JP6677412B2 - IMAGE EXTRACTION DEVICE, IMAGE EXTRACTION DEVICE CONTROL METHOD, AND PROGRAM - Google Patents

Description

The present invention relates to an image extraction device , a control method of the image extraction device , and a program.

  2. Description of the Related Art In recent years, mobile terminals such as smartphones, mobile phones, and tablets having a camera function have become widespread, and photography can be easily performed.

  Also, various kinds of application software for processing and using photographs taken using such portable terminals have been developed. For example, a technology for digitizing a business card photographed using a portable terminal has been developed (for example, see Patent Document 1).

  Further, a technique for extracting a contour from a set of edge points in an image has been developed (for example, see Patent Document 2).

JP 2003-296353 A JP-A-2004-234278

  However, application software executed on a mobile terminal is required to operate at high speed so as not to lengthen the user's waiting time, and the same applies to the case where a business card is digitized as in Patent Document 1.

  Therefore, there is a need for a technique that enables a line segment forming a contour of a polygonal object to be photographed such as a business card to be specified at higher speed from a still image photographed by a camera.

The present invention has been made in view of the above problems, and provides an image extraction device, a control method of the image extraction device , and a program that can more quickly specify a rectangular outline such as a business card from an image. For one purpose.

An image extraction device according to one aspect is an image extraction device that extracts a rectangle from an image, and includes a line segment extraction unit that extracts a line segment from the image, and a common line segment among the line segments extracted from the image. After extracting two line segments having an end point and forming an angle with each other as a line segment pair, the length of four line segments is determined for each combination in which two line segment pairs are selected from the line segment pairs. A polygon forming unit that forms a rectangle by adjusting the length, and an adjustment degree calculation unit that calculates, for each combination, a value representing the degree of adjustment of the length of the line segment required to form the rectangle And an outline specifying unit that specifies the outline of a rectangle to be extracted from the image in the order of the degree of the adjustment amount from the rectangles formed for each of the combinations.

  In addition, the problems disclosed by the present application and the solutions thereof will be clarified by the description of the embodiments for carrying out the invention, the description of the drawings, and the like.

It is possible to specify a rectangular outline from an image at higher speed.

FIG. 2 is a diagram illustrating a hardware configuration of a line segment extraction device according to an embodiment of the present invention. 1 is a diagram illustrating a functional configuration of a line segment extraction device according to an embodiment of the present invention. It is a figure showing signs that the outline of the business card was specified from the still picture concerning one embodiment of the present invention. FIG. 3 is a diagram illustrating a still image according to an embodiment of the present invention. FIG. 7 is a diagram illustrating a state where edges are detected from a still image according to an embodiment of the present invention. FIG. 4 is a diagram illustrating a state where a line segment is extracted from a still image according to an embodiment of the present invention. It is a figure showing signs that an intersection of an edge and a mesh line is specified in a still picture concerning one embodiment of the present invention. It is a figure showing signs that a line segment was extracted from a still picture concerning one embodiment of the present invention. FIG. 4 is a diagram illustrating line segments extracted from a still image according to an embodiment of the present invention. It is a figure showing signs that a polygon is formed by combining line segments concerning one embodiment of the present invention. It is a figure showing signs that a polygon is formed by combining line segments concerning one embodiment of the present invention. 5 is a flowchart illustrating a flow of a process according to an embodiment of the present invention. 5 is a flowchart illustrating a flow of a process according to an embodiment of the present invention. 5 is a flowchart illustrating a flow of a process according to an embodiment of the present invention.

  At least the following matters will be made clear by the description in the present specification and the accompanying drawings. Hereinafter, the present invention will be described in accordance with one embodiment with reference to the accompanying drawings.

== Line segment extraction device ==
1 and 2 show a configuration of a line segment extraction device 100 according to an embodiment of the present invention. The line segment extraction device 100 according to the present embodiment is a portable information processing terminal having a camera function such as a smartphone, a mobile phone, a notebook computer, and a tablet.

  Then, when the line segment extraction device 100 according to the present embodiment captures a business card, it performs image analysis of the still image 400 (described later), specifies the outline of the business card from the still image 400, and obtains an image of the business card portion (the business card The image of the part surrounded by the outline) is extracted as the business card image 410. Thus, the business card can be digitized.

<Hardware configuration>
FIG. 1 shows a hardware configuration diagram of the line segment extraction device 100. The line segment extraction device 100 includes a CPU 110, a memory 120, a communication device 130, a storage device 140, an input device 150, an output device 160, and a recording medium reading device 170.

  The storage device 140 stores the program 700 executed by the line segment extraction device 100 and various data. By reading the program 700 and data stored in the storage device 140 into the memory 120 and executing them by the CPU 110, various functions of the line segment extraction device 100 are realized. Here, the storage device 140 is a nonvolatile storage device such as a hard disk, a solid state drive (SSD), and a flash memory.

  The recording medium reader 170 reads the program 700 and data recorded on a recording medium 800 such as an SD card, and stores the program 700 and data in the storage device 140.

  The communication device 130 exchanges various data and programs 700 with another computer (not shown) via a communication network (not shown) such as a LAN (Local Area Network) or the Internet. For example, the program 700 and the data described above may be stored in another computer, and the line segment extracting apparatus 100 may download the data and the program 700 from this computer.

  The input device 150 includes various buttons and switches for receiving command and data input by the user, a keyboard, an input interface such as a touch sensor for detecting a touch position on a touch panel display, a microphone, an acceleration sensor, a temperature sensor, a camera 151, and the like. is there.

  The output device 160 is an output user interface such as a display device such as a display 161, a speaker, a vibrator, and lighting.

<Functional configuration>
FIG. 2 shows a functional block diagram of the line segment extraction device 100. The line segment extraction device 100 includes an image acquisition unit 101, an image display unit 102, and an image extraction unit 104. Further, the image extracting unit 104 includes a line segment extracting unit 900, a polygon forming unit 910, an adjustment degree calculating unit 920, a contour specifying unit 930, and a shape comparing unit 940.

  Furthermore, the line segment extracting unit 900 includes an edge detecting unit 901, a first intersection specifying unit 902, a first intersection connecting unit 903, a first line segment specifying unit 904, a second intersection specifying unit 905, and a second intersection connecting unit 906. , And a second line segment identification unit 907.

  These functions are realized by executing the program 700 according to the present embodiment by the hardware of the line segment extraction device 100.

  The image acquisition unit 101 acquires a still image 400 by photographing a business card. The image acquisition unit 101 is realized using, for example, the camera 151.

  The still image 400 may be an image of a single photograph obtained at the moment when the shutter of the camera 151 is pressed, or may be an individual image composed of a plurality of frames captured as a moving image. Furthermore, the still image 400 may be an image detected by a CCD (Charge Coupled Device), which is displayed on the display 161 before the start of capturing a moving image or before the shutter is pressed during photographing.

  The still image 400 may include not only a business card but also a background other than a business card, such as a background of a business card or a person. Further, a plurality of business cards may be included in one still image 400.

  The image display unit 102 displays the still image 400 of the business card acquired by the image acquisition unit 101. The image display unit 102 is realized using, for example, a touch panel type display 161.

  FIG. 3 shows a state in which the still image 400 is displayed on the display 161 of the line segment extraction device 100. The example shown in FIG. 3 shows a case where three business cards are photographed simultaneously. As shown in FIG. 3, in the present embodiment, when a plurality of business cards are photographed at the same time, the directions of the respective business cards do not have to be aligned.

  Then, as described in detail below, the line segment extracting apparatus 100 according to the present embodiment specifies the outline of the business card from the still image 400 at high speed, and outputs the image of the business card portion (the image of the portion surrounded by the business card outline). ) Is extracted as a business card image 410.

  Then, the image display unit 102 displays the business card marker 500 so as to overlap the business card image 410 in order to show the extracted business card image 410 to the user. In the present embodiment, the business card marker 500 is a circular image of a predetermined color, but the display mode is not particularly limited. By displaying the business card marker 500 over the business card image 410 in this manner, the user can confirm that the business card has been correctly extracted.

  Returning to FIG. 2, the image extracting unit 104 will be described.

  The image extracting unit 104 extracts a business card image 410 from the still image 400.

  Hereinafter, the functions and processing procedures of the image extracting unit 104 will be specifically described with reference to FIGS.

  First, FIG. 4 shows a still image 400 before the business card image 410 is extracted. When the image acquisition unit 101 acquires such a still image 400, the edge detection unit 901 adjusts the size of the still image 400 to a predetermined size according to the number of pixels, and then detects the edge 420 from the still image 400. . The edge detection unit 901 detects an edge by using a known technique such as a method using a Canny method or a Sobel Filter.

  FIG. 5 shows how the edge detection unit 901 has detected the edge 420 from the still image 400.

  Then, the first intersection specifying unit 902 obtains an intersection 422 between the plurality of first mesh lines (first parallel lines) 421 extending at predetermined intervals in the first direction in the still image 400 and the edge 420. . The state is shown in FIGS. 6A, 6B, and 6C.

  FIG. 6A is a diagram illustrating an edge 420 detected from the still image 400 and a plurality of first mesh lines 421 extending in the first direction in the still image 400.

  In this state, the first intersection specifying unit 902 specifies the intersection 422 between the edge 420 and the first mesh line 421 as shown in FIG.

  The edge 420 and the first mesh line 421 are a set of discontinuous points as shown in FIG. 7A when the still image 400 is viewed at a level at which individual pixels can be identified. Therefore, when the intersection angle between the edge 420 and the first mesh line 421 is small, the point where the edge 420 intersects with the first mesh line 421 is dispersed into a plurality of pixels as shown in FIG. There is.

  Therefore, in the present embodiment, when the points where the edge 420 intersects with the first mesh line 421 are distributed to a plurality of pixels, as shown in FIG. A point (for example, a middle point) is selected and specified as the intersection 422. According to such an embodiment, it is possible to specify the intersection 422 between the edge 420 and the first mesh line 421 with pixel-by-pixel accuracy. FIG. 6C shows how the first intersection specifying unit 902 finds the intersection 422 in this manner.

  Then, as shown in FIG. 6D and FIG. 6E, the first intersection coupling unit 903 performs the first mesh line operation for each pair of two adjacent first mesh lines 421. The intersection points 422 on the line 421 are connected by a straight connection line 423.

  Then, as shown in FIG. 6F, the first line segment specifying unit 904 is composed of a plurality of connection lines 423 connected at the intersection 422, and the angle difference in the stretching direction is within a predetermined range (for example, within ± 5 °). A set of connection lines 423 is extracted as a line segment 430. At this time, the first line segment identification unit 904 extracts, as the line segment 430, a set of connection lines 423 having an intersection angle of a predetermined angle or more (for example, 45 ° or more) with respect to the first mesh line 421.

  Thereafter, the second intersection specifying unit 905, the second intersection connecting unit 906, and the second line segment specifying unit 907 use a second mesh line 421 (not shown) orthogonal to the extending direction of the first mesh line 421, respectively. A line segment 430 is further extracted by performing the same processing as that of the first intersection specifying unit 902, the first intersection connecting unit 903, and the first line segment specifying unit 904.

  That is, the second intersection specifying unit 905 obtains the intersections 422 between the plurality of second mesh lines 421 extending at predetermined intervals in the second direction orthogonal to the first mesh lines 421 and the edges 420. Then, for each pair of two adjacent second mesh lines 421, the second intersection connection unit 906 connects the intersections 422 on the second mesh lines 421 with a straight connection line 423. Then, the second line segment identification unit 907 includes a plurality of connection lines 423 connected at the intersection 422, and the second line segment identification unit 907 includes a second connection line 423 of a set of connection lines 423 in which the angle difference in the extending direction is within a predetermined range (for example, within ± 5 °). A set of connection lines 423 having an intersection angle exceeding the complementary angle of the predetermined angle (eg, exceeding 45 °) with the mesh line 421 is extracted as a line segment 430.

  According to such an embodiment, when the line segment 430 is extracted, the process for the intersection 422 may be performed, and the process for a large number of points on the edge 420 can be omitted. Becomes possible.

  Also, unlike the case where a straight line is detected by using the Hough transform or the like, since the boundary of the subject in the still image 400 can be extracted as a line segment 430 of a finite length, as described below, these line segments 430 are combined. When forming the polygon 432, it is possible to calculate the ratio of the length (adjustment amount) of the adjustment portion 431 of the line segment 430 to the circumference of the polygon 432.

  Further, by determining the intersection 422 with the edge 420 using the first mesh line 421 and the second mesh line 421, even if the edge 420 extends in any direction in the still image 400, the intersection angle with the mesh line 421 is obtained. Can be specified such that the angle of intersection is equal to or greater than 45 °, so that the position of the intersection 422 on the edge 420 can be specified more accurately.

  Note that when the interval between the mesh lines 421 is reduced, the number of intersections 422 can be increased, so that the line segment 430 can be extracted more accurately, but the processing time increases accordingly. The mesh lines 421 are preferably drawn at intervals of, for example, about 50 to 100 lines in the still image 400, but can be variously changed.

  In this way, the first line segment specifying unit 904 and the second line segment specifying unit 907 specify the line segment 430 from the still image 400. FIG. 8 shows how the line segment 430 is specified from the still image 400.

  After that, the polygon forming unit 910 selects a line segment corresponding to the number of business card sides (that is, four line segments) from the line segments 430 extracted from the still image 400 for each combination. The length of the minute 430 is adjusted to form a polygon 432 having the same number of vertices as the business card.

For example, as shown in FIG. 9, when five line segments 430A to 430E are extracted from the still image 400, combinations of selecting four line segments from these line segments 430A to 430 are as follows (1) to (5). is a five types _{(5 C} 4 = 5).

(1) 430A, 430B, 430C, 430D
(2) 430A, 430B, 430C, 430E
(3) 430A, 430B, 430D, 430E
(4) 430A, 430C, 430D, 430E
(5) 430B, 430C, 430D, 430E
The polygon forming unit 910 attempts to form a square 432 using each line segment 430 for each of these combinations. However, since the polygon forming portion 910 is a parallel line in which the line segment 430B, the line segment 430D, and the line segment 430E extend in the same direction, the combinations (3) and (5) including these three line segments 430 at the same time Can be removed from the combination. According to such an embodiment, it is possible to efficiently eliminate unnecessary processing and extract business cards at higher speed.

  In this way, the polygon forming unit 910 forms a quadrangle 432 of three combinations as shown in FIGS. 10A to 10C. FIG. 10A corresponds to the combination of (1), FIG. 10B corresponds to the combination of (2), and FIG. 10C corresponds to the combination of (4).

  In this manner, for each combination shown in FIGS. 10A to 10C, a state in which the polygon forming unit 910 forms the square 432 is shown in FIGS. 11A to 11C, respectively. Show.

  In the case shown in FIG. 11A, the polygon forming unit 910 forms the square 432 by adjusting the length of the line segments 430A, 430B, 430C, and 430D. In FIG. 11A, an adjustment portion of the length of each line segment 430 is indicated by a broken line as an adjustment portion 431.

  The polygon forming unit 910 forms the square 432 by adjusting the length of the line segment 430 in the case shown in FIGS.

  Then, the adjustment degree calculation unit 920 calculates a value indicating the degree of adjustment of the length of the line segment 430 required to form the square 432 for each combination. For example, the adjustment degree calculation unit 920 calculates the ratio of the length (adjustment amount) of the adjustment portion 431 to the circumference of the rectangle 432 formed by adjusting the length of the line segment 430.

  Then, the contour specifying unit 930 assigns each line segment 430 constituting the combination as a contour of the business card in order from the combination having the smallest adjustment amount.

  For example, when it is assumed that the value indicating the degree of the adjustment amount is small in the order of FIGS. 11A, 11B, and 11C, the contour specifying unit 930 firstly determines the line segment 430 of the combination shown in FIG. (Line segments 430A, 430B, 430C, and 430D) are specified as the outline of the business card.

  In this way, when the line segments 430A, 430B, 430C, and 430D are assigned as the outlines of the business card, the remaining line segment is only one line segment 430E, and the square 432 can no longer be formed.

  Thus, the line segment extraction device 100 according to the present embodiment specifies the outline of the business card. As shown in FIG. 3, even when a plurality of business cards exist in the still image 400, the outline of the business card can be specified in ascending order of the adjustment amount for each combination of the line segments 430. It is.

  Note that the line segment extraction device 100 according to the present embodiment further includes a shape comparison unit 940.

  The shape comparing unit 940 compares an index value representing the shape of the square 432 formed by adjusting the length of the line segment 430 with a reference value representing the shape of the business card.

  Then, the contour specifying unit 930 combines, in order from the combination of the line segments 430 where the square 432 is formed such that the index value is within a predetermined range with respect to the reference value, the combination in which the degree of the adjustment amount is small, in order. Is specified as the outline of the business card.

  For example, the shape comparison unit 940 compares the intersection angle of the adjacent line segments 430 with a reference value in the square 432 formed by combining four line segments 430. If the reference value is set to, for example, 90 ° ± 5 °, the contour specifying unit 930 sets the inner angle of 90 ° ± 5 ° in various rectangles 432 formed by combining four line segments 430. A rectangle or a square within the range is selected, and among these, the line segment 430 constituting the combination is specified as the outline of the business card in order from the combination having the smallest adjustment amount. According to such an embodiment, it is possible to more accurately specify a business card.

  Alternatively, the shape comparison unit 940 compares the length of each line segment 430 with a reference value in a square 432 formed by combining four line segments 430. As a reference value, for example, if the short side is set to 50 mm to 60 mm and the long side is set to 85 mm to 95 mm, the contour specifying unit 930 includes various rectangles 432 formed by combining four line segments 430. A rectangle having such a side length is selected, and a line segment 430 constituting the combination is specified as the outline of the business card in order from the combination having the smallest adjustment amount. According to such an embodiment, the business card can be specified more accurately.

  Next, the flow of processing of the line segment extraction device 100 according to the present embodiment will be described with reference to the flowcharts shown in FIGS.

  First, the line segment extracting apparatus 100 acquires a still image 400 by photographing a business card (S1000). The image acquisition unit 101 is realized using, for example, the camera 151.

  Next, the line segment extraction device 100 adjusts the size of the still image 400 to a predetermined size determined according to the number of pixels (S1010). For example, the line segment extraction device 100 adjusts one side of the still image 400 to be 1000 pixels or less. According to such an embodiment, it is possible to reduce the processing amount when the line segment extracting apparatus 100 performs the image analysis, and to quickly specify the outline of the business card from the still image 400.

  Next, after detecting the edge 420 from the still image 400 (S1020), the line segment extracting apparatus 100 extracts a line segment 430 from the still image 400 from which the edge is detected (S1030).

  FIG. 13 shows a processing flow when the line segment extracting apparatus 100 extracts the line segment 430 from the still image 400.

  First, the line segment extraction device 100 acquires the still image 400 in which the edge 420 has been detected (S2000).

  Then, the line segment extraction device 100 draws a plurality of first mesh lines 421 extending at predetermined intervals in the first direction in the still image 400 (S2010).

  Next, the line segment extraction device 100 obtains an intersection 422 between the first mesh line 421 and the edge 420 (S2020).

  Then, for each pair of two adjacent first mesh lines 421, the line segment extraction device 100 connects the intersection points 422 on the first mesh lines 421 with a straight connection line 423 (S2030).

  Then, the line segment extracting apparatus 100 sets, as candidates for the line segment 430, a set of the connection lines 423 including a plurality of connection lines 423 connected at the intersection 422 and having an angle difference in the extending direction within a predetermined range (S2040).

  At this time, the line segment extracting apparatus 100 extracts the line segment 430 from the set of the connection lines 423 that are candidates for the line segment 430, excluding the one whose edge 420 is not extracted at the position of the connection line 423. (S2050).

  When the processing for the first mesh line 421 extending in the first direction is completed in this way, the line segment extraction device 100 rotates the still image 400 by 90 ° (S2060, S2070), and performs the processing of S2010 to S2050. repeat. To perform the same processing by rotating the still image 400 by 90 ° is that the second intersection specifying unit 905, the second intersection connecting unit 906, and the second line segment specifying unit 907 use the second mesh line 421 described above. This is equivalent to performing a process of extracting the line segment 430.

  Then, the line segment extraction device 100 rotates the direction of the still image 400 by −90 ° to return to the original position (S2080), and outputs the detected line segment 430 (S2090).

  Returning to FIG. 12, the line segment extraction apparatus 100 searches for a square 432 that can be formed by combining the extracted line segments 430 (S1040).

  Specifically, as shown in FIG. 14, the line segment extracting apparatus 100 firstly has a number of line segments corresponding to the number of business card sides (that is, four line segments) from among the line segments 430 extracted from the still image 400. The length of the line segment 430 is adjusted for each combination for selecting the line segment, thereby forming a polygon 432 (that is, a square 432) having the same number of vertices as the business card (S3000).

  Then, the line segment extracting apparatus 100 removes, out of these quadrangles 432, those whose angle (inner angle) formed by the quadrilateral 432 is not in the range of, for example, 90 ± 5 ° (S3010, S3020). According to such an embodiment, it is possible to efficiently eliminate unnecessary processing and extract business cards at higher speed.

  Then, the line segment extraction device 100 calculates, for each combination of the line segments 430 constituting the rectangle 432, a value (penalty) representing the degree of the amount of adjustment of the length of the line segment 430 required to form the rectangle 432. (S3030) and output (S3040). For example, the adjustment degree calculation unit 920 calculates the ratio of the length (adjustment amount) of the adjustment portion 431 to the circumference of the rectangle 432 as a penalty.

  Then, the line segment extracting apparatus 100 rearranges the combinations of the line segments 430 in ascending order of penalty (S3050).

  Then, the line segment extraction device 100 assigns each line segment 430 constituting the combination as the outline of the business card in order from the combination having the smallest penalty (S3060 to S3090). When the line segment 430 that can form the square 432 is gone (S3100), the line segment 430 specified as the outline of the business card is displayed on the display 161 (S3110). At this time, the line segment extracting apparatus 100 also displays the business card marker 500 so as to overlap with the business card image 410 in a portion surrounded by the outline of the business card.

  Returning to FIG. 12, when the outline of the business card is specified from the still image 400 in this way, the line segment extracting apparatus 100 performs the same process on the still image 400 of the next frame. In this way, the line segment extraction device 100 can specify the outline of the business card even for a continuous still image 400 such as a moving image.

== Other Embodiments ==
In the above-described embodiment, the polygon forming unit 910 selects the number of line segments (that is, four line segments) corresponding to the number of sides of the business card from the line segments 430 extracted from the still image 400 for each combination. In addition, the length of the line segment 430 is adjusted to form a polygon (that is, a quadrangle) 432 having the same number of vertices as the business card. In this case, from the extracted n line segments 430, four When the square 432 is formed, since there are simply _four combinations of _n C, the amount of calculation is on the order of the fourth power of n.

  Therefore, in order to reduce the amount of calculation, the polygon forming unit 910 forms an angle with each other having a common end point among the line segments 430 extracted from the still image 400 (for example, in a range of 90 ° ± 5 °). The two line segments 430 are extracted as a line segment pair, and the length of the four line segments 430 is determined for each combination in which two line segment pairs are selected from these line segment pairs. The square 432 may be formed by adjusting the height.

  That is, instead of directly selecting four lines from the extracted n line segments 430, a line segment pair consisting of two line segments 430 that intersect at once at an angle that is substantially orthogonal to each other (within 90 ° ± 5 °) ( The shape of the line segment pair is "" or "" in a bracket shape, and the line segment pairs are combined to form a square 432.

According to such an embodiment, since the number of combinations is _n C ₂ + _{(n / 2)} C ₂ , the amount of calculation can be on the order of the square of n. As a result, it is possible to more quickly specify the outline of the business card.

  Note that the end points of the two line segments 430 forming a line segment pair are common only when the end point of the first line segment 430 and the end point of the second line segment 430 are located on the same pixel. But not within a predetermined range (for example, within 10 pixels).

  As described above, the line segment extraction device 100, the control method of the line segment extraction device 100, and the program 700 according to the present embodiment have been described in detail. It becomes possible to specify.

  It should be noted that the above-described embodiment is intended to facilitate understanding of the present invention, and is not intended to limit the present invention. The present invention can be changed and improved without departing from the gist thereof, and the present invention also includes equivalents thereof.

  For example, in the above-described embodiment, the case where the outline of the business card is specified from the still image 400 has been described as an example, but not only the outline of a rectangular plane such as a business card is specified, but also a triangle, a pentagon, and a hexagon The same can be applied to the case where the outline of a polygonal plane is specified.

100 line segment extraction device 101 image acquisition unit 102 image display unit 104 image extraction unit 110 CPU
Reference Signs List 120 memory 130 communication device 140 storage device 150 input device 151 camera 160 output device 161 display 170 recording medium reading device 400 still image 410 business card image 420 edge 421 mesh line 422 intersection 423 connection line 430 line segment 431 line segment adjustment part 432 many Square (square)
500 Business card marker 700 Program 800 Recording medium 900 Line segment extraction unit 901 Edge detection unit 902 First intersection specifying unit 903 First intersection specifying unit 904 First line specifying unit 905 Second intersection specifying unit 906 Second intersection connecting unit 907 2 line segment specifying section 910 Polygon forming section 920 Adjustment degree calculating section 930 Outline specifying section 940 Shape comparing section

Claims (9)

An image extraction device for extracting a rectangle from an image,
A line segment extraction unit that performs line segment extraction from the image,
After extracting two line segments having a common end point and forming an angle from each other among the line segments extracted from the image as a line segment pair, two line segment pairs are extracted from the line segment pairs. A polygon forming unit that adjusts the length of the four line segments to form a rectangle for each combination to be selected;
For each combination, an adjustment degree calculation unit that calculates a value indicating a degree of an adjustment amount of the length of the line segment necessary to form the rectangle,
From among the rectangles formed for each of the combinations, an outline specifying unit that specifies the outline of the rectangle to be extracted from the image in order from the combination having the smaller adjustment amount ,
An image extraction device comprising: The image extraction device according to claim 1 ,
A shape comparison unit that compares a predetermined index value representing a rectangular shape formed by adjusting the length of the line segment with a reference value representing a rectangular shape to be identified from the image,
Further comprising
The outline identification unit is configured such that the index value falls within a predetermined range with respect to the reference value. An image extracting apparatus, wherein each line segment constituting a combination is specified as a rectangular outline to be extracted from the image. The image extraction device according to claim 2 ,
The image extraction device, wherein the index value and the reference value include an intersection angle between two adjacent sides forming the rectangle. The image extraction device according to claim 2 or 3 ,
The image extraction device according to claim 1, wherein the index value and the reference value include a length of each side of the rectangle. The image extraction device according to any one of claims 1 to 4 ,
The image extraction device according to claim 1, wherein the value indicating the degree of the adjustment amount is a ratio of the adjustment amount to a circumference of a rectangle formed by adjusting the length of the line segment. The image extraction device according to any one of claims 1 to 5 ,
The line segment extraction unit includes:
An edge detection unit that detects an edge from the image,
A first intersection specifying unit that obtains an intersection between the plurality of first parallel lines extending at a predetermined interval in the first direction in the image and the edge;
For each pair of two adjacent first parallel lines, a first intersection connecting portion connecting the intersections on the first parallel lines with a straight connecting line;
A first line segment identification unit that is composed of a plurality of the connection lines connected at the intersections, and extracts a set of the connection lines in which the angle difference in the stretching direction is within a predetermined range, as the line segment,
An image extraction device comprising: The image extraction device according to claim 6 ,
The line segment extraction unit further includes:
A plurality of second parallel lines extending at predetermined intervals in a second direction orthogonal to the first direction in the image, and a second intersection specifying unit for obtaining an intersection with the edge;
For each pair of two adjacent second parallel lines, a second intersection connecting portion connecting the intersections on each other's second parallel lines with a straight connecting line;
A second line segment identification unit that is composed of a plurality of the connection lines connected at the intersection, and extracts a set of the connection lines in which the angle difference in the stretching direction is within a predetermined range, as the line segment,
With
The first line segment identification unit extracts, as the line segment, a set of the connection lines having an intersection angle equal to or greater than a predetermined angle with respect to the first parallel line,
The image extraction device, wherein the second line segment identification unit extracts, as the line segment, a set of the connection lines having an intersection angle exceeding the complementary angle of the predetermined angle with respect to the second parallel line. A control method of an image extraction device for extracting a rectangle from an image,
The image extraction device,
Extracting a line segment from the image,
After extracting two line segments having a common end point and forming an angle from each other among the line segments extracted from the image as a line segment pair, two line segment pairs are extracted from the line segment pairs. For each combination you choose, adjust the length of the four line segments to form a rectangle,
For each combination, calculate a value representing the degree of adjustment of the length of the line segment required to form the rectangle,
From among the rectangles formed for each of the combinations, the outline of the rectangle to be extracted from the image is specified in order from the combination having the smaller adjustment degree .
A method for controlling an image extracting apparatus, comprising: A program for extracting a rectangle from an image, comprising:
Extracting a line segment from the image;
After extracting two line segments having a common end point and forming an angle from each other among the line segments extracted from the image as a line segment pair, two line segment pairs are extracted from the line segment pairs. A procedure of adjusting the length of the four line segments to form a rectangle for each combination to be selected;
For each combination, a step of calculating a value representing the degree of adjustment of the length of the line segment required to form the rectangle;
A procedure for identifying the outline of a rectangle to be extracted from the image , in the order of the degree of the adjustment amount, from among the rectangles formed for each combination,
A program for executing