixels to the total number of pixels of the block = 85% or more Non-blank block condition: Ratio of the number of blank pixels to the total number of pixels of the block = less than 85%

  Further, the CPU 41A executes a second block classification process for the document color discrimination function (S18). In the second block classification process, the CPU 41A classifies each block B1 into one of a color block, a gray block, and a black and white block using the following block determination condition.

An example of the block determination condition is as follows.
Color block condition: Ratio of the number of chromatic color pixels to the total number of pixels of the block = 3% or more Gray block condition: The block does not satisfy the color block condition, and the ratio of the number of gray pixels to the total number of pixels of the block = 20% or more Black and white block: None of the above color block condition, gray block condition, or blank block condition is satisfied. For block category classification, various known classification methods other than the method of this embodiment should be used. Can do.

  When the block classification process is completed for the target block row, the CPU 41A determines whether the trailing edge of the document M is detected based on the level inversion of the detection signal SG2 from the rear sensor 27 (S19). If the trailing edge of the document M is not detected (S19: NO), the CPU 41A adds 1 to the target block column number K (S20), and performs pixel classification processing and block classification processing (S15 to S18) for the next block column. Execute. If the CPU 41A detects the trailing edge of the document M (S19: YES), the CPU 41A ends the reading operation and conveyance of the document M (S21).

  Next, the CPU 41A determines the final block sequence number N (S22). Specifically, since the final pixel line of the read image is determined by the end of the reading operation, the CPU 41A determines the upper pixel line from the final pixel line by the analysis margin D as the rear end pixel line of the analysis range H1. And Then, the CPU 41A determines the number of block columns including the rear end pixel line as the final block column number N. Thereafter, if the CPU 41A determines that the target block column number K is not the final block column number N (S23: NO), the CPU 41A proceeds to S20, and if it determines that the target block column number K is the final block column number N (S23). : YES), this block analysis processing is terminated. [m2]

(2) Document Determination Process After completing the block analysis process, the CPU 41A proceeds to S6 in FIG. 3, executes the document determination process shown in FIG. 7, and ends the reading process. The CPU 41A classifies the document image G1 into one of a blank document, a color document, a gray document, and a monochrome document using the following document determination condition (an example of document classification processing). The document determination condition includes, for example, that the ratio of the number of blank blocks to the total number of blocks in the analysis range H1 is within or outside the specified range. An example of the document determination condition is as follows.
Blank document condition: Ratio of the number of blank blocks to all blocks in the analysis range H1 = 85% or more Color document condition: The number of color blocks for all blocks in the analysis range H1 that do not satisfy the blank document condition Ratio = 3% or more Gray document condition: The ratio of the number of gray blocks to all blocks in the analysis range H1 that does not satisfy the blank document condition and the color document condition = 3% or more Black and white document condition: the blank document condition described above Neither color original condition nor gray original condition is satisfied

  If the CPU 41A determines that the document image G1 is a blank document (S31: YES), the CPU 41A deletes the third read image data itself of the read image G2 corresponding to the document image G1 from the memory 41B (S32), and the original document. The determination process and the reading process are terminated. Thereby, in the memory 41B, the storage amount of the third read image data can be reduced. First, it is preferable to determine whether or not the document image G1 satisfies the blank document condition. This is because when it is determined that the document is a blank document, it is not necessary to determine other document determination conditions such as color document conditions, and the processing load can be reduced.

  When the CPU 41A determines that the document image G1 is a non-blank document (S31: NO), the CPU 41A determines whether the document image G1 is a color document (S33). If the CPU 41A determines that the document image G1 is a color document (S33: YES), it leaves the luminance data Y and the two color difference data, and leaves the third read image data as it is in the memory 41B. . If CPU 41A determines that document image G1 is not a color document (S33: NO), CPU 41A determines whether document image G1 is a gray document (S34).

  If the CPU 41A determines that the document image G1 is a gray document (S34: YES), the CPU 41A executes a gray conversion process that leaves only the luminance data Y of the third read image data and deletes the two color difference data (S35). ). Thereby, in the memory 41B, the storage amount of the third read image data can be reduced. The CPU 31A may convert the third read image data into the second read image data and leave only one color data of RGB data. If CPU 41A determines that document image G1 is not a gray document (S34: NO), CPU 41A determines that document image G1 is a monochrome document, leaving only luminance data Y in the third read image data, and deleting the two color difference data. Then, a black and white conversion process for converting the color difference data into binary data of two gradations is executed (S36). Thereby, in the memory 41B, the storage amount of the third read image data can be reduced.

(Effect of this embodiment)
In FIG. 6 showing the comparative example, the analysis range H2 includes all four edge images of the document image G1. On the other hand, in the present embodiment, as shown in FIG. 5, the analysis range H1 does not include the front end image E1 and the rear end image E2 of the document image G1, but includes the left and right end images E3 and E4. Therefore, compared with the comparative example, the influence of the edge image of the document can be suppressed with respect to the classification of the document category.

  In the comparative example, as shown in FIG. 6, the block B2 is square. Here, for example, the left end image E3 of the document M exists in the block B2X shown in an enlarged manner in the figure, but the left end image E3 is not the document image G1. For this reason, when the document image G1 is a blank document as shown in the figure, the block B2X should be determined to be a blank block.

  However, due to the presence of the left end image E3 of the document M, the block B2X is erroneously determined not to be a blank block. Specifically, as described above, the blank block condition is that the block is not a color block, and the ratio of the number of blank pixels to the total number of pixels in the block is 85% or more. However, in the block B2X, since the total number of pixels is 100, the number of black pixels is 10, the number of gray pixels is 20, and the number of blank pixels is 70, the above ratio is 70% and does not satisfy the blank block condition. . Since the block B2X satisfies the gray block determination condition, it is erroneously determined to be a gray block.

  On the other hand, as described above, the block B1 of the present embodiment forms a long rectangle in the direction orthogonal to the left image E3. In the block B1X shown in an enlarged manner in FIG. 5, the total number of pixels is 200, the number of black pixels is 10, the number of gray pixels is 20, and the number of blank pixels is 170. 85%, which satisfies the blank block condition, and the block B2X is correctly determined to be a blank block.

  Thus, the block B1 has a shape in which the length in the left-right direction is longer than the length in the up-down direction. In such a block, since the ratio of the edge image area to one block is smaller than the block B2 of the comparative example, the influence of the edge image of the document on the block category classification is suppressed. Can do. That is, in the first block classification process (S17 in FIG. 4) for classifying the block B1 into the blank block and the non-blank block and the blank document detection (S31 in FIG. 7), the left and right end images of the document image G1 are detected. The influence can be suppressed. In the second block classification process (S18 in FIG. 4) for classifying the block B1 into a color block, a gray block, and a black and white block, and in the category classification (S33 and S34 in FIG. 7) regarding the image color of the document image The influence of the left and right end images of the image G1 can be suppressed.

  Further, according to the present embodiment, the block classification process and the document classification process are executed based on the result of the common pixel classification process in S15 and S16 of FIG. 4, while suppressing the influence of the edge image of the document. Further, it is possible to efficiently perform blank document detection and category classification relating to the image color of the document image.

  The analysis range H1 includes a pair of edge images orthogonal to the relative movement direction of the reading device 24 and the document M, in other words, the sub-scanning direction, that is, upper and lower end images E1 and E2 in the document image G1. Absent. For this reason, the analysis range H1 can be determined relatively easily by adjusting the processing timing in the sub-scanning direction.

<Other embodiments>
The present invention is not limited to the embodiments described with reference to the above description and the drawings, and for example, the following various aspects are also included in the technical scope of the present invention.

  In the above embodiment, as an example of the image processing apparatus, a sheet feed scanner in which the document M moves with respect to the stationary reading device 24 is taken as an example. However, the image processing apparatus is not limited to this, and may be a configuration in which a reading device moves with respect to an original placed on a platen in a stationary state, that is, a flat bed type image reading apparatus. The image processing apparatus is not limited to a single scanner, and may be a printing apparatus, a copier, or a multifunction machine having functions such as a copy function and a facsimile function. Furthermore, the image processing apparatus may be an information processing apparatus that does not include a reading device, such as a personal computer, and that acquires read image data from an external device such as a scanner device or an external memory via an interface. In this configuration, the interface is an example of a data acquisition unit.

  In the above embodiment, the AFE 43 is taken as an example of the data acquisition unit. However, the data acquisition unit is not limited to this, and may be an external storage device that stores the read image data, or a connection unit that is communicably connected to an external device that transmits the read image data, such as a USB interface.

  In the above embodiment, the control unit 41 is taken as an example of the control unit. However, the control unit is not limited thereto, and may be configured to execute the reading process only by the CPU 41A and the memory 41B without including the image processing circuit 41C. The control unit may have a configuration including a plurality of CPUs, a configuration including a hardware circuit such as an ASIC (Application Specific Integrated Circuit), or a configuration including both the hardware circuit and the CPU. For example, a configuration in which at least two of the pixel classification process, the block division process, the block classification process, the document classification process, and the block size change process are executed by separate CPUs and hardware circuits may be employed. Further, the order of these processes may be changed as appropriate.

  In the above embodiment, the read image G2 includes all four edge images of the document image G1. However, the present invention is not limited to this, and the read image may be an image including at least two edge images orthogonal to each other in the document image G1.

  In the above embodiment, the analysis range H1 is a target range of the block analysis process, and does not include the front end image E1 and the rear end image E2 of the document image G1, but includes the left and right end images E3 and E4. However, the analysis range is not limited to this, and may include one of the front end image E1 and the rear end image E2. Even in this configuration, by using the block B1, it is possible to suppress the document image from being classified into an incorrect document category due to the influence of the edge image. For example, for some reason, one of the front end image E1 and the rear end image E2 may have a higher image density than the other. In this case, it is preferable that the analysis range does not include an edge image having a relatively high image density but includes an edge image having a relatively low image density. Thereby, it is possible to suppress a portion to be excluded from the document image.

  Furthermore, the reading device 24 of the above embodiment is configured to irradiate the light M1 from the light source 31 obliquely from behind the document M. In this configuration, since the light irradiated to the rear end of the document M is diffusely reflected, the amount of light received by the imaging unit 33 is smaller than the light irradiated to the front end of the document M. Therefore, the edge image that is read later by the reading device 24 in the document image G1, that is, the trailing edge image E2, has a lower image density than the leading edge image E1, and has little influence on the document category classification result. Therefore, the analysis range H1 preferably includes a rear end image E2 without including the front end image E1. Thus, unlike the configuration in which the analysis range H1 includes the front end image E1 and does not include the rear end image E2, the edge image to be excluded from the analysis range H1 can be determined before reading the rear end of the document M. It is possible to start block division processing at an early stage, and to simplify control processing. Note that the same effect can be obtained if the analysis range H1 does not include the front end image E1 but includes the rear end image E2 as well as the image density of the rear end image E2 and the front end image E1. it can.

  In the pixel classification process and the block classification process of the above embodiment, the block B1 is classified into one of a blank pixel, a blank block, and a non-blank block from only the Y value of each pixel included in the block B1, and a non-blank block May be classified as either a color block or a black and white block from the Cb value and the Cr value.

  In the above-described embodiment, the image reading apparatus 1 is configured to perform both blank document detection and category classification related to the image color of the document image. However, the present invention is not limited to this, and the image reading apparatus 1 classifies the block into either a blank block or a non-blank block in the block classification process, and in the document classification process, converts the document image G1 into a blank document and a non-blank document. The structure classified into either of these may be sufficient. Thus, when a blank document is detected, it is possible to suppress the document image from being classified into an incorrect document category due to the influence of the edge image of the document.

  Further, the image reading apparatus 1 classifies the block into one of a color block, a gray block, and a black and white block in the block classification process, and converts the original image G1 into a color original, a gray original, and a black and white original in the original classification process. The structure classified into either may be sufficient. Thus, when determining the type of the image color of the document image, it is possible to suppress the document image from being classified into an incorrect document category due to the influence of the edge image of the document. The block category and the document category are not limited to color, gray, and black and white, and may include two of them. Moreover, you may classify | categorize into a single color and multiple colors.

  1: Image reading device 24: Reading device 31: Light source 33: Imaging unit 43: AFE B1: Block E1 to E4: Edge image G1: Document image G2: Scanned image H1: Analysis range M: Document

Claims (7)

A light source arranged at a position where light is irradiated obliquely from behind a relatively moving document, and a light source arranged side by side in a direction orthogonal to the direction of relative movement of the document and reflected by the document A reading device having an imaging unit composed of a plurality of imaging elements that receive the received light;
Output from the reading device, a data acquisition unit that four edge sides image of the original image to obtain data including the read image,
A control unit,
The controller is
A target range in the read image, the range along the length in the orthogonal direction including the edge image of the document image read first by the reading device is excluded from the read image, and the tip A pixel classification process for classifying each pixel belonging to a target range including a pair of edge images orthogonal to the edge image of the document image to be read into one of a plurality of pixel categories based on color information of the pixels;
Block division processing for dividing the target range into a plurality of rectangular blocks whose length in the orthogonal direction is longer than the length in the relative moving direction ;
A block classification process for classifying each of the plurality of blocks into one of a plurality of block categories based on the number of pixels classified into at least one of the plurality of pixel categories in the pixel classification process;
A document classification process for classifying the document image into one of a plurality of document categories based on the number of blocks classified into at least one block category among the plurality of block categories in the block classification process ;
An image processing apparatus having a configuration for executing the above. The image processing apparatus according to claim 1,
The controller is
In the block classification process, each of the plurality of blocks is classified into one of a block category including at least two of a color block, a black and white block, and a gray block,
In the document classification process, the document image is classified into one of a document category including at least two of a color document, a monochrome document, and a gray document .
Images processing device. The image processing apparatus according to claim 1,
The controller is
In the block classification process, each of the plurality of blocks is classified into one of a block category including a blank block and a non-blank block,
In the document classification process, the document image is classified into one of a document category including a blank document and a non-blank document .
Images processing device. The image processing apparatus according to claim 1,
In the block classification process, the control unit classifies each of the plurality of blocks into one of a block category including a blank block and a non-blank block, and each of the plurality of blocks is a color block, Performing a second classification process for classifying the block category including at least two of a black and white block and a gray block;
In the document classification process, the document image is classified into one of a document category including a blank document and a non-blank document based on the result of the first classification process, and the document is classified based on the result of the second classification process. Classify the image into one of the original categories, including at least two of color, black and white, and gray
Image processing device. The image processing apparatus according to any one of claims 1 to 4, wherein:
The controller is
A block size changing process for detecting the resolution of the read image and increasing the size of the block as the detection value of the resolution is higher to increase the number of pixels included in the block ;
An image processing apparatus having a configuration for executing   A reading device that reads the document while moving relative to the document;
  A data acquisition unit that acquires data of a read image including four edge images of a document image output from the reading device;
  A control unit,
  The controller is
  Of the pair of edge images that are the target range in the read image and are orthogonal to the relative movement direction of the reading device and the document in the document image, do not include the edge image having a relatively high image density. A pixel that classifies each pixel belonging to the target range including a pair of edge images orthogonal to the edge image having a relatively high image density into one of a plurality of pixel categories based on the color information of the pixel. Classification process;
  Block division processing for dividing the target range into a plurality of rectangular blocks whose length in the orthogonal direction is longer than the length in the relative moving direction;
  A block classification process for classifying each of the plurality of blocks into one of a plurality of block categories based on the number of pixels classified into at least one of the plurality of pixel categories in the pixel classification process;
  A document classification process for classifying the document image into one of a plurality of document categories based on the number of blocks classified into at least one block category among the plurality of block categories in the block classification process;
An image processing apparatus having a configuration for executing the above. A light source arranged at a position where light is irradiated obliquely from behind a relatively moving document, and a light source arranged side by side in a direction orthogonal to the direction of relative movement of the document and reflected by the document A reading device having an image pickup unit including a plurality of image pickup elements that receive the received light, and a data acquisition unit that acquires read image data including four edge images of the document image output from the reading device. In the computer of the image processing device,
A target range in the read image, the range along the length in the orthogonal direction including the edge image of the document image read first by the reading device is excluded from the read image, and the tip A pixel classification process for classifying each pixel belonging to a target range including a pair of edge images orthogonal to the edge image of the document image to be read into one of a plurality of pixel categories based on color information of the pixels;
Block division processing for dividing the target range into a plurality of rectangular blocks whose length in the orthogonal direction is longer than the length in the relative moving direction ;
A block classification process for classifying each of the plurality of blocks into one of a plurality of block categories based on the number of pixels classified into at least one of the plurality of pixel categories in the pixel classification process;
A document classification process for classifying the document image into one of a plurality of document categories based on the number of blocks classified into at least one block category among the plurality of block categories in the block classification process ;
To execute,
Image processing program.

Images