JPH036780A - Processing system for extraction of three-dimensional position information - Google Patents

Abstract

PURPOSE:To easily extract the 3-dimensional position information on an object by providing a revolving motion information extracting means to extract the revolving motion information out of the shift information on the object between two frame images. CONSTITUTION:An image input part 101, a feature point extracting part 102, a revolving motion information extracting part 103, a reference point distance supply part 104, and a 3-dimensional position information extracting part 105 are provided. Then a 3-dimensional position is extracted with use of the coordinates set on two frame images of characteristic points based on the dis tance between an image input device and the reference point of each feature point obtained from the revolving motion information obtained from those coordinates and the coordinates set on the frame images of feature points. Thus it is possible to extract the 3-dimensional position information on the feature points of an object just by observing the moving object.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a three-dimensional position information extraction processing method performed by observing an object with one image input device.

(Prior art) In the conventional method of extracting three-dimensional position information about feature points on an object from a series of frame images obtained by observation with a single image input device, A stereo image is constructed from two input images obtained by observing before and after the movement of an object moved by one object moving means.
Three-dimensional position information was extracted by dimensional position measurement.

(Problem to be Solved by the Invention) Therefore, it is necessary to apply a predetermined movement to an object by a predetermined amount of movement, or to manually measure the amount of movement, and in order to extract three-dimensional position information, it is necessary to The problem was that it took time and effort.

The present invention (1) In order to solve these conventional problems, it is possible to easily extract three-dimensional position information of an object by observing a predetermined movement without setting a limit on the amount of movement of the object. The object of the present invention is to provide a three-dimensional location information extraction processing method.

[Means to solve the problem]

In the three-dimensional position information extraction processing method of the present invention, features on two frame images in a series of frame images obtained by observing a moving object with a single image input device such as a television camera Extract rotational motion information from the movement information of the object between two frame images from the coordinates of the point,
The present invention is characterized by having means for extracting a three-dimensional position of a feature point with one point among the feature points as a reference from the extracted rotational motion information and the coordinates of the feature point on the two frame images.

In the conventional case, the three-dimensional position information of each feature point was extracted from the coordinates of each feature point on two frame images using the principle of triangulation, so it was necessary to determine the amount of movement of the object in advance or to measure it separately. However, the three-dimensional position information extraction processing method of the present invention has a rotational motion information extraction means for extracting rotational motion information from the movement information of an object between two frame images. However, from the rotational motion information extracted by the rotational motion information extraction means and the coordinates of the feature points on two frame images, the three-dimensional position of the feature point is extracted with one point among the feature points as a reference. This method is different from that of

[Example] Hereinafter, an example of the present invention will be described in detail based on one drawing. FIG. 1 shows a configuration diagram of the three-dimensional position information extraction processing method in this embodiment, in which one of the feature points is taken as a reference point, and the distance of the reference point to the image human-powered device is given. This is shown as an example of the case. In the same figure,
101 is an image human power unit, 102 is a feature point extraction unit, 10
3 is a rotational motion information extraction unit, 104 is a reference point distance adding unit,
105 is a three-dimensional position information extraction unit.

The image human power unit 101 observes a moving object using an image input device such as a television camera, and outputs two frame images out of a series of frame images obtained through observation to the feature point extraction unit 102. . The feature point extraction unit 102 extracts feature points from the two frame images output from the nine-image input unit 101, and converts the coordinates of the extracted feature points on the two frame images to the rotation motion information extraction unit 103 and three-dimensional position information. It is output to the extraction unit 105. Extraction of feature points from two frame images is 1. For example, points where the shape or surface color of an object locally changes abruptly are extracted by image processing, and points that are the same on the object are matched. , the associated points are taken as feature points. The rotational motion information extraction unit 103 extracts rotational motion information of the object in real space between the two frame images from the coordinate values of the feature points on the two frame images output from the feature point extraction unit 102, and extracts the rotational motion information of the object in the real space between the two frame images. Location information extraction unit 1
Output to 05. The reference point distance providing unit 104 provides a reference for the three-dimensional position of each feature point.2For example, one point among the feature points is set as a reference point, and the distance from the reference point to the image input device is provided as a reference. . At this time, if only the positional relationship between feature points is required, an appropriate value may be given as the distance of the reference point to the image input device. 2 from the position of the feature point on the frame image, the rotational motion information output from the rotational motion information extraction section 103, and the distance of the reference point to the image human-powered device outputted from the reference point distance adding section 104.
Three-dimensional position information of feature points when two frame images are input is extracted and output.

Examples of the one-rotation motion information extraction section 103 and the three-dimensional position information extraction section 105 will be described below.

If the camera model is represented by the central projection shown in Figure 2, the center of the camera lens is the origin, the optical axis of the camera is the Z axis, and the direction perpendicular to the image plane of the camera is the X axis. The three-dimensional position (Xk, Yk, Zk) of Pk is Pk
The position on the O frame image can be expressed as (tJk, Vk), and the distance from the center of the camera lens to the imaging plane of the camera can be expressed as f.

As shown in Fig. 3, the first and jth feature points Pk
The points at the time of inputting two frame images are Ptk and Pj
k (k=0.1..., N-1, N: number of feature points.

Pie, PjO are reference points), and the coordinate values of these feature points on the two frame images output from the feature point extraction unit 102 are (Uik, Vik), (Ujk,
Vjk).

Here, we will explain the implementation of the rotational motion information extraction unit 103 and the three-dimensional position information extraction unit 105 in the case where an object moves between two frame images by rotational movement around the Y axis and translational movement in the X and Z directions. Explain an example.

In Fig. 3, when the object between the two frame images 81st and jth is performing rotational movement around the Y-axis at a rotation angle β and translational movement in the X and Z directions, the linear equation holds true. .

−n ~＞ +−
〉0Pjk-OPjO-R0Pik-OPio(k≠O
) , -・---(2) However.

Since the Y coordinate values of the feature points P ik and P jk are the same when the first and jth two frame images are input, the following equation holds true.

Z ik V ik 弯 ZjkVjk (3) Substituting equation (1) into equation (2) and using equation 2 (3), Zjk
By erasing.

・−・−−(4) By eliminating Z ik and Z io from Equation 6 (4), ak6 cosβ+aml sinβ+ak1=Q
-(A) Here.

a*o= f (Ujk−[JiO) VjOVik
+ (Uik-UjO) ViOVikl fak,
−(ViOVjk−■jOVik)f”+UjOUik
ViOVjk−UiOUjkVjOVikabt= (
(UiOUik) VjOVjk+ (UjO-Uj
k) ViOVikl fIf there is no translational movement, the rotation angle β of the object between two frame images is the coordinate value (Ui., ■,.), (Uj,,
Vj. ) (Urw, Vrx>, (IJ, to,
VJk) can be expressed by equation (5). When there is a translational movement, one rotation angle β is the coordinate value (
U. ,■+o)(Uto,V4o)1(U;m,V+
k), CU, v, V=+=) (U; t, V+□), (U
It can be expressed by equation (6) using tl and "tz).

si mouth β8 am+ az2 amz axIcosβ= ・−−−・−−−1−・ ・−(6) Here, all. = ((Ujm−UiO) VjOVim+
(Uim-(JjO) ViOVjs) fa*+
−(VjOVjm −VjOVim)f”+ UjO
LJ imV iOVjm -UiOtJjmVjOV
1IIaaz= ((lJio Uill)
VjOVjm+ (IJjO-Ujm) ViOVim
In other words, using the coordinates of the feature point on the frame image output from the feature point extraction unit 102 shown in FIG.
5) The rotational motion information extraction unit 103 can be realized by providing means for calculating the rotation angle β of the object between two frame images from 1 or 1 equation (6).

Next, an embodiment of the three-dimensional position information extraction unit 105 for determining the three-dimensional position of a feature point will be described.

Using the rotation angle obtained from Equation (5) or Equation 9 (6), from Equation 1 (3) and 2 (4), the feature points P ik and P jk at the time of inputting the i-th and j-th frame images are What is the X component?

It can be expressed as (k-1, 2,..., N-1) ・・−・−−−−−1・−(7) −・−・・−・−・・(8) The X component and the X component of the three-dimensional position of the feature point are obtained by substituting equations (7) and (8) into equation (1).

That is, the rotation angle outputted from the rotational motion information extraction unit 103 shown in FIG. 1, the coordinate faUik of the feature point on the two frame images outputted from the feature point extraction unit 102,
Using Vik, LJjk, Vjk, and the Z coordinate value ZiO of the reference point that can be calculated from the distance of the reference point to the image input device output from the reference point distance adding unit 104, Equation 2 (7) and Equation (8) are obtained. Accordingly, the three-dimensional position of the feature point at the time of inputting the i-th and j-th frame images can be determined.

The rotational motion information extraction unit 103 and the three-dimensional position information extraction unit 105 when an object is performing rotational movement around the X-axis and translational movement in the Y and Z directions between two frame images are as follows: The object between the two frame images is Y
Rotational motion information extraction unit 10 when performing rotational motion around an axis and translational motion in the X direction and Z direction
In the embodiment of 3 and the 3D position information extraction unit 105,
This can be realized by replacing the component with the X component.

In addition, if an object moves between two frame images by rotating around the Z axis and translating in the X and Y directions, there is no movement in the depth direction when viewed from a single image human-powered device Since the movement of a feature point on an image itself represents actual three-dimensional movement, the rotational movement information extraction unit 103 can be realized by determining the rotation angle of the rotational movement of a feature point on a frame image. The three-dimensional position information extraction unit 105 can be realized by solving equation (2) under the condition that the X component of the three-dimensional position of the feature point at the time of inputting the frame image does not change.

〔Effect of the invention〕

As explained above, according to the present invention, the coordinates of the feature points on two frame images are used to calculate each feature point from the rotational motion information obtained from these coordinates and the coordinates of the feature points on the frame image. Since it is possible to extract the three-dimensional position based on the distance to the image of the reference point and the human-powered device, there is no need to move the object a predetermined amount of movement or manually measure the amount of movement. Three-dimensional position information of feature points on an object can be extracted just by observing a moving object.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of a three-dimensional position information extraction processing method showing an embodiment of the present invention, Fig. 2 is a diagram showing a camera model, and Fig. 3 is a three-dimensional position change of feature points due to the movement of an object. FIG. 3 is a diagram showing the correspondence between two-dimensional positional changes of feature points on a frame image and a two-dimensional position change of a feature point on a frame image. 101... Image human power department Camera's agony... Feature point extraction unit. ...Rotational motion information extraction section. ...Reference point distance adding section. ...3D position information extraction unit.

Claims (1)

[Claims] Observing a moving object using an image input device
In a three-dimensional position information extraction processing method that extracts dimensional position information, two of a series of frame images obtained from the image input device are
Rotational motion information extraction that extracts feature points from each frame image, and uses the coordinates of the extracted feature points on the two frame images to extract rotational motion information regarding three-dimensional movement of an object between the two frame images. means for extracting the three-dimensional position of the feature point with one of the feature points as a reference from the rotational motion information extracted by the rotational motion information extraction means and the coordinates of the feature point on the two frame images. A three-dimensional location information extraction processing method that is characterized by: