JP2011145970A - Image processing apparatus, image processing program, and image forming apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing program, which accurately extract a document end from an image including a document without using a specific sensor. <P>SOLUTION: A mirror image conversion part 11 converts a given rear image into a mirror image, and a density boundary detection part 12 detects feature amounts such as a boundary direction and boundary strength of a density boundary respectively from a surface image and the rear image converted into the mirror image. From the feature amount of the density boundary in the surface image and the feature amount of the density boundary in the corresponding rear image converted into the mirror image, a document end detection part 13 calculates the degree of similarity of both of the feature amounts, extracts portions determined that the feature amounts are similar to each other from the degree of similarity, and detects a straight density boundary as a document end out of the extracted portions. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, an image processing program, and an image forming apparatus.

  Conventionally, document edges are detected by various methods. For example, in Patent Document 1, the edge of a document is detected by illuminating illumination light from a surface opposite to the reading surface of the document so that the document looks black in a shadow shape. Further, in Patent Document 2, an image reading unit is installed in a paper conveyance path, and the edge of the printing paper is detected from an image signal read by the image reading unit with reference to brightness and saturation.

  All of these document edge detection methods detect the document edge in conjunction with the reading device. As described above, it is considered to detect an end portion of an original from an image including the original such as an image obtained by reading the original without adding a special configuration to the reading apparatus. For example, in Patent Document 3, the spread of an effective image is determined although it is not at the document edge.

  Another conventional technique related to the present invention is a technique using a mirror image. For example, Patent Document 4 describes that the show-through data is discriminated by comparing the density of the front-side image data obtained by reading the front side of the document with the back-side image data obtained by reading the back side and mirror-inverted, and removing the show-through. Has been. Further, for example, Patent Document 5 stores a reversed right / left image of the read document surface, and compares the reverse image of the original document with the reverse image of the front surface. The density of the reverse image is lower than that of the reverse image. In the case, it is described that the show-through is removed by erasing and outputting the surface image corresponding to the location.

JP-A-6-337923 JP 2005-269124 A JP 2004-236026 A Japanese Patent No. 3359180 Japanese Patent Laid-Open No. 11-317858

  An object of the present invention is to provide an image processing apparatus and an image processing program for accurately extracting an edge of an original from an image including an original without using a special sensor, and an image forming apparatus using such an image processing apparatus. It is what.

  According to the first aspect of the present invention, there are provided density boundary detection means for detecting feature values of density boundaries whose density changes from the front image including the front surface of the document and the back image including the back surface, and the density obtained from the surface image. An image processing apparatus comprising document edge detecting means for detecting a document edge according to a feature value of a boundary and a feature value of a density boundary obtained from the back image corresponding to the density boundary.

  The invention according to claim 2 of the present application further includes mirror image conversion means for performing mirror image conversion of either the front image or the back image in the configuration of the invention according to claim 1, and the density boundary detection The means detects at least the direction of the density boundary as the feature amount using the image subjected to mirror image conversion by the mirror image conversion means as the front image or the back image, and the document edge detection means detects the density obtained from the front image. The image processing apparatus is characterized in that the document edge is detected from a similarity between a boundary feature amount and a density boundary feature amount obtained from the back image corresponding to the density boundary.

  In the invention according to claim 3 of the present application, the density boundary detection means in the invention according to claim 1 detects at least the direction of the density boundary as a feature amount, and the document edge detection means obtains from the surface image. And calculating the similarity between the density boundary feature amount and the density boundary feature amount obtained from the back image in a mirror image-related position with respect to the density boundary position in the surface image from which the feature amount was obtained, and the direction The image processing apparatus is characterized in that the similarity is calculated by mirror-transforming the direction of one of the feature values, and the document edge is detected according to the similarity.

  In the invention described in claim 4 of the present application, the document edge detecting means in the invention described in claim 2 or claim 3 of the present invention is similar in a region having a predetermined size between the front image and the back image. An image processing apparatus characterized by calculating a degree.

  In the invention according to claim 5 of the present application, the document edge detecting means according to any one of claims 1 to 4 extracts a straight line portion based on the feature amount, and An image processing apparatus is characterized in that a straight line located at the end is detected as a document edge.

  The invention according to claim 6 of the present application is a reading means for reading an image including a front surface and a back surface of a document and outputting it as a front image and a back image, and receiving the front image and the back image to detect a document edge. The image processing apparatus according to any one of claims 1 to 5, forming means for forming an image on a medium, and the document edge detected by the image processing apparatus with respect to the front surface image and the back surface image. An image forming apparatus comprising image forming means for creating an image formed by the image forming means.

  The invention according to claim 7 of the present application is an image processing program for causing a computer to execute the function of the image processing apparatus according to any one of claims 1 to 5.

  According to the first aspect of the present invention, it is possible to accurately detect the document edge from the image including the document as compared with the case where the present configuration is not provided.

  According to the second aspect of the present invention, the document edge can be detected more stably against noise or the like than in the case where the present configuration is not provided.

  According to the third aspect of the present invention, it is possible to detect the document edge stably against noise or the like as compared with the case where the present configuration is not provided.

  According to the fourth aspect of the present invention, the document edge can be detected even if the document positions in the front image and the back image are shifted.

  According to the fifth aspect of the present invention, it is possible to detect the edge of the document regardless of the contents drawn on the document.

  According to the invention described in claim 6 of the present application, the image portion of the original can be processed accurately and formed on the medium as compared with the case without this configuration.

  According to the invention of claim 7 of the present application, the effect of the invention of any one of claims 1 to 5 can be obtained.

1 is a configuration diagram illustrating a first embodiment of an image processing apparatus of the present invention. 3 is a flowchart showing an example of an operation in the first embodiment of the image processing apparatus of the present invention. It is explanatory drawing of the specific example of the image processed by an example of operation | movement in 1st Embodiment of the image processing apparatus of this invention. It is a block diagram which shows 2nd Embodiment of the image processing apparatus of this invention. It is a flowchart which shows an example of operation | movement in 2nd Embodiment of the image processing apparatus of this invention. It is explanatory drawing of the specific example of the image processed by an example of operation | movement in 2nd Embodiment of the image processing apparatus of this invention. 1 is a configuration diagram showing an embodiment of an image forming apparatus of the present invention. FIG. 11 is an explanatory diagram of an example of a computer program, a storage medium storing the computer program, and a computer when the functions described in the embodiments of the image processing apparatus of the present invention are realized by a computer program.

  FIG. 1 is a configuration diagram showing a first embodiment of an image processing apparatus according to the present invention. In the figure, 11 is a mirror image conversion unit, 12 is a density boundary detection unit, and 13 is a document edge detection unit. In the embodiment described below, it is assumed that a front image including the front surface of the document and a back image including the back surface are provided. For example, the front image and the back image may be acquired by an image reading device that reads the front and back of a document at once.

  The mirror image conversion unit 11 performs mirror image conversion on one of the given front surface image and back surface image.

  The density boundary detection unit 12 detects a feature amount of a density boundary whose density changes from the front image and the back image. One of the front image and the back image uses an image that has undergone mirror image conversion by the mirror image conversion unit 11. For example, when the mirror image is converted by the mirror image conversion unit 11, the feature value of the density boundary in the front image is detected from the image subjected to the mirror image conversion, and the feature value of the density boundary in the back image is detected from the given back image. To do. Alternatively, for example, when the rear surface image is mirror-image converted by the mirror image conversion unit 11, the feature value of the density boundary in the rear image is detected from the mirror-image-converted image, and the feature value of the density boundary in the front image from the given front image Is detected. As the feature value of the density boundary to be detected, for example, the direction and intensity of the density boundary may be extracted. Of course, it goes without saying that the length and other feature quantities may be extracted.

  The document edge detection unit 13 detects the document edge according to the feature value of the density boundary in the front image and the feature value of the density boundary in the corresponding back image. For example, what is necessary is just to calculate the similarity of both feature-values. When calculating the degree of similarity, it is preferable that the degree of similarity be calculated within an area having a predetermined size in consideration of positional deviation between the front surface image and the back surface image. For example, with respect to the density boundary existing in both the front image and the back image, it is determined from the similarity that the feature amount is similar, and it is only necessary to find such a density boundary. Even if such a density boundary exists in the document, the density boundary at the edge of the document should exist at the outermost side. May be detected as the document edge.

  FIG. 2 is a flowchart showing an example of the operation of the image processing apparatus according to the first embodiment of the present invention. In S <b> 61, the mirror image conversion unit 11 performs mirror image conversion on one of the given front surface image and back surface image. Here, it is assumed that the back image is subjected to mirror image conversion.

  In S <b> 62, the density boundary detection unit 12 detects a feature value of the density boundary whose density changes from the front surface image and the mirror image-converted back surface image. Here, at least the boundary direction is detected, and the boundary strength is also detected.

  In step S63, the document edge detection unit 13 first calculates the similarity of the feature value of the density boundary. The angle formed by the boundary direction of the density boundary at the corresponding position of the front image and the mirror image converted back image is θ, and the value may be obtained by, for example, the similarity function F (θ). Here, as the function F (θ), the smaller the θ, the larger the value. For example, the inverse of θ, cos θ, binary output function (1: 0 ≦ θ <Th, 0: Th <θ), etc. are considered. It is done. Of course, other functions may be used. When the boundary strength is obtained, the similarity may be obtained by a vector operation such as an inner product as a boundary vector together with the boundary direction.

  The document edge detection unit 13 extracts document edges based on the similarity obtained in S63. As one of the processes, here, a similarity image is created in S64, and a straight line is created from the similarity image in S65. The example which performs the process to detect is shown. For the similarity image created in S64, the similarity value calculated in S63 may be arranged as a position value corresponding to the front image for which the similarity is calculated and the back image subjected to the mirror image conversion. Alternatively, values obtained by multiplying the sum of pixel values (colors) or sum of boundary intensities at corresponding positions in the front image and the mirror image converted back image may be arranged. Of course, in addition to this, values obtained by various methods such as obtaining a linear sum or a nonlinear sum of feature amounts according to the degree of similarity may be arranged.

  A straight line is detected in S65 based on the similarity image thus obtained. The similarity value is increased for the density boundary that exists in common at the corresponding positions of the front image and the mirror image converted back image. Since the document edge is a straight line that exists in common at the corresponding position of the front image and the mirror image converted back image, if the straight line is detected from the similarity image, the detected document line includes the document edge. It will be. As a straight line detection method, for example, a known method such as Hough transform or edge tracking may be used.

  Finally, in S66, the document edge is extracted from the straight line detected in S65. Even if various straight lines are drawn in the original, the end of the original is the straight line located at the outermost periphery, and therefore, the straight line located at the end may be extracted as the original end. Alternatively, since the straight line at the document edge is longer than the straight line such as characters, the longest straight line may be extracted as the document edge. In this way, the document edge is extracted from the corresponding front and back images.

  An example of the above operation will be described using a specific example. FIG. 3 is an explanatory diagram of a specific example of an image processed by an example of an operation in the first embodiment of the image processing apparatus of the present invention. FIG. 3A shows the given front and back images. For example, when considering a case where a front image and a back image are acquired by an image reading device that reads the front and back of a document at once, an image may be read with the document tilted during reading. However, in the read front image and back image, the shape of the document has a mirror image relationship. Also in the front and back images shown in FIG. 3A, the outline of the document has a mirror image relationship.

  First, the mirror image conversion unit 11 performs mirror image conversion on either one of the given front surface image or back surface image. Here, the back image is mirror-image converted. FIG. 3B shows a mirror image-converted image as the back image.

  Next, the density boundary detection unit 12 detects a feature value of the density boundary where the density changes from the front surface image and the back image subjected to mirror image conversion. In FIG. 3B, the boundary direction detected as the feature amount for four points in the image is indicated by arrow lines. A point a in the front image is an end portion of the document, a point b is a line segment region drawn on the document, a point c is a noise portion on the document, and a point d is a noise portion outside the document. The positions in the rear image subjected to mirror image conversion corresponding to the respective positions of the front image are indicated as a ', b', c ', d'. Note that arrow lines are not shown in c ′ and d ′ in the mirror image-converted back image because there is no density boundary.

  When the feature value of the density boundary is detected, the document edge detection unit 13 first calculates the similarity of the feature value of the density boundary, and creates a similarity image here. An example of the similarity image is shown in FIG. In FIG. 3C, when the corresponding part of the front surface image and the back image subjected to the mirror image conversion is similar, it is shown in black.

  For example, when comparing the feature amounts at corresponding points of the front image and the mirror image-converted back image shown in FIG. 3B, the boundary direction of the density boundary at the point a of the front image and the point a of the mirror image-converted back image By calculating the similarity from the boundary direction at “′”, it is similar in this portion, so it is shown in black in FIG. However, the boundary direction at the point b is different from the boundary direction at the point b ', and is therefore shown in white in FIG. 3C from the calculated similarity. Further, the boundary direction is detected at the point c and the point d in the front image, but the density boundary does not exist at the corresponding point c ′ and the point d ′ in the mirror image converted back surface, and the boundary direction is also Not detected. Therefore, these portions are also shown in white in FIG. In this way, when a similar part for each point is black and a non-similar part is white, as shown in FIG. 3C, a document edge, and it happens to be a front image and a mirror image converted back image. A similarity image in which the portion where the common line segment exists is black is obtained. For example, even when the density difference at the edge of the document is lighter than a line such as a character, it appears clearly in the similarity image. Even if noise such as point c or point d exists in the image, it is deleted in the similarity image. Note that the similarity image shown in FIG. 3C is for explanation, and may actually have a binary or multivalued value instead of white or black.

  When a straight line portion is detected from the similarity image shown in FIG. 3C, in this example, the document edge is extracted as shown in FIG. 3D. For example, if the straight line located at the outermost periphery is extracted, the document edge is extracted. Alternatively, if a straight line longer than a predetermined length is detected, a line segment obtained from a character or the like is excluded. For example, even if the line segment serving as the document edge is interrupted, if the line segments obtained from the line segments overlap, the line segments serving as the document edge can be obtained by integrating the line segments.

  FIG. 4 is a block diagram showing a second embodiment of the image processing apparatus of the present invention. In this second embodiment, an example is shown in which the similarity is calculated by comparing the feature quantity of the position of the mirror image when detecting the document edge without performing mirror image conversion of the image. Note that differences from the first embodiment will be mainly described.

  Since the second embodiment does not have the mirror image conversion unit 11 and does not perform mirror image conversion on the image, the density boundary detection unit 12 determines the characteristics of the density boundary from the given front surface image and back surface image, respectively. Detect the amount.

  The document edge detection unit 13 detects the document edge according to the feature value of the density boundary in the front image and the feature value of the density boundary in the corresponding back image. The corresponding positions of the front image and the back image are positions that have a mirror image relationship with each other, and the similarity of the feature amounts at the positions that have the mirror image relationship may be calculated. When calculating the degree of similarity, it is preferable that the degree of similarity be calculated within an area having a predetermined size in consideration of positional deviation between the front surface image and the back surface image. When the boundary direction of the density boundary is used as the feature amount, the front surface image and the back surface image have a mirror image relationship with each other, and therefore the boundary direction detected by the density boundary detection unit 12 has a mirror image relationship. Therefore, when calculating the similarity, either the boundary direction obtained from the front surface image or the boundary direction obtained from the back surface image is converted into a direction having a mirror image relationship.

  FIG. 5 is a flowchart showing an example of the operation of the image processing apparatus according to the second embodiment of the present invention. In S <b> 71, the density boundary detection unit 12 detects a feature value of a density boundary whose density changes from the given front surface image and back surface image. Here, at least the boundary direction is detected, and the boundary strength is also detected.

  In step S72, the document edge detection unit 13 first performs a mirror image conversion on one of the front surface image and the back surface image in the boundary direction among the feature values of the density boundary detected in step S71. Here, the mirror image conversion of the entire image performed by the mirror image conversion unit 11 in the first embodiment is not performed.

  In step S73, the document edge detection unit 13 calculates the similarity between the feature values of the density boundary. At this time, the similarity is calculated from the feature value of the density boundary at the position of the back image in the relationship between the position of the front image and the mirror image. Since the calculation method of the similarity is described in the first embodiment, it is omitted here. However, the similarity in the boundary direction is maximum when the angle difference between the directions is 0 degree or 180 degrees, and is 90 degrees or 270 degrees. It is advisable to calculate the similarity so as to minimize it.

  The processes such as the similarity image creation, straight line detection, and document edge extraction performed by the document edge detection unit 13 in the processes after S74 are also the processes after S64 in FIG. 2 described in the first embodiment. The description is omitted here. In this way, instead of performing mirror image conversion on the image, the document edge can be obtained even if the configuration is such that the reference position of the front image and back image in the mirror image conversion in the boundary direction of the density boundary and the similarity calculation is changed. Will be.

  An example of the above operation will be described using a specific example. FIG. 6 is an explanatory diagram of a specific example of an image processed by an example of the operation in the second embodiment of the image processing apparatus of the present invention. FIG. 6A shows the given front and back images, which are shown in FIG.

  First, the density boundary detection unit 12 detects a feature value of a density boundary whose density changes from the front image and the back image. In FIG. 6A, the boundary direction detected as the feature amount is indicated by arrow lines for four points in the image. In the front image, points a and e are document edge portions, point b is a line segment region drawn on the document, point c is a noise portion on the document, and point d is a noise portion outside the document. . The positions in the back image subjected to mirror image conversion corresponding to the respective positions of the front image are indicated as a ', b', c ', d', e '. Note that arrow lines are not shown in c ′ and d ′ in the mirror image-converted back image because there is no density boundary.

  When the feature value of the density boundary is detected, the document edge detection unit 13 first performs mirror image conversion of the boundary direction among the feature values of the density boundary detected by the density boundary detection unit 12 for either the front image or the back image. deep. Here, the boundary direction of the density boundary detected from the back surface image of FIG. 6A is assumed to be mirror-image converted in the direction indicated as “conversion direction” in the figure, and the result is shown in FIG. 6B. . The back image itself is not mirror-image converted, but the direction of the arrow indicating the boundary direction is mirror-image converted.

  Comparing the feature quantities at the positions of the mirror images of the front image and the back image shown in FIG. 6B, for example, the boundary direction of the density boundary at the point a of the front image and the boundary direction after the mirror image conversion at the point a ′ of the back image By calculating the similarity from the above, it is similar in this part, so it is shown in black in FIG. However, the boundary direction at the point b is different from the boundary direction obtained by mirror image conversion at the point b ', and is therefore shown in white in FIG. 6C from the calculated similarity. Further, the boundary direction is detected at the point c and the point d in the front image, but the density boundary does not exist at the point c ′ and the point d ′ at the mirror image positions in the back image, and the boundary direction is also Not detected. Accordingly, these portions are also shown in white in FIG. Further, although the boundary direction of the density boundary at the point e of the front image and the boundary direction after the mirror image conversion at the point e ′ of the back image are directed opposite to each other, they are maximized at 0 degree and 180 degrees as described above. If the similarity is calculated, it is detected that the image is similar in this portion, and is shown in black in FIG.

  In this way, when a similar part for each point is black and a non-similar part is white, as shown in FIG. 6C, an original image, and it happens to be a front image and a mirror image-converted back image. A similarity image in which the portion where the common line segment exists is black is obtained. For example, even when the density difference at the edge of the document is lighter than a line such as a character, it appears clearly in the similarity image. Even if noise such as point c or point d exists in the image, it is deleted in the similarity image. Note that the similarity image shown in FIG. 6C is for explanation, and may actually have a binary or multivalued value instead of white or black.

  When a straight line portion is detected from the similarity image shown in FIG. 6C, in this example, the document edge is extracted as shown in FIG. 6D. For example, if the straight line located at the outermost periphery is extracted, the document edge is extracted. Alternatively, if a straight line longer than a predetermined length is detected, a line segment obtained from characters or the like is excluded and the document edge is extracted. For example, even if the line segment serving as the document edge is interrupted, if the line segments obtained from the line segments overlap, the line segments serving as the document edge can be obtained by integrating the line segments.

  FIG. 7 is a block diagram showing an embodiment of the image forming apparatus of the present invention. In the figure, 21 and 22 are reading units, 23 is an image processing unit, 24 is an image creating unit, and 25 is a forming unit. The reading units 21 and 22 read an image including the front and back surfaces of the document and output the images as a front image and a back image. A known configuration may be applied as the configuration for reading both sides of the document.

  The image processing unit 23 has been described above as the embodiment of the image processing apparatus of the present invention. The image processing unit 23 receives the front image and the back image read by the reading unit 21, detects the document edge, and detects the detection result in the image creation unit 24. give.

  The image creating unit 24 creates an image to be formed by the forming unit 25 based on the detection result of the document edge passed from the image processing unit 23 with respect to the front and back images read by the reading unit 21. For example, since the position, size, and inclination of the document can be known from the detection result of the document edge, it is only necessary to correct the inclination and cut out the image of the corrected document portion. Alternatively, an enlargement / reduction process according to the size of the medium used in the forming unit 25 may be performed. Alternatively, a medium having a size corresponding to the size of the clipped image may be specified to instruct the forming unit 25 to form an image. Further, for example, a so-called frame erasing function may be performed by performing correction processing on the document edge portion based on the detection result of the document edge to erase the document edge from the image.

  The forming unit 25 forms the image created by the image creating unit 24 on the medium.

  FIG. 8 is an explanatory diagram of an example of a computer program, a storage medium storing the computer program, and a computer when the functions described in the embodiments of the image processing apparatus of the present invention are realized by a computer program. In the figure, 31 is a program, 32 is a computer, 41 is a magneto-optical disk, 42 is an optical disk, 43 is a magnetic disk, 44 is a memory, 51 is a CPU, 52 is an internal memory, 53 is a reading unit, 54 is a hard disk, and 55 is An interface 56 is a communication unit.

  You may implement | achieve the function of each part demonstrated by each embodiment of the image processing apparatus of the above-mentioned this invention entirely or partially with the program 31 executable by a computer. In that case, the program 31 and data used by the program may be stored in a computer-readable storage medium. A storage medium is a signal of a corresponding signal that causes a change state of energy such as magnetism, light, electricity, etc., depending on the description content of the program, to the reading unit 53 provided in the hardware resource of the computer. The program description is transmitted to the reading unit 53 in a format. For example, there are a magneto-optical disk 41, an optical disk 42 (including a CD and a DVD), a magnetic disk 43, a memory 44 (including an IC card, a memory card, a flash memory, and the like). Of course, these storage media are not limited to portable types.

  The program 31 is stored in these storage media, and the program 31 is read from the computer by, for example, mounting these storage media on the reading unit 53 or the interface 55 of the computer 32, and the internal memory 52 or the hard disk 54 (magnetic disk) And the program 31 is executed by the CPU 51, so that the functions described in the embodiments of the image processing apparatus of the present invention described above are realized in whole or in part. Alternatively, the program 31 is transferred to the computer 32 via the communication path, and the computer 32 receives the program 31 by the communication unit 56 and stores it in the internal memory 52 or the hard disk 54, and the CPU 31 executes the program 31. May be.

  In addition, the computer 32 may be connected to various devices via the interface 55. For example, display means for displaying information, reception means for receiving information from the user, and the like may be connected. Further, for example, an image reading device that reads both sides of the document may be connected via the interface 55, and the document edge may be detected from the front image and the back image read by the image reading device. Further, the image forming apparatus may be connected via the interface 55, and the configuration shown in the above-described embodiment of the image forming apparatus of the present invention including the function of the image creating unit 24 may be realized.

  Of course, it may be partially configured by hardware, or all may be configured by hardware. Alternatively, it may be configured as a program including all or part of the functions described in the embodiments of the image processing apparatus of the present invention together with other configurations. As described above, the program may be configured together with a program that realizes the function of the image creating unit 24 shown in the embodiment of the image forming apparatus of the present invention. Of course, when applied to other purposes, it may be integrated with the program for that purpose.

  DESCRIPTION OF SYMBOLS 11 ... Mirror image conversion part, 12 ... Density boundary detection part, 13 ... Document edge detection part, 21, 22 ... Reading part, 23 ... Image processing part, 24 ... Image preparation part, 25 ... Formation part, 31 ... Program, 32 ... Computer 41 ... Magneto-optical disk 42 ... Optical disk 43 ... Magnetic disk 44 ... Memory 51 ... CPU 52 ... Internal memory 53 ... Reading part 54 ... Hard disk 55 ... Interface 56 ... Communication part

Claims (7)

  Density boundary detecting means for detecting density boundary feature amounts each having a density change from the front image including the front side and the back side image including the back side, and corresponding to the density boundary feature amount and the density boundary obtained from the front image An image processing apparatus comprising document edge detecting means for detecting a document edge according to a feature value of a density boundary obtained from the back image.   Furthermore, it has a mirror image conversion means for performing mirror image conversion on either the front image or the back image, and the density boundary detection means takes the image that has undergone mirror image conversion by the mirror image conversion means as the front image or the back image. And at least the direction of the density boundary is detected as a feature amount, and the document edge detection unit detects the feature amount of the density boundary obtained from the front surface image and the feature amount of the density boundary obtained from the back surface image corresponding to the density boundary. The image processing apparatus according to claim 1, wherein the edge of the document is detected from the similarity.   The density boundary detection unit detects at least the direction of the density boundary as a feature amount, and the document edge detection unit detects the feature amount of the density boundary obtained from the surface image and the density in the surface image from which the feature amount was obtained. The similarity between the boundary position and the feature value of the density boundary obtained from the back image at a mirror image-related position is calculated, and for the feature quantity having a direction, the direction of one feature quantity is mirror-image transformed to obtain the similarity The image processing apparatus according to claim 1, wherein a degree is calculated, and the document edge is detected according to the similarity.   4. The image processing apparatus according to claim 2, wherein the document edge detection unit calculates a similarity in an area having a predetermined size between the front surface image and the back surface image. 5.   5. The document edge detection unit extracts a straight line portion based on the feature amount, and detects a straight line located at the end of the image as a document edge. 6. The image processing apparatus according to item.   6. A reading unit that reads an image including a front surface and a back surface of a document and outputs the image as a front image and a back image, and receives the front image and the back image to detect a document edge. An image forming apparatus that forms an image on a medium, and an image that is formed by the forming unit based on the document edge detected by the image processing apparatus with respect to the front image and the back image. An image forming apparatus comprising image creating means for creating the image.   An image processing program for causing a computer to execute the function of the image processing apparatus according to any one of claims 1 to 5.

Images