JPH09138850A - Surface shape reconstitution device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To precisely reconstitute the shape and the structure of an object surface even if an object has the shape constituted by a smoothly curved surface by processing an outline part by means of synthesizing feature points obtained from multiple directions and detecting a density difference. SOLUTION: A picture taken by a TV camera 1-1 is stored in a picture memory 1-3. A feature point detection means 1-4 detects the point having the density difference in the picture as the feature point of the picture. A rotation control means 1-6 detects and controls the rotational angle of a rotation stage 1-5 rotating around a subject. A focal position control means 1-2 controls the focal position of the camera and a distance map generation means 1-7 generates a map shown by using information on the position where the focal point matches the feature point of the subject. A feature point synthesis means 1-8 synthesizes the feature points obtained from the picture and stores the reconstituted picture in a three-dimensional memory 1-9. A display means 1-10 displays and outputs the stored picture.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a three-dimensional shape input device that handles a three-dimensional structure of an object, and more particularly to a surface shape reconstructing device that reconstructs the surface structure.

[0002]

2. Description of the Related Art Conventionally, as a method for three-dimensionally inputting the surface structure of a subject, there has been a method of creating a "distance map" by using focus position information. An example thereof is the "focus method". The above-mentioned "distance map" is a map in which a distance to a subject corresponding to each pixel is associated with each pixel when a certain subject is photographed to obtain an image.

In order to facilitate the explanation, assume that the camera shoots a cylinder of a subject as shown in FIG. 15 (a) as a shooting situation, and a state in which the camera and the subject are viewed from above. 15 (b) and 15 (c)
Will be described sequentially.

The "focusing method" is a distance measuring method that has already been put to practical use in video cameras, silver halide cameras, etc., and is a feature point that causes contrast in an image as shown in FIG. 15 (b). In this case, the focus position of the optical image forming system when the contrast value becomes maximum is read out, and a map is created which sets the distance from the camera to the feature point.

Another method is to use "stereo matching". This is a feature point obtained by photographing the same part of the subject with respect to images observed from different viewpoints, as shown in FIG. 15C, when the image has feature points that generate contrast of light and dark. This is a method of clarifying the relative positional relationship between the camera and the feature point of the subject by applying the same principle as “triangulation” by associating with each other.

[0006]

However, in the conventional method, there is no description about the case of reconstructing the surface structure in the following situations. For example, if the subject is FIG.
This is the case when it is composed of a smooth curved surface with few features, as shown in.

When the "focus method" is applied to an object having such a curved surface, the characteristic points at which contrast occurs are limited to the contour portion of the object, and therefore measurable characteristic points are shown in FIG. 15 (b). It becomes slightly like this. In such a case, there has been no method of reconstructing a smooth curve using a method such as "interpolation" by a method using a conventional focus method. Although the focus method has been described as an example here, in general, when the distance is obtained by using the information of the in-focus position, a characteristic point that causes contrast as in the focus method is required.

Also in the case of "stereo matching", since there are few measurable feature points, a smooth curved surface is reconstructed by the conventional stereo matching method as in the focus method. There was nothing.

In addition, in stereo matching, FIG.
As shown in (c), when an attempt is made to reconstruct an outline part as a feature point, different positions on the subject are actually associated with the same position, and the same principle as "triangulation" is applied. become. In order to perform correct distance measurement in such a situation, some kind of ingenuity is required.
None of the matching explained this point. The above is the present situation to be solved by the main conventional techniques.

Another problem of the prior art that must be solved is, for example, when the surface structure is reconstructed, whether or not the characteristics of the obtained structure are isotropic is an important problem. The term “isotropic characteristic” as used herein means that the characteristic is the same in all different directions. When the surface structure is reconstructed using the focus method, the characteristics in the optical axis direction may in principle be inferior to those in other directions due to the spatial frequency characteristics of the optical imaging system. I understand. However, none of the conventional methods using the focus method have explained this point.

Therefore, an object of the present invention is to be able to handle an object whose surface structure is composed of a smooth curved surface.
It is an object of the present invention to provide a surface shape reconstructing device which is related to a three-dimensional shape input device and which can more correctly reconstruct the three-dimensional surface structure.

[0012]

The present invention adopts the measures described in the claims for the above-mentioned main problems of the prior art to be solved, and the measures described in the following supplementary notes. We are trying to solve by. That is, [1] a photographing means provided with a mechanism for controlling the focus position, a means for moving the relative position of the photographing means with respect to the subject, a memory for storing an image, and an image generated in the photographed image of the subject. The means for detecting a point having a density difference as a feature point, the means for obtaining depth information using the in-focus position information for the feature point of the subject to create a distance map and the photographing position of the photographing means A means for obtaining a relative positional relationship between the photographing means and the subject according to the change,
There is provided a surface shape reconstructing device configured by: a photographing unit, a unit for synthesizing feature points obtained from a plurality of images by using information on a relative positional relationship between a subject and a distance map. (The embodiment of the present invention corresponding to this [1] will be described in detail later in the first embodiment).

(Operation 1) An image can be input while scanning the focal position by a camera as a photographing means having a mechanism for controlling the focal position. It is possible to input images from multiple directions by means for moving the relative position of the camera with respect to the subject and means for obtaining the relative positional relationship between the camera and the subject as the shooting position of the camera is changed. The focus position is scanned by means for detecting a point having a density difference generated in the image of the subject as a feature point and means for obtaining depth information by the focus method for the feature point of the subject and creating a distance map. It is possible to create a distance map by obtaining the distances of the feature points where the density difference occurs in the obtained image. The feature points can be synthesized from the distance maps obtained in the multiple directions by the means for synthesizing the feature points obtained from the plurality of images according to the positional relationship between the camera and the subject and the distance map.

Even if there are few feature points obtained in one direction, by combining feature points obtained from multiple directions,
A more correct surface structure can be obtained. Especially when the surface structure is composed of a smooth curved surface, the feature points of the contour portion obtained by the density difference detection move according to the image input direction, so the feature points obtained in different image input directions are positional. And the resulting surface structure of the synthesis is more correct.

[2] The surface shape reconstructing apparatus according to [1], wherein the moving means for moving the relative position of the photographing means with respect to the subject is a rotary stage that rotates around the subject. (The embodiment of the present invention corresponding to this [2] will be described in detail later in the first embodiment).

(Operation 2) The rotating stage, which is the moving means, rotates with the photographing means mounted thereon to change the relative position with respect to the subject. Therefore, it is possible to provide a surface shape reconstructing apparatus which has a constant radius of gyration, that is, a relative distance to the object, and which is capable of easily inputting data regarding a three-dimensional shape by photographing multi-directional surfaces of different objects.

[3] A means for synthesizing feature points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and a distance map is a feature point obtained from a plurality of images according to the positional relationship between the photographing means and the subject and a distance map. The surface shape reconstructing apparatus according to [1], which is a linear feature point synthesizing unit that arranges and synthesizes as points, is provided. (The embodiment of the present invention corresponding to this [3] will be described in detail in the first embodiment described later).

(Operation 3) When the feature point synthesizing means synthesizes the feature points obtained from a plurality of images based on the positional relationship between the photographing means and the subject and the distance map, the feature points are first arranged as "points" on the map. To do. Therefore, it is possible to provide a surface shape reconstructing device that can reconstruct a correct surface shape by combining these feature points.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION A surface shape reconstructing device relating to a three-dimensional shape input device of the present invention will be described in detail below by taking a plurality of embodiments as examples. [First Embodiment] A first embodiment of the surface shape reconstructing apparatus of the present invention will be described with reference to FIGS. FIG.
FIG. 2 shows a block diagram of the apparatus as the first embodiment.
Shows the spatial relationship between the subject, the rotary stage, and the TV camera.

(Structure 1) The surface shape reconstructing device as the first embodiment is structured as follows. That is, FIG.
, The TV camera 1-1, the focus position control means 1-2 that controls the focus position of the camera, the image memory 1-3 that stores the captured image, and the point where there is a density difference in the image. Feature point detecting means 1-4 for detecting the feature points of the image,
A rotation stage 1-5 that rotates around the subject, and rotation control means 1-6 that detects and controls the rotation angle of the stage,
A distance map generating means 1-7 for generating a map represented by using information on a position where the feature point of the subject is in focus, an image memory 1-11 for storing the obtained distance map, and an image memory 1-11. Feature point synthesizing means 1-8 for synthesizing feature points,
It is composed of a three-dimensional image memory 1-9 for storing a reconstructed image and a display device 1-10 for displaying and outputting the stored image.

Further, as shown in FIG. 2, there is a rotary stage 1-5 which rotates around an object to be photographed, for example, and has a circular shape. The rotary stage 1-5 has a TV camera 1-1 as photographing means. Is fixed at a certain distance from the subject. Further, the rotation stage 1-5 is connected with rotation control means 1-6 for controlling and detecting the rotation angle. The rotation control means 1-6 is connected to the focus position control device 1-2 in order to output a signal for notifying that the detection of the rotation angle is completed. Further, the TV camera 1-1 has an image memory 1-3 for storing captured images, and a TV
A focus position control means 1-2 for controlling the focus position of the camera 1-1 is connected.

The focus position control means 1-2 is connected to the image memory 1-3 for storing the focus position value in association with the photographed image. The feature point detecting means 1-4 is connected to the image memory 1-3, and is connected to the distance map generating means 1-7 for outputting the result. Distance map generating means 1
-7 is connected to the image memory 1-3 and the feature point detecting means 1-4, and is connected to the image memory 1-11 for outputting the result. The feature point composition means 1-8 is the rotation control means 1-6.
Is connected to the image memory 1-11 in which the distance map is stored, and is connected to the three-dimensional image memory 1-9 to output the combined result. Also, this feature point synthesis means 1-8
Is connected to the rotation control means 1-6 to output a signal indicating that the synthesis is completed. The three-dimensional image memory 1-9 is connected to the display means 1-10 for displaying and outputting the reconstructed image obtained by the above-mentioned combining process.

(Operation and Effect 1) Next, the operation and effect of the first embodiment of the present invention will be described. FIG. 3 (a)
Each of the flowcharts (1) to (c) represents a processing procedure according to the first embodiment, FIG. 3 (a) is a main routine showing the processing procedure of the apparatus, and FIGS. 3 (b) and 3 (c) are subroutines. The procedure of "image input" processing and "distance map generation" processing is shown.

4 (a) to 4 (e) illustrate the method of reconstructing the apparatus of this embodiment, and FIG. 4 (a) shows the characteristics of the density profile function and the contrast in the image. FIG. 4B shows an image memory in which the image obtained by scanning the focus position is stored in association with each focus position, and FIG. The function representing the one-dimensional profile of concentration is shown in FIG. 4 (d).
In the table, the focal position and the distance α are shown in correspondence with each other.
A profile of the distance α is represented in (e).

Further, FIG. 5 (a) conceptually shows a memory for a three-dimensional image and shows a diagram in which each element of the memory is allocated so as to correspond to a three-dimensional space, and FIG. 5 (b) shows a memory for a three-dimensional image. The relationship between the memory and the distance map is conceptually shown in FIG.
The distance (that is, the radius d) between the feature point in the three-dimensional image memory and the distance map is conceptually represented.

First, the feature point synthesizing means 1-8 initializes a three-dimensional image memory 1-9 for storing a three-dimensional image. The three-dimensional memory 1-9 is a memory means for storing a three-dimensional image in which elements of the memory are allocated so as to correspond to a three-dimensional space as shown in FIG. 5 (a). By the initialization, the three-dimensional memory 1-9 becomes a state indicating that there is no object. For example, if there is no object here, it is represented by "0" (null or zero), and the value "0" is input to all the pixels of the 3D memory 1-9 after the initialization is completed. It (The above corresponds to step S1-1 in the flowchart of FIG. 3A).

The rotation control means 1-6 detects the rotation angle of the rotation stage 1-5 and sends the result to the feature point composition means 1-8.
Further, the rotation control means 1-6 outputs a signal indicating that the detection of the rotation angle is completed to the focus position control means 1-2. In the case of the present embodiment, a rotating stage is used to move the TV camera, and the TV camera is kept at a distance of radius d with respect to the center of rotation as shown in the schematic view of FIG. In order to obtain the change in the position due to, it is sufficient to detect only the rotation angle. (The above operation corresponds to step S1-2 in the flowchart of FIG. 3A).

Step S in the flowchart of FIG.
The subroutine "image input" processing called in 1-3 is performed in steps S1-9 to S1-13 in the flowchart of FIG.
In detail. That is, the focus position control means 1-2 which receives the signal from the rotation control means 1-6 initializes the focus position and sets the focus position at the start position of the focus position scanning. (The above corresponds to step S1-9).

The TV camera 1-1 photographs a subject and sends the obtained image to the image memory 1-3. The focus position control means 1-2 sends the focus position to the image memory 1-3. Image memory 1
In -3, the obtained image and the focus position are associated and stored.
(The above corresponds to steps S1-10 and S1-11).

Then, the focus position control means 1-2 moves the focus position. (The above corresponds to step S1-12). The focus position control means 1-2 checks whether or not the focus position scanning is completed. (The above corresponds to step S1-13). If the scanning is not completed yet, the process returns to step S1-10, and steps S1-10 to S1.
By repeating steps 1 to 13, the image memory 1-3 is displayed in FIG.
An image obtained by scanning the focal position as shown in (b) is stored in association with each focal position.

On the other hand, when the scanning of the focal position is completed, this image input routine is completed and the process returns. The characteristic point detecting means 1-4 detects all points where contrast occurs in the image as characteristic points as shown in FIG.
The coordinates in the image are sent to the distance map generating means. FIG.
As shown in (a), letting L (x) be a function that represents a one-dimensional profile of concentration, the feature points are first-order differential dL (x) / dx
It can be detected by searching for a point where is not zero. Also,
When the second derivative d ² L (x) / dx ² is used, the feature point can be detected by searching for a point where the value becomes zero between the peaks whose signs are reversed.

Here, in order to simplify the description, 1
Although the description has been made for the three-dimensional profile, in the case of a two-dimensional image, the processing of the one-dimensional profile described above is performed.
It may be performed in two orthogonal directions. (The above is Fig. 3
This corresponds to step S1-4 in the flowchart of (a)).

Step S in the flowchart of FIG.
Regarding the "distance map generation" processing subroutine shown in 1-5, step S1 of the flowchart of FIG.
-14 to S1-16 are shown in detail. That is,
The distance map generating means 1-7 initializes the image memory 1-11 for storing a so-called "distance map". By this initialization, the image memory 1-11 is in the initial state indicating that the distance has not been obtained yet. For example, if the state in which the distance is not obtained is represented by "0", the value "0" is input to all the pixels of the image memory 1-11 after the initialization is completed. (The above is step S1
-14 equivalent).

The distance map generating means 1-7 is for the small area around the coordinates of the feature points obtained from the feature point detecting means 1-4 in the image shown in FIG. 4B stored in the image memory. And apply the “focus method”. This focus method is applied in the following procedure. Specifically, if the coordinates of the feature points obtained from the feature point detecting means 1-4 are x1 as shown in FIG. 4 (c), when the contrast is expressed by the first derivative of Lfn (x), the contrast at x1 Find the focus position that maximizes. That is, the focus method is performed by obtaining fn that gives the following equation.

[0035]

(Equation 1)

Assuming that the focus position at which the contrast is maximized is fn, the distance at the coordinate x1 can be obtained as α from the relationship between the focus position and the distance shown in FIG. 4 (d). The relationship between the focal position and the distance shown in FIG. 4 (d) is specific to the optical image forming system of the TV camera used and is stored in advance in the distance map generating means 1-7. The distance map generating means 1-7 stores the obtained distance in the distance map as shown in FIG. (The above corresponds to step S1-15 in the flowchart of FIG. 3C).

The distance map generating means 1-7 checks whether or not the processing has been completed for all the characteristic points, and if not completed, the process returns to step S1-15, while if completed, the "distance map" is calculated. The "map generation" processing routine is terminated and the process returns.

The feature point synthesizing means 1-8 is the image memory 1-11.
By using the distance map stored in and the rotation angle obtained by the rotation control means 1-6, a three-dimensional image with a value other than "0" representing a state in which there is no object, for example, "1" as a feature point. Plot it on memory 1-9. It should be noted that the plotting of the characteristic points is performed in the geometrical relationship shown in FIG. That is, a distance map is assumed at a distance of radius d with the rotation angle obtained by the rotation control means 1-6 with respect to the center of the combined area, and for a three-dimensional image according to the value of the distance map for each feature point from the assumed position. Backproject on memory 1-9. However, FIG.
For simplification of description, (b) shows only a two-dimensional plane parallel to the rotation plane of the rotation stage.

The geometrical relationship of the three-dimensional "back projection" is shown in FIG.
The two-dimensional relationship shown in (b) may be stacked. As an example, FIG. 5C shows a state in which back projection in one direction is performed.
An example is shown below. The characteristic point synthesizing means 1-8 outputs a signal indicating that the synthesizing of the characteristic points is completed to the rotation control means 1-6.
(The above is step S1-6 in the flowchart of FIG.
Equivalent).

The rotation control means 1-6 performs a rotation operation on the rotary stage to move the rotary stage, and renews the relative positional relationship between the subject and the TV camera fixed to the rotary stage. (The above corresponds to step S1-7 in the flowchart of FIG. 3A).

The rotation control means 1-6 checks whether or not the rotation stage repeats a predetermined rotation, and if it has not finished yet, returns to step S1-2, and if the predetermined rotation has finished, Terminates all processing. (The above corresponds to step S1-8 of the flowchart).

As a result, since the three-dimensional image structure of the object is obtained in the three-dimensional image memory at this point, an appropriate viewpoint is set for the three-dimensional structure to make the two-dimensional image. By converting the image into an image, the display means 1-10 can be used to correctly display and output even a smooth curved surface shape.

(Modification 1) The photographing means provided with a mechanism for controlling the focus position corresponds to the TV camera and the focus position control means in this embodiment, but an electronic still camera is used instead of this TV camera. And so on. Further, the moving means for moving the relative position of the photographing means with respect to the subject corresponds to the rotary stage, but means for performing the same action may be used.

Further, the memory means for storing the image corresponds to the image memory and the memory for the three-dimensional image in the present embodiment, but it may be a storage or recording means having an equivalent operation. Further, the detection means for detecting a point having a density difference generated in an image of a subject as a feature point corresponds to the feature point detection means in the present embodiment, but a feature point other than the density difference is used. May be.

Similarly, the map creating means for obtaining the depth information by the focus method for the feature points of the object and creating the distance map corresponds to the distance map generating means in the present embodiment, and the photographing position of the photographing means is changed. According to the above, the means for obtaining the relative positional relationship between the photographing means and the subject corresponds to the rotation control means in this embodiment, and further. The means for synthesizing the feature points obtained from a plurality of images in accordance with the positional relationship between the photographing means and the subject and the distance map corresponds to the feature point synthesizing means in the present embodiment, but is not necessarily limited thereto and equivalent What is necessary is that the action and effect can be obtained.

[Second Embodiment] Next, a second embodiment of the surface shape reconstructing apparatus of the present invention will be described with reference to FIGS. The second embodiment is a modification of the feature point synthesizing means 1-8 of the first embodiment described above.

FIG. 6 shows a block diagram of an apparatus as a second embodiment, FIG. 7 shows a processing procedure of the apparatus of the second embodiment in a flow chart, and FIG. 8 shows a modified feature point synthesizing means. The figure explaining operation | movement of is shown.

(Structure 2) The structure of the second embodiment will be described in comparison with the first embodiment described above.
The configuration and connection relationship of each means of the embodiment are almost the same as those of the first embodiment described above. However, the difference from the first embodiment is that the feature point synthesizing means 2 as shown in FIG.
-2 and the three-dimensional image memory are connected, and in this embodiment, the connection between the two is different from that of the first embodiment described above in that bidirectional input / output is possible. Therefore, the operations and actions of the related means also have the following features.

(Operation and Effect 2) First, the operation and operation of the apparatus according to the second embodiment will be described. FIG. 7 to be referred to
Steps S1-1 to S1-5 of the flow chart are the same as those in the first embodiment described above. On the other hand, in the second embodiment,
The description about the feature point synthesizing means, that is, step S1-6 (FIG. 3A) of the first embodiment is changed, and step S2-1 is used instead.

The feature point synthesizing means 2-2 is the image memory 1-11.
The pixel position on the three-dimensional image memory 1-9 of the feature point is obtained by using the distance map stored in the table and the rotation angle obtained by the rotation control means 2-1. Feature point synthesis means 2-2
Represents the pixel position value corresponding to the feature point in the three-dimensional image memory 1
-9 is called, "1" is added to count the number of times the pixel becomes a feature point, and the result is stored in the memory 1-9 for three-dimensional image. "1" performed for the pixel that became the feature point
Is added in the geometric relationship shown in FIG.

That is, a distance map is assumed for the distance of radius d with the rotation angle obtained by the rotation control means 1-6 with respect to the center of the composite area, and the distance map is 3 for each feature point from the assumed position. Back projection is performed on the 1D image memory 1-9. At this time, it is assumed that a straight line having a length corresponding to the range where the feature points are expected to exist is parallel to the direction in which back projection is performed, centered on the position of the feature points obtained from the distance map, and this straight line passes. The above "1" is added to all the pixels. However, FIG. 8A shows only a two-dimensional plane parallel to the rotation plane of the rotation stage for simplification of description. The geometrical relationship of the three-dimensional backprojection is shown in FIG.
The two-dimensional relationship shown in the above may be accumulated. The characteristic point synthesizing means 2-2 outputs a signal indicating that the synthesizing of the characteristic points is completed to the rotation control means 2-1. (The above corresponds to step S2-1 in the flowchart of FIG. 7).

The rotation control means 2-1 performs a rotation operation on the rotary stage to renew the relative positional relationship between the subject and the TV camera fixed to the rotary stage. (The above corresponds to step S1-7 in the flowchart of FIG. 7 and is similar to the first embodiment described above).

The rotation control means 2-1 checks whether or not the rotation stage repeats a predetermined rotation, and if not completed, returns to step S1-2 of the flowchart. When the predetermined rotation is completed, a signal indicating that is output to the feature point synthesizing means 2-2. (The above corresponds to step S2-2 in the flowchart of FIG. 7).

The feature point synthesizing means 2-2 calls the data on the three-dimensional image memory 1-9, performs the threshold processing shown in FIG. 8 (b), and detects the intersections of straight lines equal to or more than the threshold. , The result is stored in the memory 1-9 for three-dimensional image. (The above corresponds to step S2-3 in the flowchart of FIG. 7).

As a result, since the three-dimensional image structure of the object is obtained in the three-dimensional image memory at this point, processing such as setting an appropriate viewpoint for the three-dimensional structure is performed. By converting the image into a two-dimensional image by using the display means 1-10, the surface shape of a solid having a smooth curve can be correctly displayed and output.

[Third Embodiment] Next, a third embodiment of the surface shape reconstructing apparatus of the present invention will be described with reference to FIGS.

FIG. 9 shows a block diagram of an apparatus as a third embodiment, and FIG. 10 shows a processing procedure of the third embodiment in a flow chart. Further, FIG. 11 shows an explanatory view of the operation of the feature point synthesizing means.

(Structure 3) The structure of the apparatus according to the third embodiment will be described in comparison with the embodiment described above. In the third embodiment illustrated in FIG. 9, the feature point synthesis of the second embodiment described above is performed. The configuration is different in that the means 2-2 (FIG. 6) is changed to the feature point composition means 3-1. However, the connection of each component is the same as in the second embodiment described above.
Therefore, the operations and actions of the related means also have the following features.

(Operation and Effect 3) First, the operation and operation of the device of the third embodiment will be described. See FIG. 10
Steps S1-1 to S1-5 of the flow chart are the same as those in the first embodiment described above. On the other hand, the description regarding the feature point synthesizing means, that is, step S2-1 in the second embodiment described above.
The operation procedure is different in that (FIG. 7) is changed to step S3-1 in the third embodiment.

The feature point synthesizing means 3-1 includes the image memory 1-11.
The pixel position on the three-dimensional image memory 1-9 of the feature point is obtained by using the distance map stored in 1) and the rotation angle obtained by the rotation control means 1-6. Feature point synthesis means 2
In step -2, the density value at the pixel position corresponding to the feature point is called from the three-dimensional image memory 1-9, the density values given to the feature points are added, and the result is stored in the three-dimensional image memory 1-9.

The addition of the density values given to the above-mentioned characteristic points is performed in the geometrical relationship shown in FIG. 11 (a). That is, the rotation control means 2 with respect to the center of the composite area
A distance map is assumed for the distance of radius d with the rotation angle obtained by -1, and back projection is performed from the assumed position on the three-dimensional image memory 1-9 according to the value of the distance map for each feature point. "Length" corresponding to the range in which this feature point is expected to exist
It is assumed that a straight line having is parallel to the direction in which back projection is performed, centered on the position of the feature point obtained from the distance map. The "density distribution" on the straight line is the largest at the characteristic points obtained from the distance map, and becomes lower as the distance from the characteristic points increases.

The density values are added to all the pixels passing through this straight line. However, FIG. 11A shows only a two-dimensional plane parallel to the rotation plane of the rotation stage for simplification of description. Therefore, the geometrical relationship of the three-dimensional back projection may be performed by stacking the two-dimensional relationship shown in FIG. The characteristic point synthesizing means 3-1 outputs a signal indicating that the synthesizing of the characteristic points is completed to the rotation control means 2-1.
(The above corresponds to step S3-1 in the flowchart of FIG. 10).

The rotation control means 2-1 performs a rotation operation on the rotary stage to renew the relative positional relationship between the subject and the TV camera fixed to the rotary stage. (The above is step S1-7 in the flowchart of FIG.
And is similar to the first embodiment described above).

The rotation control means 2-1 checks whether or not the rotation stage repeats a predetermined rotation, and if it has not finished yet, it returns to step S1-2 in the flowchart of FIG. On the other hand, when the predetermined rotation is completed, a signal indicating that is output to the feature point synthesizing means 3-1. (The above is the step S2-2 in the flowchart of FIG.
And is similar to the second embodiment described above).

The feature point synthesizing means 3-1 calls the data stored in the three-dimensional image memory 1-9 and performs the "threshold processing" shown in FIG. 11 (b) to detect the density value above a predetermined threshold value. Then, the detection result is stored in the three-dimensional image memory 1-9.
(The above corresponds to step S2-3 in the flowchart of FIG. 10).

As a result, the three-dimensional image structure of the object is obtained in the three-dimensional image memory at this point.
By performing a process for setting an appropriate viewpoint on the three-dimensional structure and converting it into a two-dimensional image, the display means 1-10 can be used to correctly display, for example, an object having a smooth curved surface shape. Can be output.

[Fourth Embodiment] A fourth embodiment of the apparatus of the present invention will be described with reference to FIGS. FIG.
2 shows a configuration diagram of an apparatus according to the fourth embodiment, and FIG. 13 shows a processing procedure of the apparatus according to the fourth embodiment in a flow chart. Further, FIG. 14 explains the operation of the changed feature point synthesizing means.

(Structure 4) The structure of the apparatus as the fourth embodiment will be described in comparison with the above-described embodiment. FIG.
In the fourth embodiment illustrated in FIG. 2, the TV camera 1-1 and the distance map generating means 1 of the second embodiment shown in FIG.
-7, feature point synthesizing means 2-2, TV camera 4-1,
The distance map generating means 4-2 and the feature point synthesizing means 4-3 are structurally different.

The connection relationship of each means constituting the device of the fourth embodiment is almost the same as that of the second embodiment described above, except that the second embodiment differs from the second embodiment described above. In addition to the connection relationship of the embodiment, between the TV camera 4-1 and the distance map generating means 4-2, and the TV camera 4-1.
And the feature point synthesizing means 4-3 are respectively connected. More specifically, the TV camera 4-1 is configured so that the distance map generating means 4-2 can directly supply the relationship data between the focal position and the distance peculiar to the optical image forming system of the TV camera 4-1. That is different. Also,
It is configured that the data relating to the depth of focus value of the optical image forming system of the TV camera 4-1 can be directly supplied from the TV camera 4-1 to the feature point synthesizing means 4-3. Is structurally different from. Therefore, the operation and action of each related means have the following features.

(Operation and Effect 4) The operation and operation of the apparatus of the fourth embodiment will be described. See FIG. 13 (a).
Steps S1-1 to S1-4 of the flowchart shown in FIG. 6 are the same as those in the first embodiment described above. On the other hand, the processing procedure is different in that step S2-1 (FIG. 7) of the second embodiment described above is changed to step S4-2.

The "distance map generation" process shown in step S4-1 of the flowchart of FIG.
Steps S1-14 and S shown in the flowchart of (b)
The detailed procedure is shown as a subroutine in 4-3 and S1-16.

That is, the distance map generating means 4-2 first initializes the image memory 1-11 for storing the distance map. By this initialization, the image memory 1-11 becomes a state indicating that the distance is not obtained. For example, if the state in which the distance has not been obtained is represented by "0", all the pixels in the image memory 1-11 are input with the value "0" after the initialization is completed. . (The above corresponds to step S1-14 of the subroutine of FIG. 13B).

The distance map generating means 4-2 is a predetermined small area around the coordinates of the feature points obtained from the feature point detecting means 1-4 in the image shown in FIG. 4B stored in the image memory. The "focus method" is applied to. The focus method is performed by the following procedure. For details, feature point detection means 1-4
The coordinates of the feature points obtained from x are as shown in FIG. 4 (c).
When the contrast is set to 1 and the contrast is expressed by the first derivative of Lfn (x), the focus position that maximizes the contrast at this x1 is obtained. That is, the focus method is performed by obtaining fn that gives the following equation.

[0074]

(Equation 2)

Assuming that the focus position that maximizes the contrast is fn, the distance α at the coordinate x1 is determined by the "relationship between the focus position and the distance" shown in FIG. 4 (d).
Can be requested.

The data relating to the "relationship between focal position and distance" shown in FIG. 4 (d) is a value peculiar to the optical image forming system of the TV camera used, and is stored in advance in the TV camera 4-1. It is stored and sent from the TV camera 4-1 to the distance map generation means 4-2 before the initialization of the distance map shown in step S1-14. This distance map generation means 4-2
Stores the obtained distance α in the distance map as shown in FIG. (The above corresponds to step S4-3 of the subroutine of FIG. 13B).

Next, the distance map generating means 4-2 checks whether or not the processing for all feature points has been completed,
If the processing has not been completed, step S4-3
Return to On the other hand, if the processing has already been completed, the series of distance map generation routines are ended and the process returns.

The feature point synthesizing means 4-3 comprises the image memory 1-11.
The pixel position on the three-dimensional image memory 1-9 of the feature point is obtained by using the distance map stored in the table and the rotation angle obtained by the rotation control means 2-1.

Further, the feature point synthesizing means 4-3 calls the value of the pixel position corresponding to the feature point from the three-dimensional image memory 1-9, and in order to count the number of times the pixel becomes the feature point, "1" is set. "" Is added and stored in the memory 1-9 for three-dimensional image.

The addition of "1" to the pixel which becomes the characteristic point in this way is performed in the geometrical relation shown in FIG. That is, a distance map is assumed at a distance of radius d at the rotation angle obtained by the rotation control means 2-1 with respect to the center of the combined area, and a three-dimensional image is obtained from the assumed position for each feature point according to the value of the distance map. "Backprojection" is performed on the memory 1-9. At this time, it is assumed that a straight line having a length equal to the depth of focus peculiar to the optical imaging system of the TV camera 4-1 is parallel to the direction in which backprojection is performed, with the position of the feature point obtained from the distance map as the center. , "1" is added to all the pixels passing through this straight line. The depth of focus value is stored in advance in the TV camera 4-1 and is shown in FIG.
In the first stage of step S4-2 in the flowchart of FIG.
It is directly sent from the V camera 4-1 to the feature point synthesizing means 4-3.

Further, FIG. 14A shows only a two-dimensional plane parallel to the rotation plane of the rotation stage for simplification of the description, but the geometric relationship of the three-dimensional backprojection is shown in FIG.
It is obtained by stacking the two-dimensional relationships shown in (a). Subsequently, the characteristic point synthesizing means 4-3 outputs a signal indicating that the synthesizing of the characteristic points is completed to the rotation control means 2-1.
(The above is step S4 in the flowchart of FIG.
-Corresponds to 2).

Next, the rotation control means 2-1 performs a rotation operation on the rotation stage to move the rotation stage, thereby renewing the relative positional relationship between the subject and the TV camera 4-1 fixed to the rotation stage. To establish a proper positional relationship. (The above is Fig. 13
This corresponds to step S1-7 in the flowchart of (a) and is similar to the first embodiment described above).

The rotation control means 2-1 checks whether or not the rotation stage repeats a predetermined rotation, and if it has not finished, the process returns to step S1-2. On the other hand, if the predetermined rotation has already been completed, a signal to that effect is output to the feature point synthesizing means 4-3. (The above is Fig. 13 (a)
(Corresponding to step S2-2 in the flowchart of FIG.

Then, the feature point synthesizing means 4-3 calls the data in the memory 1-9 for the three-dimensional image,
Predetermined threshold value processing as shown in (b) is performed to detect intersections of straight lines of a certain number or more, and the detection result is stored in the three-dimensional image memory 1-9. (The above corresponds to step S2-3 in the flowchart of FIG. 13A).

Therefore, as a result of the above, since the three-dimensional image structure of the object has been obtained in the three-dimensional image memory at this point, the appropriate viewpoint is set for the three-dimensional structure. By converting the image into a two-dimensional image, it is possible to correctly display and output even an object having a smooth curved surface shape using the display means 1-10.

(Other Embodiments and Modifications) Although the embodiments and modifications of the present invention have been described above, other various embodiments and modifications are also possible without departing from the scope of the present invention. It is possible.

The above description is based on a plurality of embodiments, but the present invention includes the following inventions. [1] A photographing means having a mechanism for controlling a focus position when photographing a subject to obtain an image, a moving means for moving a relative position of the photographing means with respect to the subject, and a memory means for storing the image. And a detection unit that detects a point having a “density difference” that occurs in the image as a feature point of the image, and in-focus position information regarding a position where the feature point of the subject is in focus. Depth information "
When a certain subject is photographed to obtain an image, a map creating means for creating a distance map in which a distance to the subject corresponding to each pixel is represented by a map associated with each pixel, and a photographing position of the photographing means. Position control means that obtains a relative positional relationship between the photographing means and the subject according to the change of, and feature points obtained from the plurality of images according to the distance map corresponding to the positional relationship between the photographing means and the subject. A surface shape reconstructing device comprising: a feature point synthesizing unit for synthesizing. (Note that the embodiment of the present invention corresponding to this [1] corresponds to the first embodiment).

(Operation and Effect 1) An image can be input while scanning the focal position by using a camera as a photographing means provided with a mechanism for controlling the focal position. It is possible to input images from multiple directions by means for moving the relative position of the camera with respect to the subject and means for obtaining the relative positional relationship between the camera and the subject as the shooting position of the camera is changed. The focus position is scanned by means for detecting a point having a density difference generated in the image of the subject as a feature point and means for obtaining depth information by the focus method for the feature point of the subject and creating a distance map. It is possible to create a distance map by obtaining the distances of the feature points where the density difference occurs in the obtained image. The feature points can be synthesized from the distance maps obtained in the multiple directions by the means for synthesizing the feature points obtained from the plurality of images according to the positional relationship between the camera and the subject and the distance map.

Even if there are few feature points that can be obtained in one direction, by combining feature points obtained from multiple directions,
A more correct surface structure can be obtained. Particularly in the case of an object whose surface structure is composed of a smooth curved surface, the feature points of the contour portion obtained by the density difference detection move according to the image input direction, so the feature points obtained in different image input directions are They will no longer overlap in position and the resulting surface structure will be more correct.

[2] The surface shape reconstructing apparatus according to [1], wherein the moving means is a rotary stage that rotates around the subject. (Note that this [2]
The first embodiment corresponds to the embodiment of the present invention corresponding to (1). The rotary stage in the first embodiment corresponds to this rotary stage.

(Operation and Effect 2) The rotating stage, which is the moving means, mounts the photographing means around the subject and changes the relative position with respect to the subject. Therefore, it is possible to provide a surface shape reconstructing device capable of taking images of multi-directional surfaces of different subjects in which the radius of gyration, that is, the relative distance to the subject is within a certain range, and easily inputting data regarding the three-dimensional shape thereof.

[3] The characteristic point synthesizing means is means for arranging and synthesizing characteristic points obtained from the plurality of images as “points” according to the positional relationship between the photographing means and the subject and the distance map. The surface shape reconstructing device according to [1]. (Note that the embodiment of the present invention corresponding to this [3] corresponds to the first embodiment). And
The feature point synthesizing means for arranging and synthesizing the feature points obtained from a plurality of images as points according to the positional relationship between the photographing means and the subject and the distance map corresponds to the feature point synthesizing means in the first embodiment.

(Operation and Effect 3) When the feature point synthesizing means synthesizes the feature points obtained from the plurality of images based on the positional relationship between the photographing means and the subject and the distance map, first, the feature points are “points”.
Place it on the map as. Therefore, it is possible to provide a surface shape reconstructing device that can reconstruct a correct surface shape by combining these feature points.

(4) The characteristic point synthesizing means forms characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map in a "straight line" having a length in the optical axis direction of the photographing means. ] The surface shape reconstructing apparatus according to [1], which is a straight line feature point synthesizing means. (Note that the embodiment of the present invention corresponding to this (4) corresponds to the second embodiment).

The characteristic point synthesizing means for synthesizing the characteristic points obtained from a plurality of images as a straight line having a length in the optical axis direction of the photographing means according to the positional relationship between the photographing means and the subject and the distance map is the feature point synthesizing means in the second embodiment. It corresponds to the feature point synthesizing means.

(5) <Combining the characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing means> The straight line feature The point synthesizing means arranges characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing means, and a plurality of straight lines. The surface shape reconstructing apparatus according to (4), further comprising: a unit that detects a point where the points intersect with each other as a feature point after composition. (Note that the embodiment of the present invention corresponding to this (5) corresponds to the second embodiment).

After combining the points where the plurality of straight lines intersect with the means for arranging the characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing means. The feature point synthesizing means in the second embodiment corresponds to the means for detecting the feature point.

(Function and Effect 5) The feature points obtained from a plurality of images are arranged as a straight line having a length in the optical axis direction of the camera according to the positional relationship between the camera as the photographing means and the object and the distance map. Is a TV for image input
It is treated as a straight line with a length in the optical axis direction of the camera. When these characteristic points are combined, a more reliable part of the straight line is detected as a surface structure by means of detecting a point where a plurality of straight lines intersect as the combined characteristic point.

(6) <Synthesis of characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing means> The straight line feature The point synthesizing unit cumulatively adds feature points obtained from a plurality of images according to the positional relationship between the photographing unit and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing unit and a density different from the background. And a means for detecting, as a feature point obtained after combining, a portion having a predetermined density as a result of the cumulative addition processing, the surface according to (4) Shape reconstruction device. (Note that this (6)
The third embodiment corresponds to the embodiment of the present invention corresponding to the above).

A means for cumulatively adding feature points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and a distance map as a straight line having a length in the optical axis direction of the camera and having a density different from that of the background, and cumulative addition. The feature point synthesizing means in the third embodiment corresponds to the means for detecting, as a feature point obtained after synthesizing, a portion maintaining a constant density as a result of performing.

(7) <Combining the characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing means> The straight line feature The point synthesizing means has a characteristic point obtained from a plurality of images in accordance with the positional relationship between the photographing means and the subject and the distance map, has a length in the optical axis direction of the photographing means, and has a "density distribution depending on the position on the straight line. Is a straight line, and means for performing cumulative addition as a straight line, and means for detecting a portion that maintains a constant “density” as a result of cumulative addition as a feature point obtained after combining. The surface shape reconstructing apparatus according to (4). (Note that the third embodiment corresponds to the embodiment of the present invention corresponding to this (7)).

Characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map are cumulatively added as a straight line having a length in the optical axis direction of the photographing means and having a density distribution according to the position on the straight line. The feature point synthesizing means in the third embodiment corresponds to the means for moving and the means for detecting a portion having a constant density as a result of cumulative addition as a feature point obtained after synthesis.

(Operation and Effect 7) The characteristic points obtained from a plurality of images according to the positional relationship between the camera as the photographing means and the object and the distance map have a length in the optical axis direction of the camera and a density distribution according to the position on a straight line. The characteristic points are treated as a straight line having a length in the optical axis direction of the TV camera at the time of image input and having a density distribution according to the position on the straight line. Further, the probability of the position where the characteristic point exists can be represented by the density distribution corresponding to the position of the straight line. When these feature points are combined, a more reliable part in the straight line is detected as a surface structure by means of detecting a part having a constant density as a result of cumulative addition as a feature point obtained after combining. .

(8) <Combining the characteristic points obtained from a plurality of images according to the positional relationship between the photographing means and the subject and the distance map as a straight line having a length in the optical axis direction of the photographing means> The straight line feature The point synthesizing unit has a characteristic point obtained from a plurality of images according to the positional relationship between the photographing unit and the subject and the distance map, which is equal to the “depth of focus” determined by the characteristics of the optical imaging system in the optical axis direction of the photographing unit. The surface shape reconstructing apparatus according to (4), which is a means for synthesizing a straight line having a length. (This (8)
The fourth embodiment corresponds to the embodiment of the present invention corresponding to (4).

Characteristic points obtained from a plurality of images are synthesized as a straight line having a length equal to the depth of focus determined by the characteristics of the optical imaging system in the optical axis direction of the photographing means according to the positional relationship between the photographing means and the subject and the distance map. The means corresponds to the feature point synthesizing means in the fourth embodiment.

(Effect 8) A characteristic point obtained from a plurality of images in accordance with the positional relationship between the camera as the photographing means and the object and the distance map has a length equal to the depth of focus determined by the characteristics of the optical imaging system in the optical axis direction of the camera. The feature point is treated as a straight line having a length equal to the depth of focus of the optical imaging system in the optical axis direction of the TV camera at the time of image input by the means for synthesizing as a straight line having a height. Since the depth of focus is an index showing the accuracy of the focusing method in the optical axis direction, it is possible to effectively detect the surface structure by making the length of the straight line equal to the depth of focus.

[0107]

As described above, according to the present invention, for example, even for an object having a shape formed of a smooth curved surface, processing of a contour portion by combining feature points obtained from multiple directions and detecting a density difference is performed. As a result, it is possible to provide a surface shape reconstructing device relating to a three-dimensional shape input device capable of more faithfully and correctly reconstructing the shape and structure of the object surface.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a surface shape reconstructing apparatus as a first embodiment relating to a three-dimensional shape input apparatus of the present invention.

FIG. 2 is a schematic diagram showing a spatial relationship between a subject, a rotary stage, and a TV camera relating to the present apparatus.

3A to 3C show a processing procedure as the first embodiment, FIG. 3A is a flowchart showing a processing procedure of the apparatus of the first embodiment, and FIG. A flowchart showing the procedure of the image input processing of
(C) is a flowchart showing a procedure of a distance increasing generation process as a subroutine.

4 (a) to 4 (e) illustrate a method of reconstructing the apparatus of the first embodiment, and FIG. 4 (a) shows a profile function of density and characteristic points at which contrast in an image occurs. Explanatory drawing showing a region and differentiation, (b) is an explanatory view conceptually showing the structure of an image memory that stores an image obtained by scanning a focal position in association with each focal position, and (c) is a diagram of density Explanatory drawing which shows the function showing a one-dimensional profile, (d)
Is a table in which focal positions correspond to distances, and (e) is an explanatory diagram showing a profile of distances in a distance map.

5 (a) to 5 (c) illustrate the operation of the feature point synthesizing means of the device of the first embodiment, and FIG. 5 (a) conceptually shows a memory for three-dimensional images, and FIG. The figure which allocated each element corresponding to three-dimensional space, (b) is explanatory drawing which represents notionally the relationship between the memory for three-dimensional images, and a distance map, (c).
FIG. 3 is a diagram conceptually showing a distance (radius) between a feature point in a memory for a three-dimensional image and a distance map.

FIG. 6 is a configuration diagram showing a surface shape reconstructing apparatus as a second embodiment relating to a three-dimensional shape input apparatus of the present invention.

FIG. 7 is a flowchart showing a processing procedure of an apparatus according to the second embodiment.

8A to 8C illustrate the operation of the feature point synthesizing unit of the apparatus according to the second embodiment, and FIG. 8A illustrates the relationship between the three-dimensional image memory and the distance map. Figure, (b) shows
Explanatory drawing which shows a threshold value process notionally.

FIG. 9 is a configuration diagram showing a surface shape reconstructing apparatus as a third embodiment relating to a three-dimensional shape input apparatus of the present invention.

FIG. 10 is a flowchart showing a processing procedure of the apparatus of the third embodiment.

11 (a) and 11 (b) explain the operation of the feature point synthesizing means of the apparatus of the third embodiment, and FIG. 11 (a) shows the relationship between the three-dimensional image memory and the distance map. Figure, (b)
FIG. 4 is an explanatory diagram conceptually showing threshold processing.

FIG. 12 is a configuration diagram showing a surface shape reconstructing device as a fourth embodiment relating to a three-dimensional shape input device of the invention.

13A and 13B show a processing procedure as an apparatus of the first embodiment, FIG. 13A is a flowchart showing a processing procedure of an apparatus of the fourth embodiment, and FIG. 10 is a flowchart showing the procedure of a distance increase generation process as a subroutine.

14 (a) and 14 (b) explain the operation of the feature point synthesizing means of the apparatus of the fourth embodiment, and FIG. 14 (a) shows the relationship between the three-dimensional image memory and the distance map. Figure, (b)
FIG. 4 is an explanatory diagram conceptually showing threshold processing.

15A to 15C illustrate a method of reconfiguring a conventional device, and FIG. 15A is an explanatory diagram showing a shooting condition;
(B) is an explanatory view showing the “focus method”, and (c) is
Explanatory drawing which shows "stereo matching".

[Explanation of symbols]

1-1, 4-1 ... TV camera, 1-2 ... Focus position control means, 1
-3 ... Image memory, 1-4 ... Feature point detecting means, 1-5 ...
Rotation stage, 1-6, 2-1 ... Rotation control means 1-7,
4-2 ... Distance map generating means, 1-8, 2-2, 3-1 ... Feature point synthesizing means, 1-9 ... Three-dimensional image memory, 1-10 ... Displaying means, 1-11 ... Image memory, S1-1 ... Initializing a three-dimensional image memory, S1-2 ... Detecting the position of the rotary stage, S1-3 ... Inputting an image (subroutine), S1-4 ... Extracting feature points, S1
-5 ... Step of generating distance map (subroutine), S1-6 ... Step of three-dimensionally plotting feature points according to position and distance map, S1-7 ... Step of moving rotary stage, S1-8 ... End judgment Step. S1-9
... step of initializing focus position, S1-10 ... step of storing image in image memory, S1-11 ... step of storing focus position in image memory, S1-12 ... step of moving focus position, S1-13 ... Step of determining the end of scanning of the focal position, S1-14 ... Step of initializing the distance map, S1-15 ... Step of storing distances to the feature points in the distance map, S1-16 ... End of processing of all feature points Step to judge. S2-1 ... 3D plotting as a straight line of feature points according to position and distance map and step of counting the number of plots, S2-2 ... Step of determining end of rotation operation, S2-3 ... in memory for 3D image Steps for thresholding. S3-1: A step of adding the density value to the corresponding pixel and the assumption as a straight line having density according to the position and distance map. S4-1: step of generating a distance map (subroutine), S4-2: step of plotting and counting the number of plots as a straight line having a length equal to the depth of focus of the optical imaging system according to the position and distance map, S4
-3 ... A step of storing the distances to the feature points in the distance map.

Claims (3)

[Claims] 1. A photographing means having a mechanism for controlling a focal position when a subject is photographed to obtain an image, a moving means for moving a relative position of the photographing means with respect to the subject, and a storing means for storing the image. Utilizing memory means, detection means for detecting a point having a "density difference" occurring in the image as a characteristic point of the image, and in-focus position information relating to a position where the characteristic point of the subject is in focus A map creating unit that obtains “depth information” and creates a distance map in which a distance to a subject corresponding to each pixel is represented by a map corresponding to each pixel when a certain subject is photographed to obtain an image; Position control means for obtaining a relative positional relationship between the photographing means and the subject as the photographing position of the photographing means is changed, and the distance marker corresponding to the positional relation between the photographing means and the subject. And a feature point synthesizing means for synthesizing feature points obtained from a plurality of the images according to the above method. 2. The moving means is a rotary stage that rotates around the subject.
The surface shape reconstructing device according to. 3. The feature point synthesizing means is means for arranging and synthesizing feature points obtained from the plurality of images as “points” according to the positional relationship between the photographing means and the subject and the distance map. The surface shape reconstructing device according to claim 1.