JP4404042B2 - Image reading device - Google Patents

Description

  The present invention relates to an image reading apparatus, and more particularly to an image reading apparatus that reads an original while conveying the original.

  In recent years, in scanners, facsimiles, copiers, etc., a line sensor in which a plurality of photoelectric conversion elements are arranged in a main scanning direction is fixed and installed, and a line sensor is conveyed by conveying a document in a sub scanning direction orthogonal to the main scanning direction. There is an image reading apparatus that employs a reading method for reading a document. The image reading apparatus includes a transparent document table between the document and the line sensor in order to fix the conveyed document at the reading position. The light reflected from the original passes through the original table and is received by the line sensor. For this reason, if dust adheres to the document table, the line sensor reads the dust instead of reading the document, and there is a problem that noise is generated in the image data.

  Japanese Patent Application Laid-Open No. 2002-185767 (Patent Document 1) and Japanese Patent Application Laid-Open No. 2002-271671 (Patent Document 2) are techniques for detecting noise generated due to dust adhering to the document table from a read image. It is described in. In the technique described in Japanese Patent Laid-Open No. 2002-185767, a straight line perpendicular to the main scanning direction in image data is determined as noise appearing in the image data by reading dust on the document table. For this reason, the straight line actually perpendicular to the main scanning direction drawn on the document may be erroneously determined as noise.

Japanese Patent Laid-Open No. 2002-271631 discloses a first reading unit that reads a color image from a document, and corresponds to a spectral sensitivity different from that of the first reading unit, and subscans the document with respect to the first reading unit. Second reading means arranged offset in the direction, data comparison means for comparing the density value of the reading result by the first reading means and the density value of the reading result by the second reading means, and reading by the first reading means The result of reading only by the first reading means based on the edge detection means for judging whether or not the result includes an edge component in the main scanning direction of the document, and the comparison result by the data comparison means and the judgment result by the edge detection means An image reading apparatus provided with noise detecting means for detecting a noise component included in the image reading apparatus is described. The image reading apparatus described in Japanese Patent Application Laid-Open No. 2002-271631 needs to include a second reading unit in addition to the first reading unit that reads a color image from a document, and the number of image data to be processed increases and processing is performed. There is a problem that it becomes complicated.
JP 2002-185767 A Japanese Patent Application Laid-Open No. 2002-271631

  The present invention has been made to solve the above-described problems, and provides an image reading apparatus having improved accuracy in detecting noise generated by reading dust existing on a document table from an image obtained by reading the document. It is.

  In order to achieve the above-described object, according to one aspect of the present invention, an image reading apparatus includes filters having different spectral sensitivities, arranged in a predetermined order at a distance in the sub-scanning direction, and A plurality of line sensors that scan in the sub-scanning direction, a document table provided between the document and the plurality of line sensors, and a plurality of lines at a relative speed different from the relative speed between the document and the plurality of line sensors. A moving means that moves relative to the sensor, a line-to-line correction means that synchronizes a plurality of data output from a plurality of line sensors so as to be pixels read from the same position of the document, and a plurality of lines synchronized from the line-to-line correction means. Noise pixel detecting means for sequentially inputting the data of each line, the noise pixel detecting means is a main scanning direction in which the pixel value changes in the main scanning direction from each of a plurality of data A main scanning direction edge detecting unit for detecting a wedge, a gradation detecting unit for detecting a gradation region in which a pixel value of a predetermined number of pixels continuous in the sub scanning direction changes from each of a plurality of data, and a main scanning direction edge detecting unit Noise in the main scanning direction edge including the first gradation area detected from the data in which the main scanning direction edge is detected by the gradation detecting means, which is detected from only one of the plurality of data. First noise detecting means.

  According to the present invention, the plurality of line sensors that have filters having different spectral sensitivities and that scan the document in the sub-scanning direction are arranged in a predetermined order at a distance in the sub-scanning direction. A document table provided between the plurality of line sensors moves relative to the plurality of line sensors at a relative speed different from the relative speed between the document and the plurality of line sensors. Since the document table moves relative to the plurality of line sensors at a relative speed different from the relative speed between the document and the plurality of line sensors, dust on the document table is read at different positions on the document by each of the plurality of line sensors. . This image reading apparatus synchronizes a plurality of data output from a plurality of line sensors so as to be pixels read from the same position of the document. Then, the image reading apparatus detects a main scanning direction edge in which the pixel value changes in the main scanning direction from each of the plurality of data, and the pixel values of a predetermined number of pixels continuous in the sub-scanning direction from each of the plurality of data. A main scanning direction including a first gradation area detected from data in which main gradation direction edges are detected by detecting a changing gradation area and detected from only one of a plurality of data. Let the edge be noise. Since dust on the document table is read at different positions on the document by each of the plurality of line sensors, the data obtained by reading the dust on the document table is the main scanning direction detected from only one of the plurality of data. Become an edge. Furthermore, since the document table moves relative to the plurality of line sensors, the main scanning direction edge in which the pixel value at the edge of the edge changes among the main scanning direction edges detected from only one of the plurality of data. There is a case. For this reason, it is possible to provide an image reading apparatus with improved accuracy in detecting noise generated by reading dust existing on the document table from an image obtained by reading the document.

  Preferably, the first noise detection means uses noise in the main scanning direction edge after the line to which the first gradation area belongs, which is detected from the data in which the main scanning direction edge is detected by the gradation detection means.

  Preferably, the first noise detecting means is provided on the condition that the second gradation area is detected after the line to which the first gradation area is detected, which is detected from the data in which the edge in the main scanning direction is detected by the gradation detecting means. The main scanning direction edge after the line to which the first gradation area belongs and before the line to which the second gradation area belongs is defined as noise.

  Preferably, the first gradation area and the second gradation area are opposite in the direction in which the pixel value changes.

  Preferably, the noise pixel detecting unit detects a sub-scanning direction edge detecting unit that detects a sub-scanning direction edge whose pixel value changes in the sub-scanning direction in a region around the main scanning direction edge detected by the main scanning direction edge detecting unit. If the main scanning direction edge is detected from only one of a plurality of data by the main scanning direction edge detecting means, the main scanning is performed on condition that the sub scanning direction edge is detected by the sub scanning direction edge detecting means. And a second noise detecting means that uses the direction edge as noise.

  According to this invention, when the main scanning direction edge is detected from only one of a plurality of data, the image reading device changes the pixel value in the sub-scanning direction to the area around the detected main scanning direction edge. The edge in the main scanning direction is set as noise on the condition that the edge in the sub scanning direction is detected. When the document has a high density region and a low density region, the edge in the sub scanning direction indicates a boundary where the position at which the document is read changes from the high density region to the low density region or from the low density region to the high density region in the sub scanning direction. . Even if high-concentration dust adheres to the platen, noise generated by dust is not detected as an edge in the main scanning direction while the high-density area of the document is being read. The generated noise is detected as an edge in the main scanning direction. Conversely, even if low-density dust adheres to the platen, noise generated by dust is not detected as an edge in the main scanning direction while reading the low-density area of the document, but while high-density area is being read. The noise generated by dust is detected as an edge in the main scanning direction. In this case, since the change in the pixel value at the end of the main scanning direction edge is small, the main scanning direction edge is generated by dust on the condition that the sub scanning direction edge is detected around the main scanning direction edge. Detect as noise. For this reason, it is possible to further improve the accuracy of detecting noise generated by reading dust existing on the document table from an image obtained by reading the document.

  Preferably, the second noise detecting means uses noise in the main scanning direction edge after the line to which the sub scanning direction edge detected by the sub scanning direction edge detecting means belongs.

  Preferably, the second noise detecting means detects the third gradation after the line to which the first sub-scanning direction edge belongs, on condition that the third gradation area is detected after the line to which the edge in the sub-scanning direction belongs by the gradation detecting means. The edge in the main scanning direction before the line to which the region belongs is defined as noise.

  Preferably, the second noise detection unit converts the data in the main scanning direction edge to a line different from the one data in which the main scanning direction edge is detected by the main scanning direction edge detection unit among the plurality of data. The edge in the main scanning direction is defined as noise on the condition that an edge in the sub scanning direction in which the pixel value changes is detected.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, the same parts are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

  FIG. 1 is a perspective view of an MFP (Multi Function Peripheral) including an image reading apparatus according to one embodiment of the present invention. Referring to FIG. 1, MFP 100 includes an image reading device 10 for reading a document, and an image forming device 20 provided below the image reading device 10. A part of the image reading apparatus 10 is housed in a main body 103 and includes an automatic document feeder (ADF) 101. The image forming apparatus 20 is housed below the image reading apparatus 10 in the main body 103, and forms an image on a recording medium such as paper based on image data that the image reading apparatus 10 reads and outputs a document. MFP 100 includes a communication interface for connecting to a network such as a facsimile, a local area network (LAN), and a public switched telephone network (PSTN).

  FIG. 2 is a diagram showing an outline of the internal configuration of the image reading apparatus 10. The ADF 101 includes a timing roller pair 201 for transporting the document 200 to the document reading position L, an upper restricting plate 203 for guiding document transportation in the vicinity of the document reading position L, and the document 200 that has passed through the document reading position L. And a pair of rollers 202 for conveying the document 200 to discharge the document.

  The ADF 101 separates one document from the top of the stacked plurality of documents 200 and supplies it to the timing roller pair 201. Therefore, the ADF 101 conveys a plurality of documents 200 one by one to the document reading position L in order.

  A portion housed in the main body 20 of the image reading apparatus 10 includes a document table 205 made of a transparent member, a sheet passing guide 207 that forms a part of a document transport path, and a light source for irradiating light. 206, a reflection member 208 that reflects light from the light source, a reading unit 213 in which three line sensors are arranged in the sub-scanning direction, and a reflection mirror that reflects reflected light from the document and guides it to the reading unit 213. 209, a lens 211 for forming an image of reflected light from the reflection mirror 209 on the reading unit 213, an image processing unit 215 for processing image data output from the reading unit 213, and the document table 205 are vibrated. And a motor control unit 217 for controlling the driving of the motor 219 based on control data from the image processing unit 215.

  The document 200 is conveyed between the document table 205 and the upper regulating plate 203 in the direction of the arrow D1 by the timing roller pair 201. Then, the image is read by the reading unit 213 at the document reading position L while the document is being conveyed. The direction in which the ADF 101 transports the document is the sub-scanning direction at the document reading position L. The motor control unit 217 drives the motor 219 during the image reading operation to vibrate the document table 205 in the direction of the arrow D2. The vibration direction of the document table 205 and the sub-scanning direction are substantially parallel.

  The reading unit 213 includes three line sensors. Each of the three line sensors includes a plurality of photoelectric conversion elements arranged in a main scanning direction substantially perpendicular to the sub-scanning direction. Each of the three line sensors has filters having different spectral sensitivities. Light reflected from the original passes through the filter and is received by a plurality of photoelectric conversion elements. Specifically, each of the three line sensors has a filter that transmits light of each wavelength of red (R), green (G), and blue (B). Therefore, the line sensor having a red (R) filter outputs R data indicating the intensity of red light out of the light reflected from the document, and the line sensor having a green (G) filter reflects from the document. The line data having the blue (B) filter outputs B data indicating the intensity of the blue light out of the light reflected from the document.

  The three line sensors are arranged in a predetermined order at a predetermined distance in the sub-scanning direction. Here, the document reading lines are arranged in the order of red, green, and blue in the document transport direction with a distance of 3 lines in terms of the document reading line. Note that the intervals and the order in which the three line sensors are arranged are not limited to these.

  Since the three line sensors are arranged in the order of red, green, and blue at a distance of three lines, the three line sensors simultaneously receive light reflected at different positions on the document. Therefore, the light reflected at a certain position of the document is first received by the line sensor that receives red light, and then received by the line sensor that receives green light after the document is conveyed for three lines, and the document is further received. After being conveyed for three lines, it is received by a line sensor that receives blue light. This delay is adjusted by an image processing unit 215 described later.

  In the present embodiment, the reading unit 213 is provided with three line sensors. However, four or more line sensors may be provided.

  FIG. 3 is a perspective view showing a mechanism for vibrating the document table. Referring to FIG. 3, document table 205 is held by document table holder 221. The document table holder 221 is held by the guide 220 so as to be slidable in the sub-scanning direction. The guide 220 is fixed to the main body of the image reading apparatus 10. Two arms 222 are joined to one surface of the document table holder 221. The other end of the arm 222 has a circular hole.

  Two cams 223 are attached to the shaft 224 at positions corresponding to the two arms 222. A gear 225 is attached to one end of the shaft 224. The gear 225 is disposed so as to mesh with a gear 226 joined to the drive shaft of the motor 219 with a belt. When the motor 219 rotates, the rotation is transmitted to the gear 226 via the belt, and the gear 226 rotates. As the gear 226 rotates, the gear 225 and the shaft 224 rotate. The cam 223 is disposed in a circular hole provided in the arm 222. Therefore, the rotational motion of the two cams 223 accompanying the rotation of the shaft 224 is converted into the reciprocating motion of the document table holder 221. The mechanism for vibrating the document table 205 is not limited to this. For example, a mechanism using a drive source that causes a linear motion such as a piston using electromagnet, air pressure, hydraulic pressure, or the like may be used.

  The document table 205 vibrates in parallel with the sub-scanning direction. While the document table 205 is moving in the direction opposite to the document conveyance direction, the document table 205 and the document are moved in the opposite direction. Therefore, the relative speed of the document table 205 to the line sensor and the relative value of the document to the line sensor are compared. The speed is different. On the other hand, while the document table 205 is moving in the document conveyance direction, the direction of the document table 205 and the document conveyance speed are the same. It is preferable to vary the speed. Here, the document table 205 is vibrated in parallel with the sub-scanning direction, but the direction is not limited to this.

  Here, the principle that the image reading apparatus 10 according to the present embodiment detects noise generated by dust attached to the document table 205 from the image data obtained by reading the document will be described. First, correction between lines will be described. FIG. 4 is a diagram for explaining interline correction. Referring to FIG. 4, the document is conveyed in the direction of the arrow in the drawing. Here, the three line sensors are separated from each other by a distance of three lines in the document conveyance direction in the order of a line sensor that receives red light, a line sensor that receives green light, and a line sensor that receives blue light. It is assumed that it is arranged. The output of the line sensor that receives red light is indicated by R, the output of the line sensor that receives green light is indicated by G, and the output of the line sensor that receives blue light is indicated by B.

  First, a partial image of a document is read by a line sensor that receives red light arranged at the most upstream in the document transport direction. Then, a partial image of the original is conveyed by a distance corresponding to four lines and read by a line sensor that receives green light. Further, a partial image of the original is conveyed by a distance corresponding to four lines and read by a line sensor that receives blue light.

  As described above, since a part of an image of an original is read at different times by the three line sensors, the three data output from the three line sensors at the same time are data output by reading different parts of the original. . In the inter-line correction, the output timings of the three data output by the three line sensors are adjusted so that the three line sensors read and output the same part of the document. Specifically, the output R is delayed by 8 lines, and the output G is delayed by 4 lines. The combined output is an output obtained by combining the output R, the output G, and the output B that are corrected between lines.

  FIG. 5 is a diagram for explaining the principle of detecting noise generated by dust adhering to the document table 205 from image data obtained by reading the document. Here, the document and document table 205 are conveyed in the direction of the arrow in the figure, and the movement speed of the document table 205 is the same as the document conveyance speed and is 1/5 times faster. Further, the size of dust attached to the document table 205 is set to one line.

  First, a line sensor that receives red light disposed at the most upstream in the document transport direction reads dust. Since the document table 205 moves in the same direction at a speed that is 1/5 times the document conveyance speed, dust moves by one line in the time required for the line sensor to read the document for five lines. Therefore, the line sensor that receives red light reads dust while the original is conveyed for five lines. The line sensor that receives red light reads 1/3 of the dust on the first first line while the document is being conveyed for 5 lines, and 2/3 of the dust on the next second line. Read, the next third line reads the entire dust, the next fourth line reads 2/3 of the dust, and the last fifth line reads 1/3 of the dust.

  Then, after the dust is conveyed by a distance of 3 lines, it is read by a line sensor that receives green light. Since the document table 205 moves in the same direction at a speed that is 1/5 times the document conveyance speed, the document is conveyed by 15 lines while the dust moves for three lines. For this reason, the time when the red line sensor reads dust and the time when the green line sensor reads dust are shifted by the time required to read the original for 15 lines. The line sensor that receives the green light reads dust while the document is conveyed for five lines. The line sensor that receives green light reads 1/3 of the dust on the first first line while the document is being conveyed by 5 lines, and 2/3 of the dust on the next second line. Read, the next third line reads the entire dust, the next fourth line reads 2/3 of the dust, and the last fifth line reads 1/3 of the dust.

  Further, after the dust is conveyed by a distance of three lines, it is read by a line sensor that receives blue light. Since the document table 205 moves in the same direction at a speed that is 1/5 times the document conveyance speed, the document is conveyed by 15 lines while the dust moves for three lines. Therefore, the time when the green line sensor reads dust and the time when the blue line sensor reads dust are shifted by the time required to read the original for 15 lines. The line sensor that receives blue light reads dust while the original is conveyed for five lines. The line sensor that receives blue light reads 1/3 of the dust on the first first line while the original is being conveyed by 5 lines, and 2/3 of the dust on the next second line. Read, the next third line reads the entire dust, the next fourth line reads 2/3 of the dust, and the last fifth line reads 1/3 of the dust.

  Then, due to the interline correction, the output R that the line sensor that receives red light reads and outputs dust is delayed by 8 lines, and the output G that the line sensor that receives green light reads and outputs dust is 4 Delayed by line. Therefore, in the combined output obtained by combining the output R, the output G, and the output B corrected between lines, if the output R from which dust is read, the output G from which dust is read, and the output B from which dust is read are at the same timing. Each is shifted by 16 lines.

  In the figure, white dust such as paper dust is attached to the document table 205, and a combined output when a black document is read is shown. In this case, although white dust is read, the combined output is not white but red, green, and blue, which are divided into three colors. In this way, dust adhering to the document table 205 is divided into output R, output G, and output B in the image. For this reason, it becomes possible to distinguish between a straight line in a document and noise generated by reading dust.

  Further, when the dust is read by each of the three line sensors, 1/3 of the dust is read on the first first line while the original is conveyed for 5 lines, and 2/2 of the dust is read on the next second line. 3 is read, the next third line reads the entire dust, the next fourth line reads 2/3 of the dust, and the last fifth line reads 1/3 of the dust. For example, if each of the three line sensors reads and outputs dust on the third line as 255, the pixel value read and output on the first and fifth lines becomes 255/3, and the second and The pixel value output by reading dust on the fourth line is 255 × 2/3. By detecting this change in pixel value, it is possible to distinguish between a straight line in a document and noise generated by reading dust.

  FIG. 6 is a plan view of the document table viewed from the back side. Referring to FIG. 6, document table 205 has a mark 205A at one end. The mark 205 </ b> A has a shape in which the length in the main scanning direction varies depending on the position in the sub-scanning direction and is a single color. Here, the mark 205A has a triangular shape and is black. Further, one side of the mark 205A is arranged in parallel with one side of the document table 205.

  By using the reading unit 213 or using a sensor provided separately from the reading unit 213 and fixed to the main body unit 103, the length of the mark 205A in the main scanning direction is detected, thereby reading the document table 205. It is possible to detect a relative position with respect to the part 213.

  FIG. 7 is a diagram showing a reading area on the document table 205 read by the reading unit 213. The reading unit 213 includes a line sensor 213R having a red (R) filter, a line sensor 213G having a green (G) filter, and a line sensor 213B having a blue (B) filter. The line sensors 213R, 213G, and 213B are arranged in the order of the line sensors 213R, 213G, and 213B in the document transport direction D1.

  The line sensor 213R receives light transmitted through the region 205R of the document table 205. The line sensor 213G receives light transmitted through the region 205G of the document table 205. The line sensor 213B receives light transmitted through the area 205B of the document table 205. In the areas 205R, 205G, and 205B, the line sensors 213R, 213G, and 213B are arranged so as to have an interval of three lines. The document first passes through the region 205R, then passes through the region 205G, and finally passes through the region 205B. Therefore, the light reflected at a certain position of the document is first received by the line sensor 213R that receives red light, then received by the line sensor 213G that receives green light, and finally the line that receives blue light. Light is received by the sensor 213B. As described above, the line sensors 213R, 213G, and 213B are arranged at a distance of three lines, so the line sensors 213R, 213G, and 213B do not simultaneously receive the light reflected at the same position on the document. .

  Here, it is assumed that dust 300 having a length of 4 lines or less is attached on the document table 205. In this case, since the document table 205 vibrates and moves in parallel with the sub-scanning direction, the dust 300 does not exist in two or more of the areas 205R, 205G, and 205B at the same time. FIG. 6 shows a case where the dust 300 exists in the area 205G. In this case, the light reflected by the dust 300 is received only by the line sensor 213G and not received by the line sensors 213R and 213B.

  Since the document table 205 vibrates, the document table 205 may move in the document conveyance direction D1, and the document table 205 may move in the direction opposite to the document conveyance direction D1. While the document table 205 moves in the document conveyance direction D1, dust moves in the order of the area 205R, then the area 205G, and finally the area 205B. Conversely, while the document table 205 moves in the direction opposite to the document conveyance direction D1, dust moves first in the order of the area 205B, then in the area 205G, and finally in the area 205R. Therefore, while the document table 205 moves in the document transport direction D1, the light reflected by the dust 300 is first received by the line sensor 213R, then received by the line sensor 213G, and finally the line sensor 213B. Is received. While the document table 205 moves in the direction opposite to the document conveyance direction D1, the light reflected by the dust 300 is first received by the line sensor 213B, then received by the line sensor 213G, and finally the line. Light is received by the sensor 213R.

  While the document table 205 is moving in the document conveyance direction, noise due to reading of dust is caused by the first R data output by the line sensor 213R, the next G data output by the line sensor 213G, and finally the line sensor. It appears in order in B data output by 213B. Further, when the document table 205 is moved in the direction opposite to the document conveyance direction, noise caused by reading dust is the B data output first by the line sensor 213B and then the G data output by the line sensor 213G. Data, and finally appear in the R data output by the line sensor 213R. In other words, the order of data in which noise generated by reading dust appears is determined by the moving direction of the document table 205. By determining the order in which noise is detected from R data, G data, and B data, the accuracy of detecting noise can be improved.

  FIG. 8 is a block diagram illustrating the configuration of the image processing unit of the image reading apparatus according to the present embodiment. With reference to FIG. 8, R data, G data, and B data are input to the image processing unit 215 from the reading unit 213. The image processing unit 215 includes an analog / digital conversion unit (A / D conversion unit) 251 for converting analog data R data, G data, and B data input from the reading unit 213 into digital signals, and illumination of the light source 206. Noise from shading correction unit 253 for correcting unevenness, inter-line correction unit 255, chromatic aberration correction unit 257 for correcting distortion in the main scanning direction by lens 211, and R data, G data, and B data, respectively. A noise detection processing unit 259 for detecting noise, a noise correction unit 260 for executing processing for removing noise, a control unit 263 for controlling the entire image processing unit 215, and an image output to the image forming apparatus 20 A printer interface 261. A position detector 265 for detecting the position of the document table 205 is connected to the controller 263. The position detection unit 265 detects the length in the main scanning direction of the mark 205 </ b> A that the document table 205 has.

  The inter-line correction unit 255 delays the R data by 8 lines and delays the G data by 4 lines. As a result, the R data, G data, and B data that the line sensors 213R, 213G, and 213B read and output the document are synchronized so as to correspond to the same line of the document. This is because the line sensors 213R, 213G, and 213B are arranged at a distance of three lines in the sub scanning direction as described above.

  The noise detection processing unit 259 receives R data, G data, and B data from the chromatic aberration correction unit 257. The noise detection processing unit 259 detects a noise pixel as a pixel from which dust is read for each of R data, G data, and B data input from the chromatic aberration correction unit 257. Then, logical data with the noise pixel set to “1” and the other pixels set to “0” is output to the noise correction unit 260 and the control unit 263. Details thereof will be described later.

  The noise correction unit 260 receives R data, G data, and B data from the chromatic aberration correction unit 257, and from the noise detection processing unit 259, logical data that sets the noise pixel to “1” and the other pixels to “0” is R data. , G data and B data are input. The noise correction unit 260 corrects the noise pixels of the R data, the G data, and the B data based on the logical data corresponding to the R data, the G data, and the B data. For each of the R data, G data, and B data, the noise correction unit 260 replaces the pixel value of the noise pixel with the pixel value of a pixel that is not a surrounding noise pixel, based on the corresponding logical data. You may make it replace with the average value of the some pixel which is not a surrounding noise pixel, the maximum value, or the minimum value. The noise correction unit 260 outputs R data, G data, and B data obtained by replacing noise pixels with peripheral pixels to the printer interface 261.

  The control unit 263 receives the position of the document table 205 from the position detection unit 265, and receives logical data from the noise detection processing unit 259 that sets the noise pixel to “1” and the other pixels to “0”. The control unit 263 specifies the position of dust on the document table 205 from these data. More specifically, the position of the document table 205 in the sub-scanning direction is specified from the position of the document table 205 and the line number of the logical data, and the position of the document table 205 in the main scanning direction is determined from the position of the noise pixel of the logical data. Identify.

  FIG. 9 is a diagram illustrating an example of RGB data after line-to-line correction. FIG. 9A shows an example of RGB data when black dust adhering to the document table is in the region 205R corresponding to the line sensor 213R while reading the white region of the document. Note that although black dust is detected as noise here, the dust may be a chromatic color. Although a case where a white document is read will be described as an example here, the color of the document is not limited to white and may be any other color.

  Referring to FIG. 9A, the line sensor 213R reads black dust, so that the R data output from the line sensor 213R has low brightness. Since the line sensors 213G and 213B read the white area of the document, the G data and B data output by the line sensors 213G and 213B have high brightness. Here, the output values of the three line sensors 213R, 213G, and 213B corresponding to the reflected light are referred to as brightness.

  The combination of RGB data shown in FIG. 9A is rarely read and output. On the other hand, among the RGB data combinations output by reading a document, the closest combination to the RGB data combination shown in FIG. 9A is when a blue-green region which is a complementary color of red is read. FIG. 9B is a diagram showing RGB data output by the reading unit 213 when a blue-green region of a document is read. The brightness of R data is greatly reduced, but the brightness of G data and B data is also reduced. There is a great difference between the RGB data shown in FIG. 9A and the RGB data shown in FIG. 9B whether the B data and the G data are affected or not. By detecting this difference, black dust can be detected as noise without erroneously detecting the blue-green line as noise.

  Therefore, a change in the brightness of the R data whose brightness is greatly reduced can be detected using the threshold value Ref1 (R). Further, a change in the brightness of the B data is detected using a threshold value Ref2 (B). The threshold value Ref2 (B) may be the smallest value among the following values. In the following, threshold values Ref2 (R), Ref2 (G), and Ref2 (B) are shown.

(1) When detecting achromatic dust with high brightness When reading blue-green, which is a complementary color of red, so that a blue-green line is not erroneously detected as noise, a line other than the line sensor 213R is detected. The difference between the lightness output from one of the sensors 213G and 213B and the maximum lightness value (255) may be set as Ref2 (G) and Ref2 (B). The brightness and maximum output by either one of the line sensors 213R and 213B other than the line sensor 213G when reading the magenta complementary color of green so that the magenta line is not erroneously detected as noise. Differences from brightness (255) may be set as Ref2 (R) and Ref2 (B). In order to prevent the yellow line from being erroneously detected as noise, the brightness and maximum brightness (one of the line sensors 213R and 213G other than the line sensor 213B output when reading yellow, which is a complementary color of blue) ( And Ref2 (R) and Ref2 (G).

(2) When detecting achromatic dust with low lightness When reading red, one of the line sensors 213G and 213B other than the line sensor 213R is detected so that the red line is not erroneously detected as noise. The difference between the lightness output by one and the minimum value (0) of lightness may be Ref2 (G) and Ref2 (B). The difference between the lightness output by one of the line sensors 213R and 213B other than the line sensor 213G and the minimum value (0) when green is read so that the green line is not erroneously detected as noise. Ref2 (R) and Ref2 (B) may be used. The difference between the lightness output by one of the line sensors 213R and 213G other than the line sensor 213B and the minimum value (0) when blue is read so that the blue line is not erroneously detected as noise. Ref2 (R) and Ref2 (G) may be used.

  In this way, a plurality of threshold values Ref2 (R), Ref2 (G), and Ref2 (B) are obtained, but the minimum values may be used.

  FIG. 10 is a block diagram illustrating a configuration of a noise detection processing unit of the image reading apparatus according to the present embodiment. Referring to FIG. 10, the noise detection processing unit 259 includes first brightness difference detection units 301R, 301G, and 301B for extracting regions having predetermined characteristics from input R data, G data, and B data, respectively. Second lightness difference detection units 302R, 302G, and 302B, detection result expansion processing units 303R, 303G, and 303B for extending the regions extracted by the second lightness difference detection units 302R, 302G, and 302B to the periphery; Sum elements 305R, 305G, 305B, AND elements 306R, 306G, 306B, gradation detectors 311R, 311G, 311B, background edge detectors 312R, 312G, 312B, OR elements 313R, 313G, 313B, Determination units 315R, 315G, 315B and detection area expansion processing units 316R, 316 , And a 316B.

  Each of R data, G data, and B data is input to the noise detection processing unit 259 one line at a time. In addition, R data, G data, and B data may be input collectively for a plurality of lines, or the entire image may be input collectively.

  The first brightness difference detection units 301R, 301G, and 301B differ only in the data that they handle, and their functions are the same, so the first brightness difference detection unit 301R will be described here. The first brightness difference detection unit 301R receives R data and a threshold value Ref1 (R). The first brightness difference detection unit 301R extracts a region having a first level predetermined feature from the R data. The region having the first level predetermined feature is a region where the change in lightness is small and a difference in lightness from the surrounding region is a threshold value Ref1 (R) or more. Such a region may have a size of one pixel or more. Here, a pixel included in a region having a predetermined feature at the first level is referred to as a first feature pixel.

  The first brightness difference detection unit 301R extracts an area having a first level predetermined feature using an edge extraction filter. The edge extraction filter includes a plurality of filters corresponding to a plurality of sizes of the edge region. The first lightness difference detection unit 301R performs the filter process for each of the plurality of filters, and compares the value obtained as a result with the threshold value Ref1 (R). When the absolute value of the value obtained as a result of the filter processing satisfies a condition larger than the threshold value Ref1 (R), the first brightness difference detection unit 301R sets the center pixel of the filter processing as the center pixel of the edge region, The size of the edge region is obtained from the edge extraction filter that satisfies the condition.

  FIG. 11 is a diagram illustrating an example of an edge extraction filter. Here, the edge extraction filter is described using R data, but the same edge extraction filter as that used for R data is also used for G data and B data. FIG. 11A shows an edge extraction filter used for detecting an edge region having a size of one pixel. FIG. 11B shows an edge extraction filter used for detecting an edge region having a size of 2 to 3 pixels. FIG. 11C shows an edge extraction filter used for detecting an edge region having a size of 4 to 5 pixels or less. FIG. 11D shows an edge extraction filter used for detecting an edge region having a size of 6 to 7 pixels or less. FIG. 11E shows an edge extraction filter used to detect an edge region having a size of 8 to 9 pixels or less. FIG. 11G shows an edge extraction filter used for detecting an edge region having a size of 12 to 13 pixels or less. FIG. 11H shows an edge extraction filter used to detect an edge region having a size of 14 to 15 pixels or less. The conditions for establishing these edge extraction filters are as follows.

(1) The determination condition for the edge region with high brightness is when the value obtained by subtracting the average brightness of pixel C from the average brightness of pixels A and B is equal to or greater than threshold value Ref1 (R).
Average (pixel A and pixel B) -average (pixel C)> Ref1 (R)
The central pixel in this case is the pixel having the maximum brightness among the pixels A, B, and C.

(2) The determination condition for the edge region with low brightness is when the value obtained by subtracting the average brightness of the pixels A and B from the average brightness of the pixel C is equal to or greater than the threshold value Ref1 (R).
Average (pixel C) -average (pixel A and pixel B)> Ref1 (R)
The central pixel in this case is a pixel having the minimum brightness among the pixel A, the pixel B, and the pixel C.

  Returning to FIG. 10, the first brightness difference detection unit 301 </ b> R compares the value calculated using the above-described edge extraction filter with the threshold value Ref <b> 1 (R). The first brightness difference detection unit 301R outputs logical data in which the first feature pixel is set to “1” and the other pixels are not set to “0” to the AND element 306R. The logical data corresponds to the pixel value of the R data and is configured with the same number as the pixel value. The first brightness difference detection unit 301G compares the value calculated using the above-described edge extraction filter with the threshold value Ref1 (G). The first lightness difference detection unit 301G outputs logical data in which the first feature pixel is “1” and the other pixels are “0” to the AND element 306G. The logical data corresponds to the pixel value of the G data and is configured with the same number as the pixel value. The first brightness difference detection unit 301B compares the value calculated using the above-described edge extraction filter with the threshold value Ref1 (B). The first brightness difference detection unit 301B outputs logical data in which the first feature pixel is “1” and the other pixel is “0” to the AND element 306B. The logical data corresponds to the pixel value of the B data and is configured with the same number as the pixel value.

  The second lightness difference detection units 302R, 302G, and 302B differ only in the data to be handled, and their functions are the same, so the second lightness difference detection unit 302R will be described here. The second brightness difference detection unit 302R receives R data and a threshold value Ref2 (R). The second brightness difference detection unit 302R extracts a region having a second level predetermined feature from the R data. The region having the second level predetermined feature is a region where the change in lightness is small, and is a region where the difference in lightness from the surrounding region is equal to or greater than the threshold value Ref2 (R). Such a region may have a size of one pixel or more. Here, a pixel included in an area having a second level predetermined feature is referred to as a second feature pixel. The threshold value Ref2 (R) is smaller than the threshold value Ref1 (R).

  The second brightness difference detection unit 302R extracts an area having a second level predetermined feature using an edge extraction filter. The edge extraction filter includes a plurality of filters corresponding to a plurality of sizes of the edge region. The second lightness difference detection unit 302R executes the filter process for each of the plurality of filters, and compares the value obtained as a result with the threshold value Ref2 (R). Then, when the condition that the absolute value of the value obtained as a result of the filter process is larger than the threshold value Ref2 (R) is satisfied, the second brightness difference detection unit 302R sets the center pixel of the filter process as the center pixel of the edge region. The size of the edge region is obtained from the edge extraction filter that satisfies the conditions.

  The second brightness difference detection unit 302R compares the value calculated using the edge extraction filter shown in FIG. 11 with the threshold value Ref2 (R). The second brightness difference detection unit 302R outputs logical data in which the second feature pixel is set to “1” and other pixels are set to “0” to the detection result extension processing unit 303R. The logical data corresponds to the pixel value of the R data and is configured with the same number as the pixel value. The detection result expansion processing unit 303R expands an area having a predetermined feature of the second level by setting a pixel around the second feature pixel extracted by the second brightness difference detection unit 302R as the second feature pixel. . That is, the value of the pixel having the value “0” around the pixel having the logical data value “1” input from the second brightness difference detection unit 302R is changed to “1”. Thereby, the accuracy of noise detection can be improved. The detection result expansion processing unit 303R outputs the logical data in which the area is expanded to the negative OR elements 305G and 305B. The detection result expansion processing units 303G and 303B differ only in the data to be handled, and their functions are the same as those of the detection result expansion processing unit 303R. Therefore, description thereof will not be repeated here.

  The logical data obtained by extending the area from each of the detection result expansion processing units 303G and 303B is input to the negative logical sum element 305R. The negative logical sum element 305R outputs logical data obtained by inverting the logical sum of the two input logical data to the logical product element 306R. In other words, the negative OR element 305R performs logical AND on logical data that sets a pixel that is not the second feature pixel to “1” and at least one pixel that is the second feature pixel to “0” in either G data or B data. Output to the element 306R. The AND element 306R outputs the logical product of the logical data input from the first brightness difference detection unit 301R and the logical data input from the negative OR element 305R to the determination unit 315R. That is, the AND element 306R sets the pixel that is the first feature pixel in the R data and is not the second feature pixel expanded in either the B data or the G data to “1”, and sets the other pixels to “0”. The logical data to be output is output to the determination unit 315R. Here, logical data output from the AND element 306R to the determination unit 315R is referred to as EDGE. The logical data EDGE output from the AND element 306R indicates an edge pixel that is the first feature pixel in the R data and is not the second feature pixel expanded in either the G data or the B data. This edge pixel is a pixel in which dust attached to the document table 205 is read by the line sensor 213R, and constitutes a plurality of edges in the main scanning direction that are continuous in the sub scanning direction.

  Since the logical OR elements 305G and 305B only handle different data and their functions are the same as those of the logical NOR element 305R, description thereof will not be repeated here.

  Since the gradation detection units 311R, 311G, and 311B have the same functions except for the handled data, the gradation detection unit 311R will be described here. The gradation detection unit 311R receives R data and a threshold value Ref3 (R). The gradation detection unit 311R detects the presence or absence of changes in the pixel values of a plurality of pixels arranged in the sub-scanning direction from the R data. The threshold value Ref3 (R) is defined as SignRef (K), DRRef1 (K), DRRef2 (K), DRRef3 (K), DRRef4 (K), SignRef (W), DRRef1 (W), DRRef2 (W), DRRef3. (W), DRRef4 (W) is included.

  FIG. 12 is a functional block diagram illustrating an outline of functions of the gradation detection unit 311R. Referring to FIG. 12, gradation detection unit 311 </ b> R includes a primary differential filter calculation unit 331, differential comparison filter units 332 and 333, and an OR element 334. The primary differential filter calculation unit 331 receives R data. The primary differential filter calculation unit 331 differentiates the R data input using the primary differential filter shown in FIG. 12 and outputs the differential values to the differential comparison filter units 332 and 333, respectively. The primary differential filter is a pixel whose processing target is a position indicated by hatching in the center of the drawing.

  The differential comparison filter units 332 and 333 detect the presence / absence and direction of change of the pixel value using the differential comparison filter shown in FIG. With reference to FIG. 13, pixels indicated by hatching are pixels to be processed, and reference numerals are assigned to the respective pixels for description. The pixel D05 is a processing target pixel, and the pixel on the previous line is D04. The differential comparison filter units 332 and 333 detect the presence / absence of the change in the pixel value and the direction of the change using the first-order differential values of the pixels in the subsequent 10 lines including the processing target pixel.

The differential comparison filter unit 332 detects the direction in which the pixel value decreases in order to detect black noise. Specifically, the differential comparison filter unit 332 determines the pixel value if the number of pixels in which the first-order differential values of the pixels in the subsequent 10 lines including the processing target pixel are plus signs exceeds a predetermined threshold value SignRef (K). Judge that there is a decreasing change. However, the differential comparison filter unit 332 uses the dynamic range to satisfy the following constraints in order to prevent erroneous detection of characters as noise. Here, the change in the pixel value of the pixels in the subsequent 10 lines including the pixel to be processed is detected, but the number of pixels for detecting the change in the pixel value is not limited to 10 pixels. The number of pixels for detecting a change in value may be determined by the moving speed of the 205th document.
(1) The dynamic range of the pixel values of the pixels on the five lines before the processing target pixel D05 is a predetermined threshold value DRRef1 (K).
MAX (D00, D01, D02, D03, D04) −MIN (D00, D01, D02, D03, D04) <DRRef1 (K)
Note that MAX (D00, D01, D02, D03, D04) indicates the maximum pixel value of the pixels D00, D01, D02, D03, D04, and MIN (D00, D01, D02, D03, D04) indicates the pixel. The minimum value of the pixel values of D00, D01, D02, D03, and D04 is shown.
(2) The dynamic range of the pixel values of the pixels of the subsequent five lines including the processing target pixel D05 is a predetermined threshold value DRRef2 (K).
MAX (D05, D06, D07, D08, D09) −MIN (D05, D06, D07, D08, D09) <DRRef2 (K)
(3) The dynamic range of the pixel values of the pixels on the 6th line to the 10th line after the processing target pixel D05 is a predetermined threshold value DRRef3 (K).
MAX (D10, D11, D12, D13, D14) −MIN (D10, D11, D12, D13, D14) <DRRef3 (K)
(4) The dynamic range of the pixel values of the pixels of the 6th to 10th lines after the processing target pixel D05 is a predetermined threshold value DRRef4 (K).
MAX (D15, D16, D17, D18, D19) −MIN (D15, D16, D17, D18, D19) <DRRef4 (K)
The differential comparison filter unit 332 includes the number of pixels in which the first-order differential values of the pixels in the last 10 lines including the processing target pixel exceed a predetermined threshold value SignRef (K), and the above four constraint conditions are satisfied. On the condition that all are satisfied, logical data with the value of the processing target pixel being “1” is output to the logical sum element 334. When the logical data does not satisfy any one of the above conditions, the value of the processing target pixel is “0”.

The differential comparison filter unit 333 detects the direction in which the pixel value increases in order to detect white noise. Specifically, the differential comparison filter unit 333 determines the pixel value if the number of pixels in which the first-order differential values of the pixels in the subsequent 10 lines including the pixel to be processed have a minus sign exceeds a predetermined threshold value SignRef (W). Judge that there is a rising change. However, the differential comparison filter unit 333 uses the dynamic range to satisfy the following constraints in order to prevent erroneous detection of characters as noise.
(1) The dynamic range of the pixel values of the pixels on the five lines before the processing target pixel D05 is a predetermined threshold value DRRef1 (W).
MAX (D00, D01, D02, D03, D04) −MIN (D00, D01, D02, D03, D04) <DRRef1 (W)
(2) The dynamic range of the pixel values of the five lines of pixels including the processing target pixel D05 is a predetermined threshold value DRRef2 (W).
MAX (D05, D06, D07, D08, D09) −MIN (D05, D06, D07, D08, D09) <DRRef2 (W)
(3) The dynamic range of the pixel values of the pixels on the 6th to 10th lines after the processing target pixel D05 is a predetermined threshold value DRRef3 (W).
MAX (D10, D11, D12, D13, D14) −MIN (D10, D11, D12, D13, D14) <DRRef3 (W)
(4) The dynamic range of the pixel values of the 6th to 10th pixel after the processing target pixel D05 is a predetermined threshold value DRRef4 (W).
MAX (D15, D16, D17, D18, D19) −MIN (D15, D16, D17, D18, D19) <DRRef4 (W)
The differential comparison filter unit 333 has the number of pixels in which the first-order differential values of the pixels in the subsequent 10 lines including the processing target pixel have a minus sign exceed a predetermined threshold value SignRef (W), and the four constraint conditions On the condition that all the conditions are satisfied, logical data with the value of the processing target pixel being “1” is output to the logical sum element 334. When the logical data does not satisfy any one of the above conditions, the value of the processing target pixel is “0”.

  The logical sum element 334 receives logical data from each of the differential comparison filter units 332 and 333, calculates a logical sum of the logical data, and outputs logical data obtained by the calculation. The logical data output from the logical sum element 334 is called GRA. The logical data GRA output from the gradation detection unit 311R specifies a gradation region in which the pixel value changes in the sub-scanning direction.

  Next, the background edge detection unit 312R will be described. The background edge detection units 312G and 312B differ from the background edge detection unit 312R only in the data handled, and their functions are the same as those of the background edge detection unit 312R. Therefore, description thereof will not be repeated here.

  FIG. 14 is a diagram for explaining the principle of detecting the background edge region. FIG. 14A is a diagram illustrating noise appearing in an image when a document has a black solid area on a white background and white dust is attached to the document table 205. In FIG. 14A, a region with low brightness in the R data output by the line sensor 213R is indicated by hatching. Since the R data that the line sensor 213R reads and outputs the background of the document is close to the R data that reads and outputs the white dust, the first brightness difference detection unit 301R reads and outputs the white dust R Data cannot be detected as the first feature pixel. For this reason, in the portion indicated by the dotted line from the noise front end in the figure, the line sensor 213R does not detect white dust, but it is not detected as the first feature pixel. On the other hand, since the R data that the line sensor 213R reads and outputs the black solid area of the document is different from the R data that reads and outputs the white dust, the first brightness difference detection unit 301R reads the white dust. R data to be output is detected as a first feature pixel. A line segment shown as an edge in the main scanning direction in the figure indicates the first feature pixel. If white dust does not adhere to the document table 205, the line sensor 213R reads it as a rectangular black solid, and the line sensor 213R reads the white dust so that the line sensor 213R makes the black solid area rectangular. Cannot be read. When attention is focused on the target pixel in the figure, white dust covers the entire reading area of the line sensor 213R at this position, and therefore there is no change in pixel value between pixels adjacent to the target pixel in the sub-scanning direction. For this reason, the target pixel is not detected as a gradation region by the gradation detection unit 311R.

  On the other hand, a plurality of pixels adjacent to the target pixel on the same line as the target pixel constitute an edge in the sub-scanning direction that changes from a white background of the document to a solid black region. The background edge detector 312R detects an edge in the sub-scanning direction that should be on the same line as the target pixel.

  However, if only the sub-scanning direction edge existing on the same line as the target pixel is detected, a part of the character may be erroneously detected as noise. A character including a straight line parallel to the sub-scanning direction and a straight line parallel to the main-scanning direction (for example, the letter “T”), which is a single color character of red, blue, green, blue-green, magenta, and yellow . FIG. 14B is a diagram for describing an example in which a character is erroneously detected as noise. In the figure, the red letter “T” is shown. In many cases, the character “T” has a horizontal straight line parallel to the main scanning direction and a vertical straight line parallel to the sub-scanning direction. For this reason, the main scanning direction edge including the target pixel is detected from the R data by the first brightness difference detection unit 301R described above. Further, the horizontal line of the character “T” is detected from the R data as an edge in the sub-scanning direction existing on the same line as the target pixel. Further, depending on the color of the character “T” and the values of Ref2 (G) and Ref2 (B), the main scanning direction edge including the target pixel may not be detected from the G data by the second brightness difference detection unit 302G. The second brightness difference detection unit 302B may not be detected from the B data. Therefore, in order not to detect the main scanning direction edge including the target pixel of the R data as noise, by comparing the difference in brightness between the target pixel and the pixel on the previous line in the G data and B data with a threshold value. , An edge in the main scanning direction including the target pixel detected from the R data is prevented from being erroneously determined as noise. Since the colors of the characters are all the same color, if the edge in the main scanning direction including the target pixel is a vertical straight line of the character “T”, the brightness of the target pixel and the pixel of the preceding line are determined in G data and B data. This is because the difference is small. In addition, since the values of Ref2 (G) and Ref2 (B) can be reduced, erroneous detection can be prevented and detection accuracy can be improved.

  The background edge detector 312R receives R data and a threshold value Ref4 (R). The threshold value Ref4 (R) includes BERef1 (K), BERef1 (W), and BEREF2.

  FIG. 15 is a diagram illustrating an example of the background edge detection filter. FIG. 15A shows a background edge detection filter for detecting noise appearing in R data due to dust having a length of one pixel in the main scanning direction. FIG. 15B shows a background edge detection filter for detecting noise appearing in R data due to dust having a length of 2 to 3 pixels in the main scanning direction. FIG. 15C shows a background edge detection filter for detecting noise appearing in the R data due to dust having a length of 4 to 5 pixels in the main scanning direction. FIG. 15D shows a background edge detection filter for detecting noise appearing in R data due to dust having a length of 6 to 7 pixels in the main scanning direction. FIG. 15E shows a background edge detection filter for detecting noise appearing in R data due to dust having a length of 8 to 9 pixels in the main scanning direction. FIG. 15F illustrates a background edge detection filter for detecting noise that appears in R data due to dust having a length of 10 to 11 pixels in the main scanning direction. FIG. 15G illustrates a background edge detection filter for detecting noise that appears in R data due to dust having a length of 12 to 13 pixels in the main scanning direction. FIG. 15H is a background edge detection filter for detecting noise that appears in R data due to dust having a length of 14 to 15 pixels in the main scanning direction.

  Here, the background edge detection filter shown in FIG. Referring to FIG. 15 (A), the background edge detection filter is denoted by a reference numeral for specifying each pixel for explanation. The background edge detection filter processes five lines of R data, G data, and B data. The target pixel for which the background edge is to be detected is indicated by the symbol TP.

  The background edge detection unit 312R uses the following determination formula for the R data in order to detect the sub-scanning direction edge.

(1) A judgment formula for detecting black dust.
MIN (A1)> BERef (B) AND (MIN (A1) −MIN (B1))> BERef1 (K) (1)
Or
MIN (A2)> BERef (B) AND (MIN (A2) −MIN (B2))> BERef1 (K) (2)
Here, MIN (A1) indicates the minimum value of the pixel values of a plurality of pixels with the code A1, MIN (A2) indicates the minimum value of the pixel values of the plurality of pixels with the code A2, and there are a plurality of MIN (B1). The minimum value of the pixel value of the pixel of the code | symbol B1 is shown, MIN (B2) shows the minimum value of the pixel value of the pixel of the code | symbol B2 which exists in multiple numbers.

(2) A judgment formula for detecting white dust.
MAX (A1)> BERef (W) AND (MAX (A1) -MAX (B1))> BERef1 (W) (3)
Or
MAX (A2)> BERef (W) AND (MAX (A2) -MAX (B2))> BERef1 (W) (4)
Furthermore, the background edge detection unit 312R uses the following determination formula for each of the G data and the B data in order to prevent the main scanning direction edge including the target pixel from being erroneously determined as noise.

(1) A judgment formula for preventing erroneous detection as black dust.
TP-MIN (C)> BERef2 (5)
(2) A judgment formula for preventing false detection as white dust.
MAX (C) -TP> BERef2 (6)
Here, TP indicates the pixel value (lightness) of the R data or G data of the target pixel.

  The background edge detection unit 312R has a condition that either the determination formula (1) or the determination formula (2) is satisfied in the R data, and that the determination formula (5) is satisfied in each of the G data and the B data. In addition, the target pixel TP is determined as the background edge. In addition, the background edge detection unit 312R satisfies that either the determination formula (3) or the determination formula (4) is satisfied in the R data, and the determination formula (6) is satisfied in each of the G data and the B data. As a result, the target pixel TP is determined as a background edge.

  The background edge detection unit 312R selects one of a plurality of background edge detection filters from the size of the filter used when the first feature difference detection unit 301R detects the first feature pixel. For example, the filter used when detecting the first feature pixel by the first brightness difference detection unit 301R may be a filter used to detect an edge region having a size of 2 to 3 pixels shown in FIG. For example, the background edge detection filter shown in FIG. The background edge detection unit 312R uses the selected background edge detection filter to determine that the target pixel to be processed has a sub-scanning direction edge on the same line as the target pixel and is not a part of the character. It is determined whether or not. If the target pixel is a background edge, the background edge detection unit 312R outputs logical data in which the value of the target pixel is “1” to the logical sum element 313R. In the logical data, if the target pixel is not a background edge, the value of the target pixel is “0”. The logical data output from the background edge detection unit 312R is referred to as BE.

  Returning to FIG. 10, the logical sum element 313R receives the logical data GRA from the gradation detection unit 311R and the logical data BE from the background edge detection unit 312R. The OR element 313R calculates a logical sum of the logical data GRA and the logical data BE, and outputs the logical data obtained by the calculation to the determination unit 315R. That is, the logical sum element 313R outputs logical data having a value of “1” when the target pixel is included in the gradation area or when the target pixel is a background edge, and a value of “0” when the target pixel is not any of them. Output. The logical data output from the logical sum element 313R is referred to as VAL. Since the logical sum elements 313G and 313B have the same function as the logical sum element 313R, description thereof will not be repeated here.

  The determination unit 315R receives the logical data EDGE from the logical product element 306R and the logical data VAL from the logical sum element 313R. The determination unit 315R includes a buffer memory for performing a logical operation between the same pixels on the same line using the logical data EDGE input from the logical product element 306R and the logical data VAL input from the logical sum element 313R. Yes.

  The determination unit 315R calculates the logical product of the logical data EDGE and the logical data VAL and sets it as the logical data flag. Further, the determination unit 315R calculates a logical product of the logical data FLAG of the previous line and the logical data EDGE and sets it as the logical data FLAG. The determination unit 315R calculates a logical sum of the calculated logical data flag and the logical data FLAG, and outputs the logical data FLAG to the detection area expansion processing unit 316R. That is, the determination unit 315R outputs the logical data FLAG having the pixel value “1” to the detection area expansion processing unit 316R if the logical data EDGE is “1” and the logical data VAL is “1”. In addition, even if the logical data EDGE is “1” and the logical data VAL is “0”, the determination unit 315R has “1” as the logical data FLAG of the previous line output to the detection area expansion processing unit 316R. If the logical data EDGE is “1”, the logical data FLAG with the pixel value “1” is output to the detection area expansion processing unit 316R. When the logical data EDGE is “1” and the logical data VAL is “1” and the logical data FLAG is once set to “1”, the determination unit 315R receives the logical data EDGE whose value is “0” next. The logical data FLAG whose value is “1” is output to the detection area expansion processing unit 316R until it is input.

  Here, a pixel that is the first feature pixel in the R data and is not the second feature pixel expanded in either the G data or the B data is referred to as a main scanning direction edge pixel. In other words, the determination unit 315R outputs the logical data FLAG in which the value of the pixel included in the gradation region among the edge pixels in the main scanning direction or the pixel of the background edge is “1”. In addition, when the main scanning direction edge pixels are continuous in the sub-scanning direction, the determination unit 315R determines that one of the plurality of continuous main scanning direction edge pixels is a pixel included in the gradation area or a background edge pixel. For example, the logical data FLAG in which the values of the pixels included in the gradation area or all the edge pixels in the main scanning direction following the background edge pixel are “1” is output.

  The detection area expansion processing unit 316R receives the logical data FLAG from the determination unit 315R. The detection area expansion processing unit 316R specifies a correction target area including pixels around the pixel determined to be noise by the determination unit 315R. That is, the value of the pixel having the value “0” around the pixel having the value “1” of the logical data FLAG input from the determination unit 315R is changed to “1”. As a result, the image quality can be improved by expanding the range of pixels to be corrected. The detection area expansion processing unit 316R outputs logical data indicating the correction target area to the noise correction unit 260. The detection area expansion processing units 316G and 316B have the same functions as the detection area expansion processing unit 316R except that they handle different data. Therefore, the description thereof will not be repeated here.

<First Modification>
In the noise detection processing unit 259 described above, the leading edge of the edge in the main scanning direction in which the first feature pixel in the R data and the pixel that is not the second feature pixel expanded in either the G data or the B data is continuous in the sub-scanning direction. In addition, if there is a gradation area, it is determined as noise. If there is no change in the background of the document, the edge in the main scanning direction should include a gradation area at the trailing edge. For this reason, the accuracy of detecting the main scanning direction edge as noise can be improved by detecting the trailing gradation region in addition to the gradation region at the leading edge of the main scanning direction edge.

  FIG. 16 is a functional block diagram showing a detailed configuration of the gradation detection unit 311R in the modification. Referring to FIG. 16, the gradation detection unit 311R in the modification includes a front end gradation detection unit 321R, a rear end gradation detection unit 322R, and a noise detection permission region determination unit 323R. The leading gradation detecting unit 321R has the same configuration as the gradation detecting unit 311R shown in FIG. Therefore, description is not repeated here.

  FIG. 17 is a functional block diagram illustrating an outline of functions of the trailing edge gradation detection unit 322R. Referring to FIG. 17, rear end gradation detection unit 322R includes a primary differential filter calculation unit 341, differential comparison filter units 342 and 343, and an OR element 344.

  R data is input to the primary differential filter calculation unit 341. The primary differential filter calculation unit 341 differentiates R data input using the primary differential filter shown in FIG. 17 and outputs differential values to the differential comparison filter units 342 and 343, respectively. The primary differential filter is a pixel whose processing target is a position indicated by hatching in the center of the drawing.

  The differential comparison filter units 342 and 343 detect the presence / absence of the change in the pixel value and the direction of the change using the differential comparison filter shown in FIG. Referring to FIG. 18, pixels indicated by hatching are pixels to be processed, and reference numerals are given to the respective pixels for description. The pixel D05 is a processing target pixel, and the pixel on the previous line is D06. The differential comparison filter units 342 and 343 detect the presence / absence of the change in the pixel value and the direction of the change using the primary differential values of the pixels in the previous 10 lines including the pixel to be processed. Here, the change in the pixel value of the pixels in the previous 10 lines including the pixel to be processed is detected, but the number of pixels for detecting the change in the pixel value is not limited to 10 pixels. The number of pixels for detecting a change in value may be determined by the moving speed of the 205th document.

The differential comparison filter unit 342 detects the direction in which the pixel value decreases in order to detect black noise. Specifically, the differential comparison filter unit 342 determines the pixel value if the number of pixels in which the first-order differential values of the pixels in the previous 10 lines including the pixel to be processed exceed a predetermined threshold value SignRef (K). Judge that there is a decreasing change. However, the differential comparison filter unit 342 uses the dynamic range to satisfy the following constraint in order to prevent erroneous detection of characters as noise.
(1) The dynamic range of the pixel values of the pixels on the five lines after the processing target pixel D05 is a predetermined threshold value DRRef1 (K).
MAX (D00, D01, D02, D03, D04) −MIN (D00, D01, D02, D03, D04) <DRRef1 (K)
(2) The dynamic range of the pixel values of the pixels on the previous five lines including the processing target pixel D05 is a predetermined threshold value DRRef2 (K).
MAX (D05, D06, D07, D08, D09) −MIN (D05, D06, D07, D08, D09) <DRRef2 (K)
(3) The dynamic range of the pixel values of the pixels on the 5th to 9th lines before the processing target pixel D05 is a predetermined threshold value DRRef3 (K).
MAX (D10, D11, D12, D13, D14) −MIN (D10, D11, D12, D13, D14) <DRRef3 (K)
(4) The dynamic range of the pixel values of the pixels on the 10th to 14th lines before the processing target pixel D05 is a predetermined threshold value DRRef4 (K).
MAX (D15, D16, D17, D18, D19) −MIN (D15, D16, D17, D18, D19) <DRRef4 (K)
The differential comparison filter unit 342 has the number of pixels whose first-order differential values of pixels in the previous 10 lines including the pixel to be processed have a plus sign exceed a predetermined threshold value SignRef (K), and the above four constraint conditions On the condition that all the conditions are satisfied, logical data with the value of the processing target pixel being “1” is output to the logical sum element 344. When the logical data does not satisfy any one of the above conditions, the value of the processing target pixel is “0”.

The differential comparison filter unit 343 detects the direction in which the pixel value increases in order to detect white noise. Specifically, the differential comparison filter unit 343 sets the pixel value if the number of pixels in which the first-order differential value of the pixels in the previous 10 lines including the pixel to be processed exceeds a predetermined threshold value SignRef (W). Judge that there is a rising change. However, the differential comparison filter unit 343 uses the dynamic range to satisfy the following constraints in order to prevent erroneous detection of characters as noise.
(1) The dynamic range of the pixel values of the pixels on the five lines after the processing target pixel D05 is a predetermined threshold value DRRef1 (W).
MAX (D00, D01, D02, D03, D04) −MIN (D00, D01, D02, D03, D04) <DRRef1 (W)
(2) The dynamic range of the pixel values of the previous five lines of pixels including the processing target pixel D05 is a predetermined threshold value DRRef2 (W).
MAX (D05, D06, D07, D08, D09) −MIN (D05, D06, D07, D08, D09) <DRRef2 (W)
(3) The dynamic range of the pixel values of the pixels on the 5th to 9th lines before the processing target pixel D05 is a predetermined threshold value DRRef3 (W).
MAX (D10, D11, D12, D13, D14) −MIN (D10, D11, D12, D13, D14) <DRRef3 (W)
(4) The dynamic range of the pixel values of the pixels on the 10th to 14th lines before the processing target pixel D05 is a predetermined threshold value DRRef4 (W).
MAX (D15, D16, D17, D18, D19) −MIN (D15, D16, D17, D18, D19) <DRRef4 (W)
The differential comparison filter unit 343 has the number of pixels in which the first-order differential values of the pixels in the previous 10 lines including the pixel to be processed have a minus sign exceed a predetermined threshold value SignRef (W), and the four constraint conditions On the condition that all the conditions are satisfied, logical data with the value of the processing target pixel being “1” is output to the logical sum element 344. When the logical data does not satisfy any one of the above conditions, the value of the processing target pixel is “0”.

  The logical sum element 344 receives logical data from the differential comparison filter units 342 and 343, calculates a logical sum of the logical data, and outputs logical data obtained by the calculation. The logical data output from the logical sum element 344 is called GRAE.

  Returning to FIG. 16, the noise detection permission area determination unit 323R receives the logical data GRA from the leading gradation detection unit 321R and the logical data GRAE from the trailing gradation detection unit 322R. After the logical data GRA having the value “1” is input from the leading gradation detecting unit 321R, the noise detection permission area determining unit 323R receives the logical data GRAE having the pixel value “1” at the same position in the sub-scanning direction. On the condition that it is input from the trailing edge gradation detection unit 322R, the value of the pixel at the same position in the sub-scanning direction is “1” from the line where the logical data GRA having the value “1” is input from the leading gradation detection unit 321R. The logical data with the value “1” is output to the determination unit 315R, with the line until the logical data GRAE of “is input from the trailing edge gradation detection unit 322R as an area where noise detection is permitted.

  The logical data GRA output from the leading gradation detection unit 321R has a value of “1” in the case of a white gradation region, a value of “2” in the case of a black gradation region, and a value of “0” in other cases. And the logical data GRAE output from the trailing edge gradation detection unit 322R is “1” in the case of the white gradation area, “2” in the case of the black gradation area, and the value in the other cases. It may be “0”. In this case, the noise detection permission area determination unit 323R receives the logic data GRA having the value “1” in the sub-scanning direction after the logical data GRA having the value “1” is input from the leading gradation detection unit 321R. On the condition that the data GRAE is input from the trailing edge gradation detection unit 322R, the value of the pixel at the same position in the sub-scanning direction from the line where the logical data GRA having the value “1” is input from the leading edge gradation detection unit 321R. The logical data GRAE having the value “1” is output to the determination unit 315R, with the line until the logical data GRAE having the value “1” input from the rear end gradation detection unit 322R is set as an area where noise detection is permitted. Further, the noise detection permission area determining unit 323R receives the logical data with the pixel value “2” at the same position in the sub-scanning direction after the logical data GRA with the value “2” is input from the leading gradation detection unit 321R. On the condition that GRAE is input from the trailing edge gradation detection unit 322R, the value of the pixel at the same position in the sub-scanning direction from the line where the logical data GRA having the value “2” is input from the leading gradation detection unit 321R. The logical data with a value of “1” is output to the determination unit 315R, with the line until the logical data GRAE of “2” is input from the trailing edge gradation detection unit 322R being an area where noise detection is permitted.

  Here, for the logical data input from the noise detection permission region determination unit 323R of the gradation detection unit 311R, the determination unit 315R inputs the logical data EDGE input from the AND element 306 and the noise detection permission region determination unit 323R. The logical product with the logical data to be calculated is calculated, and the logical data obtained by the calculation is output to the detection area expansion processing unit 316.

<Second Modification>
In the noise detection processing unit 259 described above, the leading edge of the edge in the main scanning direction in which the first feature pixel in the R data and the pixel that is not the second feature pixel expanded in either the G data or the B data is continuous in the sub-scanning direction. If there is a background edge, it is determined as noise. If there is no change in the background of the document, the edge in the main scanning direction should include a gradation area at the trailing edge. For this reason, in addition to the detection of the background edge at the front end of the main scanning direction edge, the accuracy of detecting the main scanning direction edge as noise can be improved by detecting the gradation area at the rear end.

  FIG. 19 is a functional block diagram illustrating a detailed configuration of the background edge detection unit 312R according to the modification. Referring to FIG. 19, the background edge detection unit 312R in the modification includes a front end background edge detection unit 324R, a rear end gradation detection unit 322R, and a noise detection permission region determination unit 325R. The tip background edge detection unit 324R has the same configuration as the background edge detection unit 312R illustrated in FIG. The rear end gradation detection unit 322R has the same configuration as the rear end gradation detection unit 322R shown in FIG. Therefore, description thereof will not be repeated here.

  The noise detection permission area determination unit 325R receives the logical data BE from the leading edge background edge detection unit 324R and the logical data GRAE from the trailing edge gradation detection unit 322R. The noise detection permission region determination unit 325R receives the logical data GRAE having the pixel value “1” at the same position in the sub-scanning direction after the logical data BE having the value “1” is input from the leading background edge detection unit 324R. Is input from the trailing edge gradation detection unit 322R, the position of the pixel at the same position in the sub-scanning direction from the line where the logical data BE having the value “1” is input from the leading background edge detection unit 324R. The logical data GRAE having the value “1” is output to the determination unit 315R, with the line until the logical data GRAE having the value “1” input from the trailing edge gradation detection unit 322R is set as an area where noise detection is permitted.

  The logical data BE output from the front-end / background-edge detection unit 324R is set to “1” when a background edge is detected in a determination formula for detecting white dust, and a determination formula for detecting black dust. The value when the background edge is detected is set to “2”, the value is set to “0” in other cases, and the logical data GRAE output from the trailing edge gradation detection unit 322R is set to the value in the case of the white gradation region. The value may be “1”, the value may be “2” in the case of a black gradation region, and the value may be “0” in other cases. In this case, the noise detection permission area determination unit 325R receives the logical data BE having the value “1” from the front-end background edge detection unit 324R, and the value of the pixel at the same position in the sub-scanning direction is “1”. On the condition that the logical data GRAE is input from the trailing edge gradation detection unit 322R, pixels at the same position in the sub-scanning direction from the line where the logical data BE having the value “1” is input from the leading edge background edge detection unit 324R. The logical data GRAE with the value “1” is output from the trailing edge gradation detection unit 322R to the determination unit 315R with the logical data with the value “1” as an area where noise detection is permitted. In addition, the noise detection permission area determination unit 325R receives the logical data BE having the value “2” in the sub-scanning direction after the logical data BE having the value “2” is input from the leading background edge detection unit 324R. On the condition that the data GRAE is input from the trailing edge gradation detection unit 322R, the pixel at the same position in the sub-scanning direction from the line where the logical data BE having the value “2” is input from the leading edge background edge detection unit 324R. The logical data GRAE with the value “2” is output from the trailing edge gradation detection unit 322R to the determination unit 315R with the logical data with the value “1” as an area where noise detection is permitted.

  For the logical data input from the noise detection permission region determination unit 325R of the background edge detection unit 312R, the determination unit 315R receives the logical data EDGE input from the AND element 306 and the noise detection permission region determination unit 325R. The logical product with the logical data is calculated, and the logical data obtained by the calculation is output to the detection area expansion processing unit 316.

  As described above, the dust on the document table is read at different positions on the document by the three line sensors 213R, 213G, and 213B. Therefore, the data obtained by reading the dust on the document is R data and G data. And an edge in the main scanning direction existing only in one of the B data. Further, since the document table 205 moves relative to the line sensors 213R, 213G, and 213B, of the edges in the main scanning direction detected from only one data of R data, G data, and B data, a pixel at the edge of the edge. In some cases, the value changes in the main scanning direction edge. The noise detection processing unit 259 of the image reading apparatus 10 detects an edge in the main scanning direction in which the pixel value changes in the main scanning direction from each of R data, G data, and B data output from the three line sensors 213R, 213G, and 213B. A main scanning direction edge that is detected from only one of R data, G data, and B data, and includes a gradation region in which the pixel values of a predetermined number of pixels that are continuous in the sub scanning direction change. Is noise. For this reason, it is possible to improve the accuracy of detecting noise generated by reading dust existing on the document table from an image obtained by reading the document.

  When the document has a white background and a black solid area, the boundary between the document background and the black solid area is detected as an edge in the sub-scanning direction. Even if black dust adheres to the platen, noise generated by dust is not detected as an edge in the main scanning direction while scanning the black solid area of the document, but dust is detected while scanning the background of the document. The generated noise is detected as an edge in the main scanning direction. Conversely, even if white dust adheres to the platen, noise generated by dust is not detected as an edge in the main scanning direction while scanning the background of the document, but dust is detected while scanning a black solid area. Is detected as an edge in the main scanning direction. In this case, since the detected edge in the main scanning direction has little change in the pixel value at the end, the gradation area is not detected. Therefore, when an edge in the main scanning direction is detected from only one of R data, G data, and B data, the sub scanning in which the pixel value changes in the sub scanning direction in a region around the detected edge in the main scanning direction. The edge in the main scanning direction is set as noise on the condition that the direction edge is detected. Thereby, it is possible to further improve the accuracy of detecting noise generated by reading dust existing on the document table from an image obtained by reading the document.

  In the image reading apparatus 10 according to the present embodiment, the second brightness difference detection units 302R, 302G, and 302B are provided. However, these may be omitted. In this case, logical data with the first feature pixel being “1” is output from the first brightness difference detection units 301R, 301G, and 301B to the detection result extension processing units 303R, 303G, and 303B, and the detection result extension processing unit 303R , 303G, 303B, the first feature pixel is expanded. Then, a pixel that is the first feature pixel and is not the first feature pixel expanded with other data is detected as a noise pixel.

  In this embodiment, the example in which the reading unit 213 is fixed to the main body unit 103 and the document is scanned is shown. However, the present invention can also be applied to the case where the reading unit 213 is moved to scan the document. For example, the upper regulating plate is made white or black, and the document is scanned by moving the reading unit 213 or the light source 206, the reflection mirror 209, and the reflection member 208 in the sub-scanning direction. By vibrating the document table 205 in the sub-scanning direction during this scanning, dust adhering to the document table 205 can be detected.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

1 is a perspective view of an MFP including an image reading apparatus according to one embodiment of the present invention. 2 is a diagram illustrating an outline of an internal configuration of an image reading apparatus. FIG. It is a perspective view which shows the mechanism for vibrating an original table. It is a figure for demonstrating correction | amendment between lines. It is a figure for demonstrating the principle which detects the noise which generate | occur | produces with the dust adhering to a document stand from the image data obtained by reading a document. FIG. 3 is a plan view of the document table viewed from the back side. FIG. 6 is a diagram illustrating a reading area on a document table 205 read by a reading unit. 3 is a block diagram illustrating a configuration of an image processing unit of the image reading apparatus according to the present embodiment. FIG. It is a figure which shows an example of the RGB data after correction | amendment between lines. It is a block diagram which shows the structure of the noise detection process part of the image reading apparatus in this Embodiment. It is a figure which shows an example of an edge extraction filter. It is a functional block diagram which shows the outline of the function of a gradation detection part. It is a figure which shows an example of a differential comparison filter. It is a figure for demonstrating the principle which detects a background edge area | region. It is a figure which shows an example of a background edge detection filter. It is a functional block diagram which shows the detailed structure of the gradation detection part in a modification. It is a functional block diagram which shows the outline of the function of the rear end gradation detection part 322R. It is another figure which shows an example of a differential comparison filter. It is a functional block diagram which shows the detailed structure of the background edge detection part in a modification.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 Image reading apparatus, 20 Image forming apparatus, 103 Main body part, 110 Image reading apparatus, 201 Timing roller pair, 202 Roller pair, 203 Upper restriction plate, 205 Original stand, 206 Light source, 207 Paper passing guide, 208 Reflecting member, 209 Reflection mirror, 211 lens, 213 reading unit, 213R, 213G, 213B line sensor, 215 image processing unit, 217 motor control unit, 219 motor, 251 A / D conversion unit, 253 shading correction unit, 255 interline correction unit, 257 Chromatic aberration correction unit, 259 Noise detection processing unit, 260 Noise correction unit, 261 Printer interface, 263 Control unit, 265 Position detection unit, 301R, 301G, 301B First brightness difference detection unit, 302R, 302G, 302B Second brightness difference detection Part, 303R , 303G, 303B detection result expansion processing unit, 311R, 311G, 311B gradation detection unit, 312R, 312G, 312B background edge detection unit, 315R, 315G, 315B determination unit, 316R, 316G, 316B detection area expansion processing unit, 321R tip Gradation detection unit, 322R Rear end gradation detection unit, 323R, 325R Noise detection permission area determination unit, 324R Front end background edge detection unit.

Claims (8)

A plurality of line sensors having filters having different spectral sensitivities, arranged in a predetermined order at a distance in the sub-scanning direction, and scanning a document in the sub-scanning direction;
A document table provided between the document and the plurality of line sensors;
Moving means for moving the document table relative to the plurality of line sensors at a relative speed different from a relative speed between the document and the plurality of line sensors;
Interline correction means for synchronizing a plurality of data output by the plurality of line sensors so as to become pixels obtained by reading the same position of the document;
A plurality of data synchronized from the inter-line correction means, noise pixel detection means sequentially input in line units,
The noise pixel detecting means detects main edge in the main scanning direction in which the pixel value changes in the main scanning direction from each of the plurality of data, and
A gradation detection unit that detects a gradation region in which the pixel values of a predetermined number of pixels that are continuous in the sub-scanning direction change from each of the plurality of data;
The main scanning direction edge detected from only one of the plurality of data by the main scanning direction edge detecting means, and the gradation detecting means detects the main scanning direction edge from the data detected by the gradation detecting means. An image reading apparatus comprising: a first noise detecting unit that uses the edge in the main scanning direction including one gradation area as noise.   2. The first noise detection unit according to claim 1, wherein the main scanning direction edge after the line to which the first gradation area belongs detected from the data in which the main scanning direction edge is detected by the gradation detection unit is used as noise. The image reading apparatus described.   The first noise detecting means is provided on the condition that the second gradation area is detected after the line to which the first gradation area is detected from the data in which the edge in the main scanning direction is detected by the gradation detecting means. The image reading apparatus according to claim 1, wherein the main scanning direction edge after the line to which the first gradation area belongs and before the line to which the second gradation area belongs is defined as noise.   The image reading apparatus according to claim 3, wherein the first gradation area and the second gradation area have opposite directions in which pixel values change. The noise pixel detecting unit detects a sub-scanning direction edge detecting unit that detects a sub-scanning direction edge in which a pixel value changes in a sub-scanning direction in a region around the main scanning direction edge detected by the main scanning direction edge detecting unit. When,
When the main scanning direction edge is detected from only one data of the plurality of data by the main scanning direction edge detecting means, on condition that the sub scanning direction edge is detected by the sub scanning direction edge detecting means. The image reading apparatus according to claim 1, further comprising a second noise detecting unit that uses the edge in the main scanning direction as noise.   The image reading apparatus according to claim 5, wherein the second noise detecting unit uses the main scanning direction edge subsequent to the line to which the sub-scanning direction edge detected by the sub-scanning direction edge detecting unit belongs as noise.   The second noise detecting means may detect the third gradation after the line to which the edge in the sub-scanning direction belongs, provided that the third gradation area is detected after the line to which the edge in the sub-scanning direction belongs by the gradation detecting means. The image reading apparatus according to claim 5, wherein the edge in the main scanning direction before the line to which the region belongs is defined as noise.   The second noise detecting means may include sub-data in the main scanning direction edge line that is different from the one data in which the main scanning direction edge is detected by the main scanning direction edge detecting means among the plurality of data. The image reading apparatus according to claim 5, wherein the edge in the main scanning direction is set as noise on the condition that a sub-scanning direction edge whose pixel value changes in the scanning direction is detected.

Images