JP2573347B2 - Automatic shooting position determination device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an apparatus for automatically determining a photographing position when a real object such as a product, a work of art, a flora and fauna is photographed from a plurality of directions. The image information photographed from the photographing position determined by the present invention is stored in the center device, and is used for an image information providing device that searches and provides the terminal device via a transmission line.

(Prior Art) When a real object such as a product, a work of art, a flora and fauna is photographed from a plurality of directions, a virtual hemisphere covering an object 0 as shown in FIG.
Assuming OA, there is a method of determining the photographing position at equal angular intervals in the longitude direction ψ (horizontal direction) and the latitude direction λ (vertical direction) of this virtual hemisphere.

However, in this case, the extreme point where the latitude λ of the virtual hemisphere OA increases
As approaching OC, the density of photographing positions increases, and images photographed from near the poles have substantially the same information (although the posture of the object in the screen changes). Therefore, it is understood that the distance between the photographing positions on the hemisphere needs to be constant.

However, even if a plurality of photographing positions are determined in this way, a difference in information substantially received by a viewer of the image photographed from each photographing position greatly changes depending on the shape of the object and the photographing position. .

For example, an image of a rectangular parallelepiped as shown in FIG. 9 (FIG. 10 (a)) is compared with an image obtained by rotating the rectangular parallelepiped by 10 degrees (FIG. 10 (b)). The difference between the images felt when viewed, and the comparison between the image viewed from the direction (FIG. 11 (a)) and the image rotated by 10 degrees (FIG. 11 (b)) As for the difference, the difference when viewed from the obvious direction is larger.

This means that even if the interval between the photographing positions on the hemisphere is constant, the difference in information between the images is not constant. Therefore, when these images are displayed one after another in the order in which the imaging positions are adjacent to each other, the information that the viewer substantially receives increases in one section and decreases in another section. That is, when such a set of images is stored, the storage efficiency in terms of substantially how much information is given to a viewer deteriorates.

Conventionally, in order to solve such a problem, there has been proposed a method of automatically determining a shooting position at which a difference in information between images is constant. In this conventional method, the sum of the amount of information generated and the amount of information disappeared due to the movement of the viewpoint position is obtained based only on the shape of the object to be photographed, and is determined as the difference between the images.

However, actually, pattern information such as color exists on the surface of the object to be photographed. Therefore, the information difference between the images should include the change in the pattern information, but not in the conventional method. For this reason, even if the information difference between the images due to the shape is small, the information difference between the images does not become constant when the information greatly varies depending on the pattern.

(Object of the Invention) It is an object of the present invention to provide a method for capturing a plurality of images along a virtual hemisphere that covers an object, by using a point at which a difference in information difference between the captured images is generated. It is an object of the present invention to provide an apparatus for automatically determining a shooting position at which a difference in information between images becomes constant by solving based on pattern information on a surface.

(Structure of the Invention) (Differences between the Features of the Invention and the Conventional Technique) In the present invention, when a plurality of images are taken along a virtual hemisphere covering the object O, information of the information between the taken images is obtained. The most important feature is that the amount of change in surface information corresponding to the difference is obtained based on the shape of the object and the pattern on the surface of the object, and the photographing position is determined so that this is constant between the images.

In the present invention, the object shape and the surface information change amount based on the pattern on the object surface are defined as follows. First, as shown in FIG. 3 (b), an object O to be photographed (hatching means a pattern) shown in FIG. 3 (a) is converted into a fine polygon ωi (i = 0, 1,..., M−1). It is assumed that each polygon ωi is given color information at a corresponding position of the object O.

As shown in FIG. 4, when the object O is first photographed from the position u, an area where a certain polygon ωi is projected on the screen surface among a plurality is defined as su. Next, when the object O is photographed from the position v, the projection area of the polygon ωi on the screen surface is sv.

The n kinds of colors on the surface of the object O are represented by Ij (j = 0,..., N−1).
And At this time, if the projection area of the n kinds of colors Ij projected on the area su is pj, the projection area of each color is p (su) = (p0,..., Pn-1) (1) ). Assuming parallel projection for simplicity, the change di of the projected area of each color of the polygon ωi projected on the screen surface between two images taken from positions u and v is di = p (su ) −p (sv) = (Δp0,..., Δpn−1) (2) If the component Δpj of di is positive, it can be considered that the information has increased by an amount proportional to Δpj for the n types of colors Ij due to the viewpoint movement. If it is negative,
It can be considered that the information has decreased by an amount proportional to Δpj.

Therefore, the sum Duv of the norm {di} of the change di of the projected area of each color obtained for each polygon ωi (3) can be regarded as the amount of change in information between the image photographed from the position u and the image photographed from the position v.
Duv is called a surface information change amount.

The present invention focuses on the fact that the amount of change in surface information is proportional to the amount of change in the amount of information between two images, and can be extracted from only the shape information of the object and the pattern information of the surface. It is characterized in that the photographing position is selected so as to be substantially constant between the images to be processed, thereby making the difference in information between the images constant.

On the other hand, in the related art, since the photographing positions are simply determined at equal intervals, the amount of change in the information on the object O between the images is not constant.

(Embodiment) FIG. 1 is a block diagram of a basic configuration of the present invention. In the figure, reference numeral 1 denotes a shape information input unit for inputting shape information of an object, and 2 denotes pattern information input for inputting pattern information of the surface of the object. The unit 3 approximately divides the shape information input from the shape information input unit 1 into a plurality of polygon planes ωi, and converts the pattern information input from the pattern information input unit 2 into the plurality of polygon planes ωi. The shape / pattern information approximation unit 4 that is associated with the surface / plane information is approximated by the shape / pattern information approximation unit 3, and the surface information change amount based on the viewpoint change for a plurality of polygon planes ωi associated with the pattern information. The detected surface information change amount detection unit 5 extracts a representative longitude and a representative latitude of the photographing position so that the surface information change amount detected by the surface information change amount detection unit 4 is substantially constant between adjacent images. Representative position extraction unit, Reference numeral 6 denotes a representative photographing position set to the representative longitude and representative latitude extracted by the representative position extracting unit 5.

FIG. 2 shows a specific configuration described above. The shape / pattern information approximating unit 3 includes a shape approximating unit 31, a pattern approximating unit 32, and a shape / pattern approximating unit 32.
It is a pattern information storage unit 33. Surface information change amount detection unit 4
Is composed of a latitude direction change amount measurement unit 41, an averaging unit 42, and a longitude direction change amount measurement unit 43. The averaging unit 42 further includes a storage unit 421 and a calculation unit 422. The representative position extracting unit 5 includes a representative latitude extracting unit 51 and a representative longitude extracting unit 52, and both of these units include storing units 511 and 521 and extracting units 512 and 522, respectively. The representative shooting position 6 includes a representative latitude 61 and a representative longitude 62.
Are output from the extraction units 512 and 522 of the representative position extraction unit 5, respectively.

To operate this, the shape information input unit 1 inputs the shape information of the target object obtained in advance, and similarly, the pattern information input unit 2 outputs the pattern information of the object surface (this is generally monochrome, that is, One color information or color information such as full color). The three-dimensional shape information of the object is obtained by projecting a reference light such as a laser beam or a slit light on the surface of the object, photographing a projected image of the reference light on the surface of the object with a camera or the like, and obtaining an image. From the relationship between the position of the projected image of the reference light and the emission position of the reference light, there is known a method of obtaining the position based on the principle of triangulation. The pattern information on the surface of the object can be easily obtained by imaging the object with a general photographing device such as a video camera. Of course, it is not necessary to be limited to these. In any case, here, the shape information of the target object,
It is assumed that the pattern information is obtained in advance and stored in a storage device or the like (not shown). The shape information input unit 1 and the pattern information input unit 2 are used to input shape information and pattern information of a target object from this type of storage device. It is. Shape / pattern information approximation unit 3
The shape approximation unit 31 reads the shape information of the object from the shape information input unit 1 and converts the shape of the object into m fine polygons ωi (i
= 0, 1,..., M−1), and stores the shape information of each polygon ωi and its outward normal vector information in the shape / pattern information storage unit 33.

On the other hand, the pattern approximating unit 32 reads the pattern information of the object surface from the pattern information input unit 2 and selects n types of colors Ij (j = 0,.
1), for example, by back-projecting the pattern information onto an object planarly approximated by the m polygons ωi, the polygons ωi and the polygons ωi back-projected onto the polygons ωi. The pattern information portions are stored in the shape / pattern information storage unit 33 in association with each other.

The latitude direction change amount measuring section 41 of the surface information change amount detecting section 4 reads out the shape information of the object approximated by a polygon and the pattern information associated with each polygon ωi from the shape / pattern information storage section 33. Based on this, as shown in FIG. 5, the surface information change amount D0 when the longitude of the shooting position where the object is shot is fixed at 変 化 = 0 and the latitude λ is moved by a fine angle Δλ
(Λ) is determined from equation (3). Where the angle Δλ = π /
(2 · C0) (C0 is a natural number).

Then, the longitude ψ is moved by a fine angle Δψ, and DΔψ (λ) is similarly obtained. However, Δψ = 2π / C1 (C1 is a natural number). This is repeated until the longitude φ = 2π, and the surface information change amount Dψ (λ) (ψ = Δψ, 2
Δψ, ..., 2π, λ = Δλ, 2Δλ, ..., π / 2)
The data is accumulated in the accumulation unit 421 of the averaging unit 42.

The arithmetic unit 422 reads the surface information change amount Dψ (λ) in the latitude direction stored in the storage unit 421, and calculates the average value Da (λ) for the longitude ψ. Is calculated and stored in the storage unit 511 of the representative latitude extraction unit 51.

The extraction unit 512 obtains the average value Da (λ) (λ = Δ
λ, 2Δλ,..., π / 2) and, as shown in FIG. 6, the integral of the average value D _a (λ) is a predetermined interval of the surface information change in the latitude direction. A division position in the latitude direction (λ direction) is determined so as to be equal to the value δ. Then, x latitudes Λi (i = 0, ...,
x-1) is output as the representative latitude 61, and this is output to the longitude direction change amount measuring unit 43.

The longitude direction change amount measuring unit 43 reads out shape information and pattern information obtained by planarly approximating the object shape from the shape / pattern information storage unit 33, and then outputs the latitude Λi (i = 0, ..., x-1). Then, similarly to the latitude direction change amount measuring unit 41, the surface information change amount D 変 化 i (Λ) (i = 0,..., X−1, ψ = Δψ, 2Δ) in the longitude direction.
ψ, ..., 2π), and accumulates in the representative longitude extractor 52
To accumulate.

Longitudinal surface information change amount D stored in storage unit 521
Λi (ψ) is read out, and the integral value of DΛi (ψ) is equally divided by the interval value δ as shown in FIG. Then, yi longitudes Ψij (j = 0,..., Y−1, i = 0,..., X−1) serving as division positions are obtained and output as the representative longitude 62.

In this embodiment, the description has been given assuming that the interval value δ of the surface information change amount is preset in the apparatus, but the interval value δ may be input from the outside.

If the representative latitude and the representative longitude are determined based on the shape information and the pattern information of the object in this manner, the image obtained by shooting the object from the shooting position on the virtual hemisphere determined by the representative latitude and the representative longitude is an adjacent image. The amount of change in surface information during this period becomes substantially constant.

(Effects of the Invention) As described above, the amount of change in surface information in the present invention is extracted based on shape information and pattern information. It is possible to keep the information difference between the images constant. Therefore, when these images are continuously displayed, the images can be displayed more smoothly than when the objects are photographed at equal intervals. When these images are stored in a database or the like, the storage efficiency in terms of the amount of information substantially given to a viewer can be improved as compared with the case where the images are simply taken at equal intervals.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic configuration of the present invention, FIG. 2 is a block diagram showing a specific configuration thereof, FIG. 3 is an explanatory diagram of a plane approximation of an object to be photographed by a polygon ωi, and FIG. FIG. 3 is a diagram showing a projected area pj on the screen surface when the polygon ωi is photographed from the positions u and v, FIG. 5 is an explanatory diagram for obtaining a surface information change amount of an object, and FIG. 6 is predetermined. Average value using the interval value δ of the surface information change amount in the latitude direction
FIG. 7 is an explanatory diagram when the integral value of Da (λ) is equally divided, and FIG. 7 is a diagram for equalizing the integral value of the average value DΛi (ψ) using a predetermined interval value δ of the surface information change amount in the longitudinal direction. FIG. 8 is an explanatory diagram for dividing an object, FIG. 8 is an explanatory diagram for determining a photographing position in a longitude direction and a latitude direction of an object by a conventional method, FIG. 9 is a diagram of a rectangular parallelepiped viewed from two directions, FIG. 10 and FIG. 10 is a diagram showing a difference between images when the rectangular parallelepiped in FIG. 9 is viewed from two directions. 1 ... shape information input section, 2 ... pattern information input section, 3 ...
Shape / pattern information approximation unit, 4 ... surface information change amount detection unit,
5 ... representative position extraction unit, 6 ... representative shooting position, 31 ... shape approximation unit, 32 ... pattern approximation unit, 33 ... shape / pattern information storage unit, 41 ... latitude direction change amount measurement unit, 42 ...... Averaging part,
43 …… Longitudinal change amount measurement unit, 51 …… Representative latitude extraction unit,
52 ... representative longitude extraction unit, 61 ... representative latitude, 62 ... representative longitude, 421,511,521 ... storage unit, 422 ... calculation unit, 512,522
.... Extraction unit.

Claims (1)

(57) [Claims] 1. A virtual hemisphere (hereinafter, referred to as a hemisphere) on a space covering an object.
A virtual hemisphere), and automatically determines a plurality of representative shooting positions (hereinafter, representative latitude and representative longitude) suitable for expressing the pattern and shape of the object on the virtual hemisphere. A determining device, comprising: means for inputting shape information of the object; means for inputting pattern information constituted by color information of a surface pattern of the object; and a plurality of three-dimensional shapes of the input shape information. Means for approximating a plane with a polygon, and associating the input pattern information with a plurality of polygons whose planes are approximated; and setting the plurality of polygons on the virtual hemisphere from the same position as the object. When projected onto a position and a position separated by a minute distance from the position, the sum of the absolute values of the differences between the projection areas of the color information constituting the pattern information corresponding to the plurality of polygons is defined as the surface information change amount. Define Over the entire area on the virtual hemisphere, a means for detecting each surface information change amount by separately moving the latitude and longitude of the position where the plurality of polygons are projected by a minute position, and the latitude and longitude of the virtual hemisphere. For each of the directions, the division value in the latitudinal direction was determined so that the integral value obtained as the sum of the small areas of the surface information change amount was equal to a predetermined latitudinal value, and the integral value was similarly predetermined. Means for determining a division position in the longitude direction so as to be equal to the value in the longitude direction, and determining each division position as a representative latitude and a representative longitude.