JP2009116464A - Outline information generation method, system, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To efficiently and surely extract the outline of an object existing in an image. <P>SOLUTION: A rubber band having fixed thickness is displayed between an input point and an input pointer, and a point is input so that an input segment can be inserted into the rubber band. Thus, it is possible for an operator to efficiently input the segment by saving any labor to precisely determine the position of an input pointer, and to make an input decision. When the input segment is displayed as it is, an outline whose accuracy is missing is generated. Then, it is possible to generate an outline matched with a ground object picked up in an image by performing processing to correct the input segment according to the image. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a contour information generation method and system, and a program, for example, a method for extracting a contour of a feature composed of line segments captured in an image from an image photographed from a high sky.

  In recent years, the demand for electronic maps has increased rapidly due to the development of electronic map services and car navigation systems on the Internet. However, an electronic map requires a great deal of man-hours to create and update.

  As a method for solving this problem, a technique for automatically extracting features that are constituents of a map from an image taken from a high sky such as an aerial photograph or a satellite image has been studied. For example, Patent Document 1 proposes a program for extracting the shape of a building by inputting one point in the building on the image.

  When a manual input by an operator is performed, it is widely used to display a rubber band as shown in Patent Document 2 as an input assisting means.

JP 2004-341422 A JP-A-7-168552

  However, a technique for automatically extracting the contour shape of a feature from an image as in Patent Document 1 has not reached a practical level at present. This is because if the aerial photograph has a different shooting situation or the state of the photographed feature, it is necessary to change the parameter to be set each time, and extraction may be difficult in the first place.

  Therefore, manual input by an operator is still common. In the contour input of features by the operator, the operator inputs points with the target of the contact point of the contour, that is, the part where the line segment intersects. Is required. That is, as shown in FIG. 1, when tracing the outline of the building 100 captured in the aerial photograph, the point 104 that is considered to be the intersection of the line segment 101 and the line segment 102 is first determined in order to determine the starting point of drawing. Need to recognize. In some cases, the intersection can be clearly recognized, but there are many cases where the portion where the line segment 101 and the line segment 102 intersect is not clear, and in this case, it takes time to determine the start point. After inputting the point 104, the rubber band 106 shown in Patent Document 2 follows the movement of the input pointer 107. The rubber band is aligned with the line segment 102, and the intersection of the end line segment 102 and the line segment 103 is detected. You must enter the point 105 that you think.

  As described above, there is a problem that efficiency is inevitably lowered in order to input correctly, and further, there is a problem that the accuracy of the figure varies depending on the operator.

  The present invention has been made in view of such a situation, and it is not necessary to input accurate coordinate information of the start point and end point of a line segment, and the line segment constituting the contour is extracted efficiently and accurately. The method which can be performed is provided.

  In order to achieve the above object, in the present invention, a rubber band having a certain thickness is displayed between an input point and an input pointer, and a point is input so that an input line segment is placed in the rubber band. . As a result, the operator can save time and effort to accurately determine the position of the input pointer and perform input determination, and input can be performed efficiently. However, if the input line segment is displayed as it is, an outline that is not accurate will be generated. Therefore, by performing a process of correcting the input line segment according to the image, a contour suitable for the object captured in the image is generated.

  That is, the contour information generation method according to the present invention is a contour information generation method for generating contour information of an object existing in an image using an information processing apparatus, and is a line composed of input start points and end points. Generating a plurality of line segments within a predetermined range including a minute as contour constituent line segment candidates, and contouring a line segment having the highest edge strength value among the plurality of line segments as contour constituent line segment candidates Extracting as a constituent line segment, and generating contour information of the object using the contour constituent line segment.

  In the generating step described above, a rubber band including a line segment composed of the start point and the end point corresponding to the input of the start point and the end point and having a certain width in the normal direction of the line segment is displayed on the display unit. A plurality of line segments within the rubber band range are set as contour configuration line segment candidates.

  In the above-described extracting step, for each of a plurality of line segments that are contour constituent line segment candidates, a gradient vector is calculated for pixels on the image through which the line segment passes, and the line segment for which the gradient vector is calculated is calculated. The edge intensity value in the normal direction of the line segment is calculated from the inner product of the normal vector and the gradient vector, and the line segment having the highest edge intensity value is determined as the contour constituent line segment.

  Further features of the present invention will become apparent from the best mode for carrying out the present invention and the accompanying drawings.

  According to the present invention, not only the contour can be efficiently extracted in the contour input operation of the feature from the image by the operator, but also the contour that matches the feature captured in the image can be accurately extracted.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be noted that this embodiment is merely an example for realizing the present invention, and does not limit the technical scope of the present invention. In each drawing, the same reference numerals are assigned to common components.

<Configuration of feature outline generation system>
FIG. 2 is a diagram showing a schematic configuration of the feature contour generation system according to the embodiment of the present invention. The feature outline generation system 20 includes an information processing apparatus 100 that generates outline information, and a monitor 206 that displays line information of the outline information generation process and generated outline information.

  The information processing apparatus 200 includes a storage device 201 in which image data 202 and pixel value vector data 203 of each pixel are stored. The information processing apparatus 200 superimposes the image data 202 and the pixel value vector data 203 so that the feature information can be displayed on the monitor 206.

  The information processing apparatus 200 further generates an input auxiliary line display unit (including a contour line segment candidate generation unit) 204 that generates a line group corresponding to FIG. 5 and displays it on the monitor 206, and an appropriate line segment from the line group. Input line segment correcting means (including contour component line segment extracting means) 205 for selecting.

  The operator 208 traces the outline of the feature imaged on the image data on the monitor 206 by using an input unit such as the mouse 207. At this time, the input auxiliary line display means 204 displays a rubber band having a certain thickness enough to include a plurality of line segments. When a line segment is input, the line input by the input line segment correction unit 205 is corrected.

<Example of input screen>
FIG. 3 is a diagram illustrating an example of an input screen when a line segment is drawn between the start point and the end point using the input auxiliary line display unit 204. As shown in FIG. 3, the start point and end point of the line segment to be drawn need only be included within the range covered by the rubber band 306 having a certain thickness. Therefore, even if the start point and the end point determined by the operator 208 are different from the viewpoint and end point of the actual line segment constituting the contour of the feature, the contour line segment is determined as long as it is within the range covered by the rubber band 306. There is no problem in confirming.

  For example, when drawing a line segment 302 including the start point 304 and the end point 305, the input pointer 307 is moved and clicked so that the start point 304 is included in the initial rubber band 308, and then the end point 305 is included in the rubber band 306. The input pointer 307 is moved and clicked. At that time, the line segment 302 to be drawn should be included in the rubber band 306. In this way, the rubber band 306 may include the line segment to be drawn without trying to accurately match the input pointer to the point to be hit.

<Contour figure confirmation processing>
FIG. 4 is a flowchart for explaining a process of determining a figure based on a line segment determined by an input start point and end point.

  First, in step S400, based on the input line segment information, the input auxiliary line display unit 204 lists all line segments that are line segment candidates. More specifically, processing as shown in FIG. 5 is executed, and all candidate line segments are obtained. That is, with respect to the input line segment 500, line segments 504 and 505 perpendicular to the line segment 500 are drawn from the end points 501 and 502 constituting the line segment 500 until they intersect the end of the rubber band 503. The line segment is divided equally to create a plurality of points 506 and 507 on the line segment 504 and the line segment 505. The division method is such that the interval is about the size of the pixel of the image. A combination of all line segments connecting the point group 506 and the point group 507 is obtained, and these are set as candidate line segment groups 508. Therefore, for example, if there are 10 division points (pixels) on the line segments 505 and 506, the number of line segments is 10 × 10 = 100.

  Next, in step S401, the input line segment correcting unit 205 calculates the edge intensity value on the image for each line segment candidate obtained in step S400, and compares the edge intensity values to obtain the largest value. A line segment having a value is determined as a corrected line segment. More specifically, processing as shown in FIG. 6 is executed. That is, as shown in FIG. 6, first, in step S600, all the pixels 605 on the image through which each candidate line segment 604 passes are obtained. In step S601, a gradient vector (vi) 606 of each pixel is obtained using a Sobel filter for each pixel. The configuration of the Sobel filter is shown in FIG. A gradient vector is generated by multiplying each pixel elementary value vector around the pixel of interest read from the vector data storage unit 203 by the filter coefficient shown in FIG. If there is a strong edge with respect to the straight line 604, the gradient vector (vi) 606 is a long vector that faces the normal direction of the straight line 604. In step S602, the unit normal vector of the straight line 604 is (n) 607, and the inner product of the gradient vector (vi) 606 and the normal (n) 607 is obtained. This inner product is obtained for all the pixels 605. Then, the total value of the inner products is set as the edge strength value (S) 608, and the edge strength value (S) is obtained for all line segment candidates. In step S603, the line segment having the largest edge strength value (S) is set as a corrected line segment (step 603).

  Finally, in step S402, the information processing apparatus 200 determines the figure by using the corrected intersection of the line segments to obtain the contact point of the figure. In step S402, the trimming process may be executed together. In other words, if at least one of the line segments is too short to be sufficient for the intersection, or if at least one of the line segments is too long and the line segment protrudes from the intersection, the intersection is the start point (end point) of the line segment The trimming process is executed so that

<Summary>
In the above embodiment, the example of extracting the contour of the target feature composed of line segments from the image taken from the high sky has been described. However, the present invention is not limited to this, and the contour of the target existing in the image is extracted. It can be widely applied.

  In the present embodiment, a plurality of line segments within a predetermined range including a line segment constituted by a start point and an end point input by an operator are generated as contour constituent line segment candidates. Also, a line segment having the highest edge strength value is extracted as a contour constituent line segment from among a plurality of line segments as contour constituent line segment candidates. And the outline information of a target object is generated using a plurality of outline composition line segments obtained by repeating these processes. By doing so, it is not necessary for the operator to accurately input the start point and end point of the line segment constituting the contour, and the contour determination operation of the object becomes very efficient.

  Further, within the above-mentioned predetermined range, a rubber band including a line segment constituted by a start point and an end point and having a certain width (thickness) in the normal direction of the line segment is displayed on the display unit, and the rubber is displayed. A plurality of line segments within the band range are set as contour configuration line segment candidates. By doing in this way, the operator can properly configure the line segment that forms the outline of the object as long as the start point and end point of the line segment are included in the rubber band. Therefore, it is very efficient and less stress is felt in the contour determination work.

  Further, for each of a plurality of line segments that are contour constituent line segment candidates, a gradient vector is calculated for pixels on the image through which the line segment passes, and the normal vector and gradient vector of the line segment for which the gradient vector is calculated, The edge strength value in the normal direction of the line segment is calculated from the inner product of, and the line segment having the highest edge intensity value is determined as the contour constituent line segment. By doing in this way, the line segment which comprises an outline can be decided reliably from candidates.

  The present invention can also be realized by a program code of software that implements the functions of the embodiments. In this case, a storage medium in which the program code is recorded is provided to the system or apparatus, and the computer (or CPU or MPU) of the system or apparatus reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the program code itself and the storage medium storing the program code constitute the present invention. As a storage medium for supplying such a program code, for example, a floppy (registered trademark) disk, CD-ROM, DVD-ROM, hard disk, optical disk, magneto-optical disk, CD-R, magnetic tape, non-volatile A memory card, ROM, or the like is used.

  Also, based on the instruction of the program code, an OS (operating system) running on the computer performs part or all of the actual processing, and the functions of the above-described embodiments are realized by the processing. May be. Further, after the program code read from the storage medium is written in the memory on the computer, the computer CPU or the like performs part or all of the actual processing based on the instruction of the program code. Thus, the functions of the above-described embodiments may be realized.

  In addition, the program code of the software that realizes the functions of the embodiment is distributed via a network, and the code is stored in a storage means such as a hard disk or memory of a system or apparatus or a storage medium such as a CD-RW or CD-R. A stored computer and a computer (or CPU or MPU) of the system or apparatus may read the program code stored in the storage means or the storage medium and execute the function.

It is an example of an input screen using a general rubber band function. It is a figure which shows the schematic structural example of the feature outline production | generation system by embodiment of this invention. It is an example of an input screen using the input auxiliary line display means of this invention. It is the figure which showed the flow which fixes a figure based on the input line segment. It is the figure which showed the method of calculating | requiring all the candidate line segments. It is the figure which showed the method of determining the line segment after correction | amendment from a candidate line segment group. It is the figure which showed the Sobel filter.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... The feature currently imaged on the image 101 ... The line segment which comprises a feature 102 ... The line segment which comprises a feature 103 ... The line segment which comprises a feature 104 ... The starting point at the time of drawing a line segment 105 ... Line End point 106 when drawing a minute 106 ... General rubber band 107 ... Input pointer 200 ... Information processing device 201 ... Storage device 202 ... Image data 203 ... Vector data 204 ... Input auxiliary line display means 205 ... Input line segment correction means 206 ... monitor 207 ... mouse 208 ... operator 300 ... features picked up in image 301 ... line segment constituting feature 302 ... line segment constituting feature 303 ... line segment constituting feature 304 ... line segment Starting point when drawing 305 ... End point when drawing a line segment 306 ... Rubber band having a certain thickness 307 ... Input pointer 308 ... First point Rubber band for input 400 ... Step for listing all line segment candidates 401 ... Steps for calculating and comparing edge strength values of each line segment, and making a line segment with the highest edge intensity value a corrected line segment 402 Step of obtaining the contact point of the figure as the intersection of the corrected line segment 500 ... Input line segment 501 ... Input line segment start point 502 ... Input line segment end point 503 ... Rubber band having a certain thickness 504 ... Input line A line segment that forms a right angle with the input line segment through the start point of the minute 505 ... A line segment that forms a right angle with the input line segment through the end point of the input line segment 506 ... A point group obtained by dividing the line segment 504 at equal intervals 507 ... Point group obtained by dividing line segment 505 at equal intervals 508 ... Candidate line group group 600 ... Step for obtaining pixels on image through which each line segment passes 601 ... Using Sobel filter Step 602 for obtaining a pixel gradient vector Step 602 for obtaining an edge intensity value obtained by summing the inner product of the unit normal vector of the straight line and the gradient vector 603. Candidate line segment 605 ... Pixel on image through which line segment passes 606 ... Gradient vector 607 ... Linear unit normal vector 608 ... Calculation formula of edge intensity value

Claims (7)

A contour information generation method for generating contour information of an object existing in an image using an information processing device,
Generating a plurality of line segments within a predetermined range including a line segment constituted by the input start point and end point as contour component line segment candidates;
Extracting the line segment having the highest edge strength value as the contour line segment from among the plurality of line segments as the contour line segment candidates;
A contour information generation method comprising: generating contour information of the object using the contour constituent line segment.   In the generating step, a rubber band including a line segment composed of the start point and the end point corresponding to the input of the start point and the end point and having a certain width in the normal direction of the line segment is displayed on the display unit. The contour information generation method according to claim 1, wherein a plurality of line segments within the rubber band range are set as the contour constituent line segment candidates.   In the extracting step, for each of the plurality of line segments that are the contour line segment candidates, a gradient vector is calculated for pixels on an image through which the line segment passes, and the line segment for which the gradient vector is calculated is calculated. An edge strength value in a normal direction of a line segment is calculated from an inner product of a normal vector and the gradient vector, and a line segment having the highest edge intensity value is determined as the contour constituent line segment. Item 3. The contour information generation method according to item 1 or 2. An outline information generation system for generating outline information of an object existing in an image,
Contour component line segment candidate generating means for generating, as contour component line segment candidates, a plurality of line segments within a predetermined range including a line segment composed of the input start point and end point;
Out of the plurality of line segments as the contour line segment candidates, a contour line segment extracting unit that extracts a line segment having the highest edge strength value as a contour line segment;
A contour information generation system comprising: generating contour information of the object using the contour constituent line segment.   The contour constituent line segment candidate generating means includes a line segment composed of the start point and the end point corresponding to the input of the start point and the end point, and a rubber band having a certain width in the normal direction of the line segment. 5. The contour information generation system according to claim 4, wherein a plurality of line segments displayed on a display unit and within a rubber band range are set as the contour constituent line segment candidates.   The contour component line segment extraction unit calculates a gradient vector for each pixel on the image through which the line segment passes for each of the plurality of line segments that are the contour component line segment candidates, and calculates the gradient vector. An edge intensity value in the normal direction of the line segment is calculated from an inner product of a normal vector of the line segment and the gradient vector, and a line segment having the highest edge intensity value is determined as the contour constituent line segment. The contour information generation system according to claim 4 or 5.   A program for causing a computer to function as the contour information generation system according to any one of claims 4 to 6.

Images