JP2012170049A - Image processor and document reading system provided with the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow acquisition of a cut-out image by an operation of directly designating a region with a pointer such as a finger on a paper surface of a document.SOLUTION: An image processor acquires: a first image obtained by capturing an image of the paper surface of the document without the pointer such as a finger, with a document-placing surface; and a second image obtained by capturing the pointer pointing out a region to be cut out on the paper surface of the document, with the paper surface of the document. A cut-out region acquisition section 33 acquires the region to be cut out by detecting a point pointed out with the pointer from the second image, and the cut-out region acquisition section 33 acquires the region to be cut out on the basis of a designated region expressed with a track of one point pointed out with the pointer.

Description

  The present invention relates to an image processing apparatus that processes an image obtained by reading a paper surface of a document such as a book, and a document reading system including the image processing apparatus.

  2. Description of the Related Art Book scanners (document cameras) that can pick up images from above and read a page image in a state in which a page of a book is naturally opened are widely used (see Patent Document 1). When such a book scanner is used, the images of the pages can be read one after another while turning the pages, so that the work of digitizing the book can be performed efficiently.

JP 2001-103240 A

  A book scanner can also be used when an image of only a part of the paper surface of a book is required. In this case, the entire page of the book is imaged or an image is captured so as to include a necessary area in the page. After that, it is necessary to take the image data into the PC, and to cut out the image of the necessary area while viewing the screen of the PC, which is troublesome.

  The present invention has been devised to solve such problems of the prior art, and its main purpose is to cut out an image by directly specifying an area with an indicator such as a finger on the surface of a document. It is an object of the present invention to provide an image processing apparatus configured to be able to acquire image data and a document reading system including the image processing apparatus.

  In order to achieve the above object, an image processing apparatus according to the present invention includes a first photographed image obtained by photographing a paper surface of a document together with a placement surface in the absence of a pointing material, and an indicator indicating a region to be cut out on the paper surface of the document. A captured image acquisition unit that acquires a second captured image that is captured together with a paper surface of a document, a cutout region acquisition unit that detects a point indicated by an indicator from the second captured image and obtains a cutout region, and the first A cut-out image generation unit that obtains a cut-out image by flattening an image portion corresponding to at least the cut-out region obtained by the cut-out region acquisition unit in one captured image, and the cut-out region acquisition unit includes the indicator The cut-out area is obtained based on a designated area represented by a locus of one point indicated by.

  According to the present invention, various areas can be designated as cut-out images by the operation of directly designating areas with a pointing object such as a finger or a pen on the surface of a document. In addition, since a flattened cut-out image can be acquired, user convenience can be improved. In particular, if the configuration is such that only the image portion corresponding to the cutout region is planarized, it is possible to reduce the computational burden required for planarization and increase the processing speed.

1 is an overall configuration diagram showing a document reading system according to an embodiment. The block diagram which shows schematic structure of the book scanner 1 and PC2 Flow chart showing image cutout procedure in this document reading system FIG. 3 is a flowchart showing the procedure of contour acquisition (ST103) shown in FIG. Schematic diagram explaining the outline of outline acquisition FIG. 3 is a flowchart showing the procedure of primary mesh model generation (ST104) shown in FIG. Schematic diagram explaining how to create a mesh model FIG. 3 is a flowchart showing the procedure for obtaining a cutout area (ST105) shown in FIG. FIG. 3 is a flowchart showing the procedure for obtaining a cutout area (ST105) shown in FIG. Schematic diagram explaining the point of cutout area acquisition FIG. 3 is a flowchart showing a procedure of secondary mesh model generation / image conversion (ST106) shown in FIG. Schematic diagram explaining the point of secondary mesh model generation and image conversion The schematic diagram which shows the state of the paper surface of the book at the time of designating the area | region cut out with the indicator F The schematic diagram which shows the designated area | region obtained by connecting the point which the indicator F points to, and the cut-out area | region which actually cuts out an image FIG. 3 is a flowchart showing the procedure of cutout image correction (ST107) shown in FIG. Schematic diagram showing the status of each process performed in cutout image correction Schematic diagram showing the status of each process performed in cutout image correction Schematic diagram showing the status of each process performed in cutout image correction Schematic diagram showing another example of processing performed in cutout image correction

  A first invention made to solve the above-described problem is that a first photographed image obtained by photographing a paper surface of a document together with a placement surface in the absence of a pointing material, and an indicator indicating an area to be cut out on the paper surface of the document A captured image acquisition unit that acquires a second captured image that is captured together with a paper surface of a document, a cutout region acquisition unit that detects a point indicated by an indicator from the second captured image and obtains a cutout region, and the first A cut-out image generation unit that obtains a cut-out image by flattening an image portion corresponding to at least the cut-out region obtained by the cut-out region acquisition unit in one captured image, and the cut-out region acquisition unit includes the indicator The cut-out area is obtained based on the designated area represented by the locus of one point indicated by.

  In the second invention, the cut-out area acquisition unit indicates a position where the indicator has stopped for a predetermined period when the indicator stops for a predetermined time after the operation of the indicator and further starts to operate. A plurality of position data is stored, and a cutout area is obtained based on the position data. According to the first and second aspects of the present invention, various areas can be designated as cut-out images by directly designating areas on the paper surface of the original with an indicator such as a finger or a pen. In addition, since a flattened cut-out image can be acquired, user convenience can be improved. In particular, if the configuration is such that only the image portion corresponding to the cutout region is planarized, it is possible to reduce the computational burden required for planarization and increase the processing speed.

  According to a third aspect of the present invention, in the first or second aspect of the invention, a primary mesh model generation unit that generates a primary mesh model based on the first photographed image, and the primary mesh model. A secondary mesh model generating unit that generates a secondary mesh model having a mesh interval smaller than this, and the cut-out region acquiring unit obtains the cut-out region using the primary mesh model, and the cut-out image The generation unit is configured to planarize an image using the secondary mesh model.

  According to this, by increasing the mesh interval of the primary mesh model, it is possible to reduce the calculation burden of the process for obtaining the cutout region. Further, since the image is planarized using a secondary mesh model with a small mesh interval, the conversion accuracy of the planarization can be improved, and a high-quality cutout image with small distortion can be acquired. In this case, a secondary mesh model may be generated by equally dividing the intersection points of the primary mesh model.

  In a fourth aspect based on the third aspect, the secondary mesh model generation unit generates the secondary mesh model only for the cutout region, and the cutout image generation unit includes the secondary mesh model. Only the image portion corresponding to the cut-out area is planarized by using.

  According to this, it is possible to reduce the calculation burden of each process of generating the secondary mesh model and planarizing.

  According to a fifth aspect of the invention, in the third or fourth aspect of the invention, the cut-out area acquisition unit obtains the cut-out area by demarcating a mesh line surrounding the specified area in the primary mesh model. To do.

According to this, since a necessary margin is ensured by appropriately setting the mesh interval of the primary mesh model, when the area is designated by the indicator, the image is displaced due to the paper being displaced by being pressed by the indicator. Regardless, the image of the area designated by the indicator can be reliably cut out. In this case, the mesh interval may be set based on the image shift amount when the paper surface is displaced by being pressed by the indicator.
A sixth invention relates to a document reading system, and includes the image processing device according to the first to fifth inventions, and an image input device having a camera unit for taking a captured image. .

  According to this, since the image of the specified area can be cut out by the operation of directly specifying the area with an indicator such as a finger or a pen on the paper surface of the document, a flattened cut-out image can be acquired. User convenience can be improved.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is an overall configuration diagram showing a document reading system according to the present embodiment. This document reading system reads a paper image of a book (original) B and acquires image data on the paper surface. The document reading system includes a book scanner (image input device) 1 and a PC (image processing device) 2. Yes.

  The book scanner 1 includes a camera unit 3 that captures the paper surface of the book B and a stand unit 4 that holds the camera unit 3. The stand unit 4 is installed on a mounting surface 5 such as a desk, and the camera unit 3 includes The book B is placed on the placing surface 5 immediately below, and the paper surface of the book B is imaged by the camera unit 3. The book B is imaged on the paper surface in a naturally opened state, and a two-page spread image of book B is obtained. Then, a process of flattening an image on a distorted paper surface is performed, and a flattened paper image for two pages can be acquired.

  Furthermore, here, in the operation of directly designating an area on the paper surface of Book B using an indicator F such as a finger or a pen, an image of the designated area is cut out to obtain a flattened cut out image. Can do. At this time, the book scanner 1 photographs the paper surface of the book B together with the placement surface 5 without the indicator F, and an image of the paper surface is generated in the PC 2 based on the photographed image. Further, in the book scanner 1, an indicator F indicating the area to be cut out on the paper surface of the book B is photographed together with the paper surface of the book B, and the cut out area is recognized in the PC 2 based on the photographed image.

FIG. 2 is a block diagram showing a schematic configuration of the book scanner 1 and the PC 2. The book scanner 1 includes an imaging processing unit 11 including a camera unit 3, an operation instruction unit 12, and an external interface 13. The PC 2 includes an external I / F 21, an image data input unit (captured image acquisition unit) 22, an image processing unit 23, an operation system control unit 24, a display data generation unit 25, a display 26, and an input unit 27. have. Note that the image processing unit 23 and the display data generation unit 25 of the PC 2 are realized by software processing in which a program such as an image processing application is executed by the CPU.

  In the PC 2, operating conditions such as the resolution and frame rate of an image captured by the book scanner 1 are input by operating the input unit 27 such as a keyboard, and the operating conditions are transmitted from the operation system control unit 24 to the book scanner 1. In the book scanner 1, the imaging processing unit 11 performs a required operation in accordance with an instruction from the operation instruction unit 12 based on the operation condition transmitted from the PC 2.

  The image data input unit 22 of the PC 2 stores the image data transmitted from the book scanner 1 in a memory, and outputs the image data to the image processing unit 23 as necessary. In the image data input unit 22, the first photographed image obtained by photographing the paper surface of the book B together with the placement surface 5 without the indicator F and the indicator F indicating the area to be cut out on the paper surface of the book B are booked. A second photographed image photographed together with B paper is acquired.

  The image processing unit 23 includes a contour acquisition unit 31, a primary mesh model generation unit 32, a cutout region acquisition unit 33, a secondary mesh model generation unit 34, a cutout image generation unit 35, an image correction unit 36, have.

  The contour acquisition unit 31 acquires the contour of book B from the first captured image. The primary mesh model generation unit 32 generates a primary mesh model from the contour information acquired by the contour acquisition unit 31 based on the first captured image. The cutout area acquisition unit 33 detects a point indicated by the pointing object from the second captured image and obtains a cutout area. The secondary mesh model generation unit 34 generates a secondary mesh model having a smaller mesh interval based on the primary mesh model generated by the primary mesh model generation unit 32.

  The cut-out image generation unit 35 planarizes the image portion corresponding to the cut-out area obtained by the cut-out area acquisition unit 33 in the first captured image, and acquires the cut-out image. The cut-out image generation unit 35 includes an image conversion unit 37 that converts an image on a distorted paper surface into an image without distortion. In the image conversion unit 37, the image is planarized by projective transformation (affine transformation). An image similar to that read by a flatbed scanner can be acquired. The image correction unit 36 performs image correction to delete unnecessary image portions that appear on the outer periphery of the cut image acquired by the cut image generation unit 35.

  The cut-out area acquisition unit 33 determines a cut-out area using the primary mesh model, and the cut-out image generation unit 35 planarizes the image using the secondary mesh model. This will be described in detail later.

  FIG. 3 is a flowchart showing the procedure of image cropping in the document reading system. First, when the book scanner 1 is activated and a required application is activated on the PC 2, image input, that is, imaging is started by the book scanner 1, and the imaging data is transmitted to the PC 2 (ST101). Next, when the user sets a book under the camera unit 3 of the book scanner 1 (ST102), the paper surface of the book B is imaged and the imaged data is transmitted to the PC 2, and the image data input unit 22 of the PC 2 uses the indicator. A first photographed image obtained by photographing the paper surface of book B together with the placement surface 5 without F is acquired.

In the PC 2, processing for acquiring the contour of the book B from the first photographed image is performed by the contour acquisition unit 31 (ST 103), and then the entire book B is targeted from the contour information acquired by the contour acquisition unit 31. Then, the primary mesh model generation unit 32 performs processing for generating a coarse primary mesh model (ST104).

  In addition, the user performs an operation of pointing an area to be cut out on the paper surface of the book B using the indicator F such as a finger or a pen, and obtains a second photographed image obtained by photographing the indicator F together with the paper surface of the book B Then, the cut-out area acquisition unit 33 performs processing for detecting the point indicated by the pointing object F from the second captured image and determining the cut-out area (ST105).

  Then, a secondary mesh model having a finer mesh than the primary mesh model is generated based on the primary mesh model generated by the primary mesh model generation unit 32 for the cut region acquired by the cut region acquisition unit 33. The secondary mesh model generation unit 32 performs the processing to perform, and the cut-out image generation unit 34 performs image conversion processing for planarizing the image based on the secondary mesh model (ST106).

  Next, image correction is performed in the image correction unit 35 to delete unnecessary image portions appearing on the outer peripheral portion of the cut image acquired by the cut image generation unit 35 (ST107), and the cut image obtained thereby is displayed on the display 26. (ST108). As a result, the user can confirm whether or not the image has been appropriately cut out.

  FIG. 4 is a flowchart showing a procedure of contour acquisition (ST103) shown in FIG. FIG. 5 is a schematic diagram for explaining the outline acquisition procedure. Here, from the first photographed image obtained by photographing the paper surface of the book B together with the placement surface 5 without the indicator F, the main information for roughly grasping the placement state of the book B as the outline information of the book B is shown. Detects linear components and singular points that are the top, bottom, left and right end points of the page

  First, the main linear component of the outline of the entire book is detected (ST201). Here, as shown in FIG. 5A, in the captured image, two straight lines indicating the left and right side edges (usually the side edges of the cover) in the entire outline of the book B, and the stitches of the left and right pages Is detected. The detection of this linear component may be performed by Hough transformation.

Also, edge detection is performed on the captured image (ST202). This edge detection may be performed using the Canny method. Next, contour components (pixels constituting the contour of the document) in the acquired edge image are extracted (ST203). Next, singular points that are the top, bottom, left, and right end points of the page are detected based on the acquired contour component (ST204). Here, as shown in FIG. 5B, six singular points are detected as the top, bottom, left, and right end points of the two-page spread image.

  Then, the validity of the singularity acquired in singularity detection (ST204) is determined (ST205). Here, the obtained singular point is compared with the linear component obtained in the straight line component detection (ST201), and the validity of the singular point, which is the upper, lower, left, and right end points of the page is determined. If it is determined that the singular point is valid, the singular point is determined (ST206).

  Note that the processing of edge detection (ST202), contour component extraction (ST203), and singularity detection (ST204) may be performed in parallel with the processing of linear component detection (ST201).

FIG. 6 is a flowchart showing a procedure of primary mesh model generation (ST104) shown in FIG. FIG. 7 is a schematic diagram for explaining a procedure for generating a primary mesh model. Here, a primary mesh model for the entire book surface is generated from the first photographed image obtained by photographing the paper surface of book B together with the placement surface 5 in the absence of the indicator F.

  First, a distortion component representing a distorted outline of the page of book B is extracted (ST301). As shown in FIG. 7A, the contour lines of the upper and lower edges of the page are displayed in a distorted state bulging outward, and the distorted contour components (contours of the upper and lower edges of the page are configured). Are extracted from the edge image acquired by the edge detection (ST202) shown in FIG. Then, the validity of the distortion component is determined based on the continuity of the distortion component (ST302). If it is determined that the distortion component is not valid, interpolation processing is performed for the interrupted portion (ST303). As a result, the distortion component becomes continuous without interruption.

  Next, an operation for converting a distortion component into a height component is performed (ST304). The reason why the upper and lower contours of the page are displayed in a distorted state on the photographed image is that the paper is curved so that it is convex upward in three dimensions. Based on the distortion component indicating the contour lines of the upper edge and the lower edge, the height component indicating the three-dimensional curved state of the upper edge and the lower edge of the page can be obtained. Specifically, the heights (Z-axis coordinates) of the upper and lower edges of the page with respect to the placement surface 5 are obtained at regular intervals. With this height component, the actual length of the upper and lower edges of the page can be estimated.

  Next, vertical and horizontal mesh lines are generated (ST305), and the coordinates of mesh intersections where the vertical and horizontal mesh lines intersect are stored in the mesh table (ST306). Here, as shown in FIG. 7B, based on the height components of the upper and lower edges of the page, the base point for dividing the curve indicating the upper and lower edges of the page at equal intervals according to the number of meshes. The vertical mesh lines are obtained by connecting the corresponding base points of the upper edge and the lower edge. Next, the coordinates of points (mesh intersections) at which the vertical mesh lines are divided at equal intervals according to the number of meshes are obtained, and the points are connected to obtain the horizontal mesh lines.

  FIGS. 8 and 9 are flowcharts showing the procedure of the cutout area acquisition (ST105) shown in FIG. 3. FIG. 8 shows the case of the one-stroke writing designation mode (first area designation mode), and FIG. The case of this finger designation mode (second area designation mode) is shown respectively. FIGS. 10A and 10B are schematic diagrams for explaining the procedure for obtaining the cut-out area. FIG. 10A shows the case of the single-stroke designation mode, and FIG. 10B shows the case of the two-finger designation mode.

  In the present embodiment, a one-stroke writing designation mode (first area designation mode) shown in FIG. 10A and a two-finger designation mode (second area designation mode) shown in FIG. 10B are provided. The user can select either.

  As shown in FIG. 10 (A), in the single stroke designation mode, the indicator F such as a finger or a pen is moved in the manner of a single stroke, and based on the designated area represented by the locus of one point indicated by the indicator F. Find the cutout area. In particular, here, the indicator F is moved in the left-right direction (a direction perpendicular to the center line serving as a page seam) and the up-down direction (a direction parallel to the center line). Specifies a polygonal area consisting of. As a result, it is possible to specify the region so that an extra image is not included.

  In this one-stroke writing designation mode, as shown in FIG. 8, first, the cut-out area acquisition unit 33 first captures the indicator F indicating the area to be cut out on the paper surface of the book B together with the paper surface of the book B. From this, the indicator F present on the paper surface of book B is detected (ST401). Here, the pointing object F is identified by the shape recognition process, and the pointing point indicated by the pointing object F is determined according to the form of the pointing object F. For example, if the pointing object is a finger, the fingertip is the pointing point.

Next, it is determined whether or not the indicator F has stopped moving for a predetermined time (for example, 1 second) (ST4).
02) If the indicator F has stopped moving, it is next determined whether or not the indicator F has started a prescribed operation (ST403). If the indicator F has started a prescribed operation, Then, the tracking process of the trajectory is started with the position where the indicator F first stops as a starting point, and the acquired coordinates of the trajectory are stored in the memory (ST404).

  Here, when the indicator F is stopped and moved in either the left or right direction, that is, in a direction substantially orthogonal to the center line that is the page seam, it is determined that the area designation has started, and the locus Start the tracking process. The locus of the pointing object is drawn on the screen of the captured image displayed on the display 26.

  Then, it is determined whether or not the specified designation operation has been completed, that is, whether or not the indicator has stopped moving in the vicinity of the start point (ST405). The process is terminated, and main points for area designation are extracted (ST406). Here, a polygonal region having a vertex at a bending point where the moving direction of the indicator changes by approximately 90 degrees is specified, and the bending point and the starting point are extracted as the main points for specifying the region, A polygonal designated area is obtained by connecting a plurality of extracted bending points with straight lines in the order of clockwise and counterclockwise directions indicated by the locus of the pointing object.

  When the indicator F for designating the area is a user's finger or a normal pen used for writing, a process for identifying the indicator F by the shape recognition process is necessary. It is also possible to use an indicator provided with a tip at the tip. In this case, since the infrared light emitted from the pointing tool appears in the captured image, the infrared light may be detected.

  On the other hand, as shown in FIG. 10B, in the two-finger designation mode, a cutout area is obtained based on a rectangular designated area having two points indicated by the indicator F as diagonal vertices. Thereby, the area designation by the indicator F can be easily performed. In particular, here, the region is specified using the left and right index fingers. A rectangular area in which the upper and lower sides are orthogonal to the center line indicating the joint of the page and the left and right sides are parallel to the center line, with the two points indicated by the left and right index fingers as diagonal vertices specify.

  In the two-finger designation mode, as shown in FIG. 9, first, the cut-out area acquisition unit 33 first shoots the pointing object F indicating the area to be cut out on the paper surface of the book B together with the paper surface of the book B. The indicator F present on the paper surface of book B is detected from the image (ST501). Then, it is determined whether or not the detected indicator F is the left and right forefinger (ST502), and if it is the left and right forefinger, the left and right forefinger stops moving for a predetermined time (for example, 1 second). (ST503) If the movement is stopped for a predetermined time, the coordinates of the pointing point indicated by the two left and right index fingers are obtained (ST504).

  Next, a straight line parallel to the center line indicating the joint of the page and a straight line perpendicular to the center line are extended from the two designated points respectively indicated by the left and right index fingers to obtain an intersection point thereof. The two intersections obtained in this way and the two indication points indicated by the index finger are extracted as the main points for specifying the region, and a rectangular region connecting the four main points becomes the specified region.

  FIG. 11 is a flowchart showing a procedure of secondary mesh model generation / image conversion (ST106) shown in FIG. FIG. 12 is a schematic diagram for explaining a procedure of secondary mesh model generation / image conversion. Here, a secondary mesh model is generated for the cutout region, and the image of the cutout region is planarized by projective transformation based on the secondary mesh model.

First, in the secondary mesh model generation unit 34, the coordinates of the main points of the designated area acquired by the cutout area acquisition shown in FIGS. 8 and 9 are combined with the coordinates of the primary mesh model (ST601). Then, a mesh intersection that defines a cut-out area where an image is actually cut out based on the designated area is extracted (ST602). Here, as shown in FIG. 12A, an area connecting mesh intersections outside the designated area is set as a cut-out area so as to surround the designated area. Then, as shown in FIG. 12B, a secondary mesh model in which the mesh interval is made finer than the primary mesh model is generated for the cutout region including the designated region (ST603).

  The secondary mesh model can be generated by obtaining mesh intersections that equally divide between mesh intersections of the primary mesh model. The average character size is about 10 points (width 3.5 mm). For example, when the mesh interval of the primary mesh model is 10.5 mm corresponding to three characters of an average 10 point character, 2 The mesh interval of the next mesh model is set to 3.5 mm, which is one character of 10 points, that is, 1/3 of the mesh interval of the primary mesh model.

  Next, planarization by projective transformation is performed in the cutout image generation unit 35 only for the image portion corresponding to the cutout region where the secondary mesh model is generated (ST604). As a result, as shown in FIG. 12C, it is possible to obtain a cut-out image in which only the cut-out area is planarized.

  Since the image is planarized using the secondary mesh model having a small mesh interval as described above, the conversion accuracy of the planarization can be improved, and a high-quality cutout image with small distortion can be acquired. Furthermore, since only the image portion corresponding to the necessary cut-out area is planarized, it is possible to reduce the calculation burden required for planarization and increase the processing speed.

  Here, an image shift that occurs when an area is designated by an indicator F such as a finger or a pen will be described. FIG. 13 is a schematic diagram illustrating a state of a book surface when an area to be cut out is designated by the indicator F. FIG. 14 is a schematic diagram showing a designated area obtained by connecting the points indicated by the indicator F and a cutout area where an image is actually cut out. FIG. 14 shows an example of an image in which a character string is written on the paper.

  As shown in FIG. 13, when a book page is naturally opened, there is a gap between the papers. Therefore, when the area is specified, if the book surface is pressed with an indicator F such as a finger or a pen, When pressed by the object F, the paper of book B bends, and the image on the paper surface shifts to the left and right according to the bending of the paper. Therefore, here, the cutout area acquisition unit 33 sets the cutout area so that a predetermined margin is secured around the designated area connecting the points indicated by the indicator F. As a result, an area larger than the designated area by the indicator F is cut out by a margin. Therefore, when the area is designated by the indicator F, the indicator F is used regardless of the image shift caused by the paper being displaced by being pressed by the indicator F. The image of the specified area can be cut out reliably.

  The margin width to be secured around the designated area when setting the cut-out area may be determined based on the amount of image shift caused by the paper being bent by being pressed by the indicator F such as a finger or a pen. The amount of image shift differs depending on the type of book (magazine, hardcover book, etc.) and paper quality, but is estimated to be about 5 mm at maximum, which is about 2 for an average 10-point (width 3.5 mm) character. It is preferable to secure a margin for at least two of these 10-point characters, corresponding to the characters.

In particular, here, as shown in FIG. 14, in the primary mesh model, a cut-out area is obtained by demarcating a mesh line surrounding a specified area. In the example shown in the figure, a mesh line that is one outer side than the mesh line immediately outside the designated area is used as the boundary line of the cut-out area. As a result, a margin greater than the mesh interval can always be secured, and the necessary margin is ensured by appropriately setting the mesh interval of the primary mesh model. Regardless of the above, it is possible to cut out the image of the area designated by the indicator without missing.

  In this case, the mesh interval of the primary mesh model may be set so that at least three characters are included in one mesh. Since the average character size is about 10 points (width 3.5 mm), if it is 3 characters of 10 point characters, it becomes a mesh of 3.5 × 3 = 10.5 mm square or more.

  FIG. 15 is a flowchart showing the procedure of the cutout image correction (ST702) shown in FIG. FIGS. 16, 17, and 18 are schematic diagrams illustrating the status of each process performed in cutout image correction. 16, 17, and 18 show examples of images in which a character string is written on a paper surface.

  As described above, the cut-out area is larger than the area actually specified by the indicator, and therefore an unnecessary image is included in the outer peripheral portion of the cut-out image acquired by the above processing. Therefore, an unnecessary portion is cut out from the cut image to obtain a final cut image.

  First, as shown in FIGS. 15 and 16, it is determined whether or not there is an image straddling the boundary line of the cut-out area (ST701). If there is a corresponding image, the boundary line is set inside the image. After setting, the corresponding image portion is cut out along the new boundary line (cutout line) (ST702). When a paper image is cut out, it is usually cut out so that the image is not divided in the middle. Therefore, an image straddling the boundary line of the cut-out area is considered as an unnecessary portion existing on an extra reserved margin and can be deleted.

  Next, as shown in FIGS. 15 and 17, it is determined whether or not there is a frame in the cut image (ST703). If there is a frame, the image outside the area surrounded by the frame is checked. A boundary line is set on the inner side, and the corresponding image portion is cut out along the new boundary line (cutout line) (ST704). When a region surrounded by a frame is cut out, usually only the region surrounded by the frame is cut out in many cases. Therefore, the image outside the area surrounded by the frame is considered as an unnecessary portion existing on an extra reserved margin and can be deleted.

  On the other hand, when there is no frame in the cut-out image, it is determined whether or not there is a space larger than the character spacing and line spacing around the character group (ST705). Edges without blanks are extracted (ST706).

  Thereafter, as shown in FIGS. 15 and 18, the phrase is determined from the position of the punctuation mark, and the image of the character group outside the character string having the punctuation mark existing on the outer periphery of the cut-out image is deleted. When cutting out a part of a sentence, usually the range delimited by punctuation marks is often cut out, and it is not cut out in the middle of a phrase, so the image of the character group outside the punctuation marks existing in the outer periphery of the cut-out image is It can be deleted because it is considered as an unnecessary part existing on an extra margin. Also, the parts that can be deleted based on punctuation marks differ depending on the writing direction (horizontal writing and vertical writing). If horizontal writing, the image of the character group above and below the phrase to be left is left if vertical writing. What is necessary is just to delete the image of the character group in the right and left of the power clause.

  Here, first, simple OCR is performed on the character string in the cut-out image (ST707), and the writing direction, that is, vertical writing or horizontal writing is determined based on the result of the simple OCR (ST708). Here, it is determined based on the positional relationship of the punctuation marks. When the punctuation mark position is on the lower right side of the previous character, it is written vertically, and when the punctuation mark position is on the lower right side of the previous character, Judged as horizontal writing. In the case of vertical writing, the subsequent processing is limited to the sides where there are no white spaces on the left and right sides (ST709), and in the case of horizontal writing, the subsequent processing is limited to the sides where there are no white space on the upper and lower sides (ST709). ST710).

  Next, it is determined whether or not there is a punctuation mark at the end of the character string on the first line (ST711). If there is a punctuation mark at the end of the character string on the first line, a boundary is placed inside the character string on the first line. A line is set, and the image portion of the character string in the first row is cut along the new boundary line (cut line) (ST712). Next, it is determined whether or not there is a punctuation mark at the end of the character string on the second line from the last line (ST713). If there is a punctuation mark at the end of the character string on the last line, the inside of the character string on the last line. A boundary line is set to the image line, and the image portion of the character string on the last line is cut out along the new boundary line (cutout line) (ST714).

  FIG. 19 is a schematic diagram illustrating another example of processing performed in cutout image correction. Here, the image outside the blank larger than the character spacing and line spacing is deleted from the cut image. When there is a space larger than the character spacing and line spacing in the text, the character group is usually cut out with the space position as the boundary, so the image outside the space exists on an extra reserved margin. It is considered an unnecessary part and can be deleted. In this case, an unnecessary character string can be deleted without performing the phrase determination as described above.

  In the above example, as shown in FIG. 2, the photographed image data is sent from the book scanner to the PC and the image is cut out by the PC. Such image processing is performed by the book scanner. That is, a configuration in which the book scanner includes an image processing apparatus is also possible. In this case, an image may be displayed on a display device such as a television receiver.

  In the above example, only the cutout area including the area specified by using an indicator such as a finger or a pen is flattened to obtain a cutout image. However, after the entire paper surface of the book is flattened, The image portion corresponding to the cutout region may be cut out.

  The image processing apparatus according to the present invention has an effect that a clipped image can be acquired by an operation of directly designating an area with a pointing object such as a finger on a document such as a book. The present invention is useful as an image processing apparatus that processes an image obtained by reading and a document reading system including the image processing apparatus.

1 Book scanner (image input device)
2 PC (image processing device)
3 Camera unit 22 Image data input unit (captured image acquisition unit)
31 Outline acquisition unit 32 Primary mesh model generation unit 33 Cutout region acquisition unit 34 Secondary mesh model generation unit 35 Cutout image generation unit 36 Image correction unit 37 Image conversion unit

Claims (6)

A first photographed image obtained by photographing the paper surface of the document together with the placement surface in the absence of an indicator is obtained, and a second photographed image obtained by photographing the indicator indicating the area to be cut out on the paper surface of the document together with the paper surface of the document. A captured image acquisition unit,
A cutout area acquisition unit for detecting a point indicated by the indicator from the second photographed image and obtaining a cutout area;
A cut-out image generation unit that obtains a cut-out image by planarizing an image portion corresponding to at least the cut-out region obtained by the cut-out region acquisition unit in the first captured image;
The image processing apparatus is characterized in that the cut-out area acquisition unit obtains the cut-out area based on a designated area represented by a locus of one point indicated by the indicator. The cutout area acquisition unit stores a plurality of position data indicating positions where the indicator has stopped for a predetermined period when the indicator stops for a predetermined time after the operation of the indicator and further starts the operation. The image processing apparatus according to claim 1, wherein a cut-out area is obtained based on position data. A primary mesh model generation unit that generates a primary mesh model based on the first captured image, and a secondary mesh model generation that generates a secondary mesh model having a smaller mesh interval based on the primary mesh model. And further comprising
The cutout area acquisition unit obtains the cutout area using the primary mesh model,
The image processing apparatus according to claim 1, wherein the cut-out image generation unit performs planarization of an image using the secondary mesh model. The secondary mesh model generation unit generates the secondary mesh model only for the cutout region,
The image processing apparatus according to claim 3, wherein the cut-out image generation unit planarizes only an image portion corresponding to the cut-out region using the secondary mesh model.   5. The image processing apparatus according to claim 3, wherein the cut-out area acquisition unit obtains the cut-out area by demarcating a mesh line surrounding the specified area in the primary mesh model. .   6. A document reading system comprising: the image processing apparatus according to claim 1; and an image input apparatus having a camera unit that captures the captured image.

Images