JP3579645B2 - Texture cutout area correction method and apparatus, and recording medium storing an execution program of this method - Google Patents

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a method for generating a three-dimensional shape model, and more particularly to a method and an apparatus for calculating and correcting a region where a texture to be mapped to a surface forming a three-dimensional shape is cut out from a target image.
[0002]
[Prior art]
2. Description of the Related Art In recent years, there has been an increasing demand for using a beautiful three-dimensional shape model with a high precision appearance for the purpose of simulating a landscape and using amusement. Since such a model is required to have a high-precision appearance and a high degree of realism, it is necessary to map a texture onto a surface constituting the model shape.
[0003]
In response to such demands, several techniques have been developed so far for mapping textures to surfaces constituting a model shape. For example, the system described in Japanese Patent Application Laid-Open No. H10-42282 or the device and system described in Japanese Patent Application Laid-Open No. H10-267671 stores images of various points in association with position information and attitude information of a camera that has taken the images. is there. Based on the position information and the posture information, it is possible to present where and what is captured in the captured image, and to use the captured image as a texture cutout target image.
[0004]
[Problems to be solved by the invention]
However, the position information and the posture information of the camera according to the related art described above have large errors depending on the shooting conditions, and it is difficult to accurately calculate what appears in the image and what looks like a three-dimensional model shape. It is difficult to effectively cut out and use a texture with high precision when the cleanliness and precision of the image are required. Therefore, in order to increase the accuracy, a sensor for acquiring such position information and posture information becomes large or expensive. For this reason, it is necessary to mount the device on a helicopter or a large wagon, and it is difficult to use the device in a situation where, for example, a person carries the device.
[0005]
As described above, it is difficult to map a texture onto a surface forming a three-dimensional shape with high accuracy in the conventional technique. For this reason, it is necessary to prepare a large-scale device for acquiring high-precision information, or to manually operate the device. Also, this operation requires considerable time and experience. Therefore, labor saving and further automation are strongly desired.
[0006]
The present invention has been made in view of the above-described problems and technical backgrounds of the related art, and has been made in consideration of a photographed image or the like formed on a surface forming a three-dimensional shape obtained by any three-dimensional shape measuring means. It is an object of the present invention to provide a texture cutout area correction method and apparatus for cutting out and mapping a texture from a target image with high accuracy.
[0007]
[Means for Solving the Problems]
The texture cutout area correction method according to the present invention is a method for calculating and correcting a texture cutout area from a texture cutout target image in order to map a texture onto a surface constituting a target object having a three-dimensional shape. A cutout area calculation processing procedure for calculating a cutout area from a shape obtained by projecting a processing surface onto a texture cutout target image, and an image feature quantity calculation for calculating an edge pair satisfying a critical pixel interval for each coordinate axis direction from the texture cutout target image It is characterized by comprising a processing procedure, and a cutout area correction processing procedure for correcting the cutout area using the calculated cutout area and the edge pair.
[0008]
Alternatively, in the above-described texture cutout area correction method, in the cutout area calculation processing procedure, a histogram is calculated for the direction of a line segment sequence forming a shape in which the surface of the target object is projected on the texture cutout target image, and the line segment is calculated. The rotation angle at which the angle between the column group and the coordinate axis is minimized is calculated.In the feature amount calculation processing procedure, the texture cutout target image is rotated from the coordinate axis at the minimum rotation angle, and in the cutout area correction processing procedure, The calculated cutout area is rotated from the coordinate axis by the minimum rotation angle.
[0009]
Alternatively, in the above-described texture cutout area correction method, in the image feature amount calculation processing procedure, an edge image in the coordinate axis direction is generated from the texture cutout target image, and the edge image is scanned in the coordinate axis direction to obtain a certain target edge pixel. It is characterized in that a pair of pixels in a case where the interval to the edge pixel in the opposite direction satisfies the critical pixel interval is calculated as an edge pair.
[0010]
Alternatively, in the above-described texture cutout area correction method, the cutout area correction processing procedure uses the cutout area and the edge pair to calculate a candidate for the amount of movement of the cutout area. A volume voting process procedure and a travel amount calculation procedure for determining the travel amount from a candidate for the calculated travel amount are provided, and the clipping region is moved and corrected by the determined travel amount. Features.
[0011]
Alternatively, in the above-described texture cutout area correction method, in the feature amount voting processing procedure, some top satisfies the critical vote rate calculated by the number of votes / the total number of votes are calculated as candidates for the movement amount. I do.
[0012]
Alternatively, in the above-described texture cutout area correction method, in the moving amount calculation processing procedure, moving amount candidates other than the determined moving amount of the object whose moving amount is already determined are deleted to form the same line segment sequence. Calculate a combination in which the difference in the movement amount of the line segment sequence is a certain value or less, and calculate a combination in which the difference in the movement amount is a certain value or less without inconsistency with the original positional relationship in the combination forming the different line segment sequence Then, a combination is calculated by the same processing for the line segment sequence that constitutes the adjacent object, and a combination that is not inconsistent with the original positional relationship in the combination is calculated. It is characterized in that a combination having the largest number of votes is calculated, and a movement amount is calculated from the combination of the maximum number of votes.
[0013]
Alternatively, in the above-described texture cutout area correction method, in the cutout area correction processing procedure, a point forming the cutout area by using a point forming a line segment sequence obtained by dividing a line segment forming the calculated cutout area. Is calculated, and the cut-out area is corrected based on the calculated movement amount.
[0014]
Further, the texture cutout region correcting apparatus according to the present invention is a device for calculating and correcting a region from which a texture is cut out from a texture cutout target image in order to map a texture onto a surface constituting a target object having a three-dimensional shape, A cutout region calculation processing means for calculating a cutout region from a shape obtained by projecting an unprocessed surface of the object onto the texture cutout target image, and an image for calculating an edge pair satisfying a critical pixel interval for each coordinate axis direction from the texture cutout target image It is characterized by comprising a feature amount calculation processing means, and a cutout area correction processing means for correcting the cutout area using the calculated cutout area and the edge pair.
[0015]
Alternatively, in the above-described texture cut-out region correction apparatus, the cut-out region correction processing unit calculates a candidate for a moving amount of how much the cut-out region should be moved using the cut-out region and the edge pair. An amount voting processing means, and a movement amount calculation processing means for determining the movement amount from a candidate for the calculated movement amount, wherein the cutout area is moved and corrected by the determined movement amount. There is a feature.
[0016]
Further, a program for causing a computer to execute the processing procedure in the above-described texture cutout area correction method is recorded on a computer-readable recording medium.
[0017]
In the present invention, when mapping a texture to a surface forming a three-dimensional shape, the shape of a cutout area calculated from a texture cutout target image such as shooting information is corrected using an edge pair extracted from the texture cutout target image. As a result, the cut-out region of the texture can be corrected with high accuracy, and a beautiful three-dimensional shape model can be generated. Here, the edge pair is, for example, a threshold (critical threshold) at which a pixel forming an edge changing from light to dark and a pixel forming an edge changing from dark to light on an image are recognized as forming a pair. (Referred to as pixel spacing). Such an edge pair always appears at the boundary of the shape of a normal captured image.
[0018]
Further, a histogram is calculated for the direction of a line segment sequence forming a shape obtained by projecting the surface of the target object on the texture cut-out target image, and the angle between the line segment sequence group and a coordinate axis such as the X axis or the Y axis is determined to be the minimum. By calculating the rotation angle, the calculation of the correction amount of the cut-out area such as the movement amount can be reduced in the coordinate axis direction, thereby realizing efficient and very high-speed processing.
[0019]
When an edge image (referred to as a directional edge image) in the coordinate axis direction of the texture extraction target image is scanned in the specified coordinate axis direction, and the interval between the target pixel and the edge pixel in the opposite direction satisfies the critical pixel interval. Is calculated as an edge pair, noise generated by image processing in estimating a correction amount such as a movement amount is reduced, and efficient and very high-speed processing is realized.
[0020]
In addition, by selecting some upper edge pairs satisfying the critical vote rate (threshold recognized as a candidate for the moving amount) calculated by the number of votes / the total number of votes as candidates for the moving amount, the moving amount is determined. In the above, efficient and very high-speed processing is realized.
[0021]
In addition, a moving amount candidate other than the determined moving amount of the object whose moving amount has already been determined is deleted, and a combination in which the difference in the moving amount of the line segment sequence forming the same line segment sequence is equal to or less than a certain value is calculated. Then, a combination having a contradiction with the original positional relationship and having a difference of a movement amount equal to or less than a certain value is calculated for the combinations constituting the different line segment sequences, and the same processing is performed on the line segment sequences constituting the adjacent object. By calculating the combination, calculating the combination that is not inconsistent with the original positional relationship in the combination, and calculating the combination with the largest number of votes among the combination that is not inconsistent, the cutout area after correction Inconsistency can be eliminated, efficient and very high-speed processing can be realized, and a beautiful three-dimensional shape model can be generated.
[0022]
Further, by using the points forming the line segment obtained by dividing the calculated line segment forming the cutout region, the moving amount of the point forming the cutout region is estimated, and the cutout region is corrected. The shape can be controlled flexibly. Further, information for estimating the movement amount is also increased, and efficient processing is realized.
[0023]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0024]
FIG. 1 shows an embodiment of an apparatus for realizing a texture cutout area correction method according to the present invention.
[0025]
The apparatus according to the present embodiment accepts processing from an operator and stores the processing result in a three-dimensional shape database 7 using an input / output processing unit 1 and a shape obtained by projecting an unprocessed surface of a target object onto a texture cut target image. A cut-out area calculation processing unit 2 for calculating a cut-out area, an image feature amount calculation processing unit 3 for calculating an edge pair satisfying a critical pixel interval for each axial direction from a texture cut-out target image, and a cut-out area calculated in the above procedure And a cutout area correction processing means 4 for correcting a cutout area using an edge pair.
[0026]
The cut-out area correction processing means 4 calculates a candidate for the amount of movement of the cut-out area by using the cut-out area and the edge pair calculated by the processing means 2 and 3. And a moving amount calculation processing unit 6 for determining a moving amount from candidates of the moving amount calculated by the processing unit 5.
[0027]
The three-dimensional shape database 7 accumulates data including data of the three-dimensional coordinate point sequence of the constituent points and the constituent surfaces of the three-dimensional model. In addition, the image database 8 stores image information (RGB values of each pixel) as a texture cutout target image, and position information (X, Y, Z) of a camera at which the frame was shot for each frame in the image. ) And photographing information in which the posture information (Roll, Pitch, Yaw) is associated.
[0028]
The input / output processing unit 1 receives a region or an object to be processed from an operator, calculates a texture cut-out region of the object to be processed and stores the corrected result in the three-dimensional shape database 7. This process is performed as follows in the procedure shown in FIG.
[Step 1] Information of an object included in the processing target range is acquired from the three-dimensional shape database 7 and stored in a buffer.
[Step 2] Information on other objects adjacent to the unprocessed object is calculated.
[Step 3] For the unprocessed surface of the target object, the image information and the photographing information of the texture cutout are acquired from the image database 8.
[Step 4] A cutout area calculation process is performed.
[Step 5] An image feature amount calculation process is performed.
[Step 6] A cutout area correction process is performed.
[Step 7] [Step 3] to [Step 6] are repeated until the processing is completed for all the target surfaces.
[Step 8] [Step 2] to [Step 7] are repeated until the processing is completed for all target objects.
[Step 9] The texture image and the cutout area are stored in the three-dimensional shape database.
[0029]
Since this [Step 3] is not particularly different from the conventional method, a detailed description thereof will be omitted here.
[0030]
The cut-out area calculation processing means 2 calculates a cut-out area from the shape of the unprocessed surface of the target object projected on the texture cut-out target image. This processing is performed as follows in the procedure shown in FIG.
[Step 1] The shape (Ts) of the unprocessed surface of the target object projected on the image of the texture cutout target is calculated.
[Step 2] A shape group (Rs (n)) on which a surface constituting another object adjacent to Ts is projected is calculated.
[Step 3] A histogram of the angles formed by the individual line segments forming the Ts with the X-axis and the Y-axis is calculated, and those having the highest appearance frequency are defined as the angles formed by the X-axis and the Y-axis.
[Step 4] A smaller angle between the line segment group and the X-axis or the Y-axis is defined as a rotation angle α.
[0031]
Since this [Step 1] is not particularly different from the conventional method, a detailed description thereof will be omitted here.
[0032]
The image feature value calculation processing unit 3 calculates an edge pair satisfying the critical pixel interval for each axial direction from the texture cutout target image. This processing is performed as follows in the procedure shown in FIG.
[Step 1] Image information (Ti) of an area obtained by adding a margin to an area including Ts and Rs (n) is obtained from image information to be cut out.
[Step 2] Ti is rotated by an angle α.
[Step 3] A direction edge image Te is calculated from Ti.
[Step 4] An edge pair that satisfies the critical pixel interval from Te in each of the X-axis direction and the Y-axis direction is calculated.
[0033]
This edge pair scans the directional edge image Te in the designated axial direction, and when the interval between a certain target pixel and the edge pixel in the opposite direction satisfies the critical pixel interval, the pair of pixels is regarded as an edge pair. Is calculated. The critical pixel interval may be determined in advance as a value for each image or each processing target area. Since the method of calculating the direction edge image in [Step 3] is not particularly different from the conventional method, a detailed description thereof is omitted here.
[0034]
The cutout area correction processing means 4 corrects the cutout area using the cutout area and the edge pair calculated in the above procedure. This process is performed as follows in the procedure shown in FIG.
[Step 1] The line segment sequence forming Ts, Rs (n) is rotated by an angle α, and the line segment sequence is classified into categories in the X-axis direction and the Y-axis direction.
[Step 2] The line segment sequence forming Ts, Rs (n) is divided into m, and this line segment sequence is defined as L (i, j, k). Here, i = Ts or Rs (n), j = X-axis direction or Y-axis direction, and k = division line segment number.
[Step 3] A feature voting process is performed on L (i, j, k) in the X-axis direction and the Y-axis direction.
[Step 4] An axial direction in which the moving amount of the moving amount candidate is large is calculated.
[Step 5] A movement amount calculation process of L (Ts, j, k) is performed with L (i, j, k) in the calculated direction.
[Step 6] The calculated movement amount is reflected in L (i, j, k).
[Step 7] The feature amount voting process is performed again with L (i, j, k) in the unprocessed direction, and the movement amount calculation process of L (Ts, j, k) is performed.
[Step 8] The movement amount of the points constituting Ts is determined.
[0035]
Note that the division number m may be determined in advance or may not be divided. When m is 3 or more, the line segment sequence between them does not include a point constituting Ts, and thus may be excluded from the processing target. It may be set as a divided line segment sequence or a point forming a line segment sequence.
[0036]
The feature amount voting processing means 5 calculates a candidate for the amount of movement of the cutout region by using the cutout region and the edge pair in the above procedure. This processing is performed as follows in the procedure shown in FIG.
[Step 1] L (i, j, k) in the specified direction from the direction edge image Te and all edge pairs in the direction opposite to that direction are acquired.
[Step 2] Obtain all points on the unprocessed line segment sequence from the three-dimensional model.
[Step 3] Scan in the specified reverse direction from all the obtained constituent points to calculate the distance to all the edge pairs.
[Step 4] Vote for the calculated frequency for each distance.
[Step 5] The top T moving amounts satisfying the critical vote rate (the number of votes / the total number of votes) are acquired and set as the moving amount candidates.
[Step 6] [Step 2] to [Step 5] are repeated until the processing is completed for all target line segment strings.
[0037]
When voting the frequency in [Step 4], if the distance is longer than a predetermined distance, the voting may not be performed and the processing efficiency may be increased. Further, the T-th place of the acquisition criterion and the critical vote rate may be determined in advance, or may be dynamically set by determining the minimum number of votes.
[0038]
The movement amount calculation processing means 6 determines the movement amount from the candidates of the movement amount calculated in the above procedure. This processing is performed as follows in the procedure shown in FIG.
[Step 1] A candidate for the moving amount is acquired, and candidates having a critical moving amount (absolute value) or more are deleted.
[Step 2] In the already determined object, the moving amount candidates other than the determined moving amount are deleted.
[Step 3] A combination of L (Ts, j, k) constituting the same line segment sequence and having a difference in movement amount equal to or less than a threshold (critical movement amount) D1 is calculated.
[Step 4] A combination is calculated that has a contradiction with the original positional relationship and has a difference in movement amount equal to or smaller than a threshold (critical movement amount) D2 in the combinations forming different line segment sequences.
[Step 5] With L (Rs (n), j, k), the processes of [Step 3] and [Step 4] are performed, and a similar combination is calculated.
[Step 6] Among these combinations, a combination that does not contradict the original positional relationship is calculated.
[Step 7] Among these consistent combinations, the combination with the largest number of votes is calculated.
[Step 8] The movement amount of each point constituting L (i, j, k) is calculated.
[0039]
The critical movement amount (absolute value) and the values of D1 and D2 may be determined in advance. Also, in this [Step 8], the movement amount is set as a point constituting a line segment sequence constituting L (i, j, k) instead of a point constituting L (i, j, k). It may be calculated.
[0040]
Hereinafter, the above-described processing procedure will be specifically described according to actual data.
[0041]
In the input / output processing, the processing is received from the operator, the information of the object included in the processing target range is acquired from the three-dimensional shape database, and stored in the buffer. Calculates information, obtains image information and shooting information of a texture extraction target for an unprocessed surface of the target object, and performs an extraction process on the unprocessed surface of the target object in a cutout region calculation process. The shape (Ts) projected above is calculated, and a shape group (Rs (n)) on which a surface constituting another object adjacent to Ts is projected is calculated. As a result, a processing result as shown in FIG. 8 is obtained. Is obtained.
[0042]
On the other hand, the histogram for the direction of the line segment sequence forming the shape Ts projected on the texture clipping target image in the clipping region calculation process is calculated, and the angle between this line segment sequence group and the X axis here is calculated. The rotation angle α is calculated assuming the minimum value, and image information (Ti) of a region obtained by adding a margin to the region including Ts and Rs (n) is obtained from the image information to be subjected to texture extraction in the image feature amount calculation process. Then, as a result of rotating Ti by an angle α, a processing result as shown in FIG. 9 is obtained. As shown in FIG. 9, in this state, the cutout region of the texture is shifted due to the accuracy of the imaging information including the position information and the posture information.
[0043]
The image information Ti is processed by an image feature amount calculation process to calculate a direction edge image Te. For example, FIG. 10 shows a direction edge image in the Y-axis direction. This Te is scanned in each of the X-axis direction and the Y-axis direction, and when the interval between a certain target pixel and an edge pixel in the opposite direction satisfies the critical pixel interval, the pixel pair is calculated as an edge pair. I do.
[0044]
Further, in the cutout area correction processing, the line segment sequence forming Ts, Rs (n) is rotated by an angle α, and the line segment sequence is classified into categories in the X-axis direction and the Y-axis direction, and Ts, Rs (n) Is divided into m, and this line segment sequence is defined as L (i, j, k). Here, it is assumed that this line segment sequence is as shown in FIG.
[0045]
A feature voting process is performed on the L (i, j, k) line segment sequence in the X-axis direction and the Y-axis direction. In the feature amount voting process, first, L (i, j, k) in the X-axis direction and all edge pairs in the Y-axis direction opposite to that direction are acquired. All points on the unprocessed line segment sequence are acquired, and scanning is performed from all the constituent points in the specified reverse direction to calculate distances to all edge pairs, and vote for the frequency for each distance. For example, in the case as shown in FIG. 12, one vote is cast for each of the distances of −21, −18, 10, 12, 24, 25, 38, and 43. From the number of votes obtained in this way, the top T movement amounts satisfying the critical vote rate (the number of votes / the total number of votes) are acquired and set as the movement amount candidates. These processes are performed on all the target line segment sequences. The same processing is performed for the Y-axis direction.
[0046]
The axial direction in which the moving amount of the moving amount candidate is large is calculated. Here, assuming that the moving amount in the Y-axis direction is large, L (Ts , J, k) is calculated. In the movement amount calculation processing, first, a movement amount candidate is acquired, and candidates having a critical movement amount (absolute value) or more are deleted. In the already determined objects, the moving amount candidates other than the determined moving amount are deleted, and the difference in the moving amount is a threshold (critical moving amount) in L (Ts, j, k) constituting the same line segment sequence. ) Calculate a combination of D1 or less, and calculate a combination in which a difference in movement amount is equal to or less than a threshold value (critical movement amount) D2 in the combination constituting a different line segment sequence, without inconsistency with the original positional relationship. The case where there is no inconsistency with the original positional relationship means that, for example, when the projected shape is Ts as shown in FIG. 11, a line composed of L (Ts, Y, 5) and L (Ts, Y, 6) This is a case where the position of the line segment line composed of the line segment and L (Ts, Y, 7) and L (Ts, Y, 8) does not interchange. Further, the same processing is performed on L (Rs (n), j, k) to calculate the same combination. Among these combinations, the combination having the largest number of votes is calculated in a combination that does not contradict the original positional relationship. The case where there is no inconsistency with the original positional relationship is, for example, in the case of Ts and Rs (2) as shown in FIG. 11, it is composed of L (Ts, Y, 5) and L (Ts, Y, 6). And the position of the line segment sequence composed of L (Rs (2), Y, 1), L (Rs (2), Y, 2), L (Rs (2), Y, 3) This is the case where the left and right are not interchanged. In this way, the movement amount for each point constituting L (i, j, k) is calculated.
[0047]
Further, in the cutout area correction processing, the movement amount calculated for L (i, j, k) is reflected, and the same processing is performed in the X-axis direction. In this way, the movement amount of the points constituting Ts is determined, and the corrected cutout area as shown in FIG. 13 is calculated.
[0048]
Finally, in the input / output processing, the result of calculating the texture cutout area of the processing target object is stored in the three-dimensional shape database.
[0049]
It is noted that some or all of the processing functions of each unit shown in FIG. 1 can be realized using a computer, or that the computer can execute the processing procedures shown in FIGS. 2 to 7. Needless to say, a program for realizing the processing function of each part in the computer or a program for causing the computer to execute the processing procedure is stored in a storage medium readable by the computer, for example, FD (floppy disk: registered trademark). ), MO, ROM, memory card, CD, DVD, removable disk, etc., and can be stored, provided, or distributed.
[0050]
As described above, the present invention has been specifically described based on the embodiment. However, it is needless to say that the present invention is not limited to the above embodiment and can be variously modified without departing from the gist thereof. Absent.
[0051]
【The invention's effect】
According to the present invention, when mapping a texture to a surface constituting a target object having a three-dimensional shape, the shape of a cutout area calculated with low accuracy from a texture cutout target image such as a shot image based on shooting information or the like is also used. Since the correction is performed by using the edge pair extracted from the target image of the texture extraction, the cut-out region of the calculated texture can be corrected with high accuracy, so that the object is cut out with high accuracy on the surface of the object. The texture can be mapped, and an effect of generating a beautiful three-dimensional shape model can be obtained.
[0052]
Further, a histogram is calculated for the direction of a line segment sequence forming a shape obtained by projecting the surface of the target object on the texture cut-out target image, and the angle between the line segment sequence group and a coordinate axis such as the X axis or the Y axis is determined to be the minimum. Since the rotation angle is calculated as follows, the calculation of the movement amount can be reduced in the coordinate axis direction, so that an effect of realizing efficient and very high-speed processing can be obtained.
[0053]
In addition, the direction edge image of the texture cutout target image is scanned in the designated coordinate axis direction, and a pair of pixels is calculated as an edge pair when an interval between a certain target pixel and an edge pixel in a direction opposite to the target pixel satisfies a critical pixel interval. Therefore, the noise generated by the image processing in estimating the movement amount can be reduced, and the effect of achieving efficient and very high-speed processing can be obtained.
[0054]
In addition, the top several satisfies the critical voting rate calculated by the number of votes / the total number of votes is selected as a candidate for the moving amount, so that in determining the moving amount, efficient and very high-speed processing is performed. The effect that can be realized is obtained.
[0055]
In addition, a moving amount candidate other than the determined moving amount of the object whose moving amount has already been determined is deleted, and a combination in which the difference in the moving amount of the line segment sequence forming the same line segment sequence is equal to or less than a certain value is calculated. Then, a combination having a contradiction with the original positional relationship and having a difference in the movement amount of a certain value or less is calculated for the combinations constituting the different line segment sequences, and the same processing is performed on the line segment sequences constituting the adjacent object. The combination is calculated in, the combination that does not contradict the original positional relationship in the combination is calculated, and the combination with the largest number of votes is calculated in the combination that does not have contradiction. The inconsistency of the cut-out areas can be eliminated, and an effect of realizing efficient and very high-speed processing and generating a beautiful three-dimensional shape model can be obtained.
[0056]
In addition, by using the points forming the line segments obtained by dividing the calculated line segments forming the cutout region, the amount of movement of the points forming the cutout region is estimated and the cutout region is corrected. An effect that the shape of the region can be flexibly controlled can be obtained. Further, since the information for estimating the movement amount also increases, an effect that an efficient process can be realized is obtained.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of an apparatus for implementing a texture cutout area correction method according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an input / output processing flow according to the embodiment.
FIG. 3 is a diagram for explaining a cutout region calculation processing flow according to the embodiment;
FIG. 4 is a diagram for explaining an image feature amount calculation processing flow according to the embodiment;
FIG. 5 is a diagram illustrating a cutout region correction processing flow according to the embodiment.
FIG. 6 is a diagram illustrating a feature amount voting process flow according to the embodiment;
FIG. 7 is a diagram for explaining a movement amount calculation processing flow according to the embodiment;
FIG. 8 is a diagram illustrating an example of a shape group Ts, Rs (n) in which a surface of a target object and a surface of an adjacent object are projected on a texture cutout target image in the embodiment.
FIG. 9 is a diagram illustrating an example of image information Ti of a region obtained by adding a margin to the group of projected shapes acquired from a texture extraction target image according to the embodiment.
FIG. 10 is a diagram illustrating an example of a direction edge image in the Y-axis direction according to the embodiment.
FIGS. 11A and 11B are diagrams illustrating an example of a line segment L (i, j, k) obtained by dividing the projected shape group according to an embodiment of the present invention.
FIG. 12 is a diagram illustrating an example of a feature amount voting process according to an embodiment of the present invention.
FIG. 13 is a diagram illustrating an example of a processing result of a texture cutout area correction process according to an embodiment of the present invention.
[Explanation of symbols]
1. Input / output processing means
2. Cutout area calculation processing means
3. Image feature amount calculation processing means
4. Cutout area correction processing means
5. Feature voting processing means
6. Moving amount calculation processing means
7 3D shape database
8. Image database

Claims (10)

A method for calculating and correcting a texture cutout area from a texture cutout target image in order to map a texture to a surface constituting a target object having a three-dimensional shape,
A cutout area calculation processing procedure for calculating a cutout area from a shape obtained by projecting an unprocessed surface of the target object on the texture cutout target image;
An image feature amount calculation processing procedure for calculating an edge pair satisfying a critical pixel interval for each coordinate axis direction from the texture extraction target image;
A cut-out area correction processing procedure for correcting the cut-out area using the calculated cut-out area and the edge pair. In the cutout region calculation processing procedure, a histogram is calculated for the direction of a line segment sequence forming a shape obtained by projecting the surface of the target object onto the texture cut target image, and the angle between the line segment sequence group and the coordinate axis is determined to be the minimum. Calculate the rotation angle
In the feature amount calculation processing procedure, the texture cutout target image is rotated at the minimum rotation angle,
2. The texture cutout area correction method according to claim 1, wherein in the cutout area correction processing procedure, the line segment sequence is rotated at the minimum rotation angle. In the image feature amount calculation processing procedure,
An edge image in the coordinate axis direction is generated from the texture extraction target image, and the edge image is scanned in the coordinate axis direction to determine a pair of pixels when the interval between a certain target edge pixel and the edge pixel in the opposite direction satisfies the critical pixel interval. 3. The texture cutout area correction method according to claim 1, wherein the calculation is performed as an edge pair. In the cutout area correction processing procedure,
A feature amount voting process procedure for calculating a candidate for a movement amount of how much the cut region should be moved using the cut region and the edge pair;
A moving amount calculation processing procedure for determining the moving amount from the candidate of the calculated moving amount,
4. The texture cutout area correction method according to claim 1, wherein the cutout area is moved and corrected by the determined movement amount. In the feature amount voting processing procedure,
5. The texture cutout area correction method according to claim 4, wherein the top several satisfies the critical vote rate calculated by the number of votes / total votes is calculated as a candidate for the movement amount. In the movement amount calculation processing procedure,
The moving amount candidates other than the determined moving amount of the object whose moving amount has already been determined are deleted, and a combination in which the difference in the moving amount of the line segment sequence forming the same line segment sequence is equal to or less than a certain value is calculated, A combination having a contradiction with the original positional relationship and having a difference in the movement amount equal to or less than a certain value is calculated for the combinations constituting the different line segment sequences, and the same processing is performed on the line segment sequences constituting the adjacent object by a similar process. Calculate the combination, calculate the combination that is not inconsistent with the original positional relationship in the combination, calculate the combination with the largest number of votes among the combination that is not inconsistent, and calculate the movement amount from the combination of the maximum number of votes. The texture cutout area correction method according to claim 4 or 5, wherein is calculated. In the cutout area correction processing procedure,
Using the points forming a line segment sequence obtained by dividing the calculated line segment forming the cutout region, calculating the movement amount of the points forming the cutout region, and correcting the cutout region from the movement amount. The texture cutout area correction method according to any one of claims 1 to 6, characterized in that: An apparatus for calculating and correcting a region from which a texture is extracted from a texture extraction target image in order to map a texture onto a surface constituting a target object having a three-dimensional shape,
A cutout region calculation processing unit that calculates a cutout region from a shape obtained by projecting an unprocessed surface of the target object onto the texture cutout target image;
Image feature amount calculation processing means for calculating an edge pair satisfying a critical pixel interval for each coordinate axis direction from the texture cutout target image;
A texture cutout area correction apparatus, comprising: cutout area correction processing means for correcting the cutout area using the calculated cutout area and the edge pair. The cutout area correction processing means,
Feature amount voting processing means for calculating a candidate for the amount of movement of the cutout region by using the cutout region and the edge pair;
A moving amount calculation processing means for determining the moving amount from the candidate of the calculated moving amount,
9. The texture cut-out area correction device according to claim 8, wherein the cut-out area is moved and corrected by the determined movement amount. 8. A texture cutout, wherein a program for causing a computer to execute a processing procedure in the texture cutout area correction method according to any one of claims 1 to 7 is recorded on a computer-readable recording medium. A recording medium on which an execution program for an area correction method is recorded.

Images