CN1205429A - Device and method for estimating three-dimensional position of moving target - Google Patents

Abstract

A device and method for estimating the three-dimensional position of a moving target. The device comprises: a plurality of markers attached to the moving object, each marker can emit or reflect a different color; at least two color cameras for photographing the moving object with the marker; An image processor that calculates the three-dimensional position of a moving target from each marker on two different images and from the position of the marker with the same color in the captured image. Therefore, the marks of different colors attached to the moving target can be easily and quickly recognized in different color images.

Description

Device and method for estimating three-dimensional position of moving target

The present invention relates to a device and method for estimating three-dimensional position, in particular to a device and method for estimating three-dimensional position of a moving object, wherein color marks are added to the moving object.

Virtual reality is a computing application that provides realistic effects through the use of advanced computer graphics and simulation techniques. For example, using a virtual reality program, when designing a building, various problems can be predicted before the building is constructed.

In the above virtual reality program, when a person or object (hereinafter, the moving object will be referred to as a target) moves, it is necessary to estimate the three-dimensional position of the target by tracking the motion of the target in order to smooth the motion of the target. In order to estimate the 3D position of a moving target, two images of the same target must be obtained simultaneously. Using conventional methods, these two images are obtained as follows.

First, black and white marks that can reflect light are attached to various parts of the target, and images of the target are continuously captured by multiple cameras at 0.01-second intervals. A computer connected to the camera checks all pixel values output from the camera. Here, since each marker only provides light and shade information, it can only be determined whether a part of the image corresponds to a marker, but different markers cannot be distinguished. Thus, identification of the markers is performed by taking a sequence of images with each camera, observing the motion of the markers in each sequence, and matching the markers between each camera sequence based on the observed motion. Only then can the three-dimensional position of each marker be calculated.

However, in the above conventional methods, it is only possible to decide whether a part of an image represents a marker or not by using the shading information. Whether the labels of two images acquired at the same time are the same can only be determined after processing the next set of images. Thus, it takes a long time to calculate the position information.

To solve the above problems, the object of the present invention is to provide a device and method for estimating the three-dimensional position of a moving target, wherein a plurality of marks of different colors are attached to the target, so that each mark can be easily recognized in a color image.

According to one aspect of the object, there is provided an apparatus for estimating a three-dimensional position of a moving object using marks attached to a plurality of parts of the moving object, the apparatus comprising: a plurality of marks attached to the moving object, and each marking emits or reflective different colors; at least two color cameras for photographing a moving target with a marker; and identifying each marker on at least two different images captured simultaneously by a plurality of color cameras, and identifying each marker from the plurality of images captured An image processor with the same colored marker position to calculate the three-dimensional position of the moving object.

According to another aspect of the object, there is provided a method for estimating the three-dimensional position of a moving target, the method comprising the following steps: (a) attaching a plurality of markers each emitting or reflecting a different color to the moving target; (b) utilizing at least two cameras, simultaneously photographing a target with a plurality of markers; (c) extracting images of the markers from the plurality of color images captured simultaneously; (d) identifying the markers by the color of the images, and in at least two of the images calculating the position of the same marker; and (e) calculating the three-dimensional position of the object using the marker position calculated in step (d).

By describing a preferred embodiment of the present invention in detail in conjunction with the accompanying drawings, the above-mentioned purpose and advantages of the present invention will become more apparent, wherein,

Fig. 1 is a schematic diagram of a moving target according to the present invention and a device for estimating the three-dimensional position of the moving target;

Fig. 2 is a flow chart explaining the method for estimating the three-dimensional position of a moving object according to the present invention; and

Fig. 3 is a schematic diagram explaining the steps of Fig. 2 for calculating the three-dimensional positions of a plurality of markers typically contained in two different images of the same object.

In FIG. 1 , the device for estimating the three-dimensional position of a moving object according to the present invention includes: a marking unit 102 ; a camera unit including a first camera 104 , a second camera 106 and a third camera 108 ; and an image processor 110 . Here, reference numeral 100 represents a moving object.

The marking section 102 includes a plurality of colored markings. Here, these marks are attached to parts of the moving object 100 that move frequently, such as a person's head, arms, hands, and legs. Color markers are light emitting diodes (LEDs) or colored objects that reflect light. Using the cameras 104, 106, and 108 of the camera section, a plurality of images of a moving object to which a plurality of color marks are attached are taken. More than one camera is needed because, firstly, at least two images are required to obtain the three-dimensional position, and secondly, images from only one direction are not enough to capture all markers.

Image processor 110 extracts images of markers from at least two input images from first, second, and third cameras 104, 106, and 108, identifies the color of the markers, and calculates the three-dimensional position of each marker.

The operation and effect of the present invention will be described below with reference to FIGS. 2 and 3 .

First, a plurality of colored marks 120 are attached to respective parts of the sporting object 100 (part 200). The target images are then input by the cameras 104, 106 and 108 to the image processor 110 (step 202).

Image processor 110 examines each pixel value of the two input images (step 204). Then, it is determined whether each pixel in the image corresponds to a marker (step 206). Here, if the pixel value is greater than 100, that point in the image is identified as a marker. Otherwise, the point in the image is determined to be the background and the next pixel value is checked. If that point in the image is determined to be a marker, the color (red, green, blue) is checked to select the same marker from different images. Here, marker images having the same color are determined to represent the same marker, and the positions of a plurality of matching marker images are calculated in screen coordinates (step 208). Then, the three-dimensional position of the marker is calculated from the position of the matched marker image by stereo matching (step 210). Here, when one object point in space is projected to two image points, stereo matching is used to match two points on multiple planes into one point in order to obtain a three-dimensional position from the two-dimensional position of the object. Figure 3 explains the steps of calculating the 3D position of the same marker in two different images of the same object by stereo matching. In Fig. 3, it is assumed that the positions and focal lengths _f1 and _f2 of the first and second cameras are known.

Referring to FIG. 3 , the position (x ₁ , y ₁ ) in the first image 304 of the target W captured from the first camera 300 and the position (x ₂ ) in the second image 306 of the target W captured by the second camera 302 , y ₂ ), the three-dimensional position W(x, y, z) of the target W can be calculated as follows.

The position (x ₁ , y ₁ ) is calculated by the following equation (1), and the equation of the line L ₁ passing through W(x, y, z) from the position of the first camera 300 is calculated therefrom. $x_1=\frac{f_{1}x}{f_1-z},{\mathrm{the\; y}}_1=\frac{f_1\mathrm{the\; y}}{f_1-z}--------\left(1\right)$

The position (x ₂ , y ₂ ), is calculated by the following formula (2), and the formula of the line L ₂ passing through w(x, y, z) from the position of the second camera 302 is calculated therefrom.

Here, the intersection point of the two lines L ₁ and L ₂ satisfies the relationship of L ₁ =L ₂ , and the intersection point corresponds to the three-dimensional position W(x, y, z) of the target W.

As described above, in the apparatus and method for estimating the three-dimensional position of a moving object according to the present invention, a color image is taken and color marks reflecting different colors are attached to the moving object. Each mark is identifiable by its color from the color image. Thus, the same marker in different images of the same object can be easily and quickly identified.

Claims (3)

1. An apparatus for estimating a three-dimensional position of a moving object using marks attached to a plurality of parts of the moving object, the apparatus comprising: Multiple markers attached to a moving target, each marker can emit or reflect a different color; at least two color cameras filming moving objects bearing said markings; and An image processor that recognizes each marker in at least two different images captured simultaneously by multiple color cameras, and calculates the three-dimensional position of the moving target from the positions of multiple markers with the same color in the multiple captured images. 2. A method for estimating a three-dimensional position of a moving target, the method comprising the following steps: (a) attaching multiple markers each emitting or reflecting a different color to a moving target; (b) using at least two cameras to simultaneously take pictures of the target with the mark; (c) extracting labeled images from simultaneously captured color images; (d) identifying the mark by the color of its images and calculating the position of the same mark in at least two images; and (e) Computing the three-dimensional position of the object using the position of the marker calculated in step (d). 3. 2. The method of claim 2, wherein step (e) of calculating the three-dimensional position is performed by stereo matching two images having the same marker.

Images