JP2785697B2 - Three-dimensional object region estimating apparatus and method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and a method for estimating a three-dimensional object area, and more particularly to a man-machine interface for photographing a person's face. The present invention relates to a three-dimensional object region estimating device and method for estimating a region of a face.

[0002]

2. Description of the Related Art Conventionally, in order to obtain an area where a target three-dimensional object appears in an image, the image is divided into continuous areas based on the color and brightness of each pixel constituting the image.
From these, the area where the object is shown is determined. As such a conventional example, a method for detecting the position of an object in a natural moving image ("Information Research Report Vol. 93, No. 62, Information Processing Society of Japan" published by Information Processing Society of Japan) by Tomitaka et al. In this example, a brightness histogram of an input natural moving image is created, and based on the histogram, an object image is separated by a local histogram growth method, and its position is detected. .

[0003]

In this conventional example, since it is necessary to perform image processing, a dedicated image processing device is required to continuously detect an area from a moving image. In the "method for detecting the position of an object in a natural moving image", a dedicated image processing device using an i860 processor manufactured by Intel Corporation of the United States is used. Further, in order to separate the area where the object is captured from the background, it is necessary that the background and the object have an image that is clearly different. For this reason, for example, it is not possible to reliably extract the face of the target person from an image taken at a street corner where many human faces are shown.

An object of the present invention is to extract a face image region by a simple device without performing image processing in view of the above points. It is another object of the present invention to reliably extract even when the background and the image of the object are similar.

[0005]

A three-dimensional object region estimating apparatus according to the present invention comprises: a position measuring means for measuring three-dimensional position information of a target three-dimensional object; Image input means for generating the image; position information measured by the position measuring means; and a known size of the three-dimensional object.
And copy the image to the image generated by the image input means.
The position and size of the three-dimensional object output in the image
Region estimating means for estimating the size.
You.

[0006]

According to the above means, when the position of the object shown in the image is obtained, the position is obtained by estimating from the three-dimensional position information of the object, not by the conventional image processing. This makes it possible to perform high-speed position estimation with simple processing without requiring the image processing device.

[0007]

Next, the present invention will be described with reference to the drawings. Referring to FIG. 1 showing a block diagram of an embodiment of the present invention, a three-dimensional object region estimating apparatus of this embodiment includes a position measuring unit 1 for measuring a position of a head 41 which is a target three-dimensional object The head 41, which is a three-dimensional object, is
The image input means 2 for taking an image and generating the image, and the position 41 in the image generated by the image input means 2 at which position in the image generated by the image input means 2 using the position information measured by the position measuring means 1 And an area estimating means 3 for estimating whether or not is captured.

FIG. 2 is a diagram for explaining the operation of this embodiment. The operation of this embodiment will be described with reference to FIG. 2 and FIG. First, the sensor 11 is attached to the top of the target head 41 in advance. As the sensor 11, a magnetic three-dimensional position sensor 3SPACE manufactured by Polhemus Inc. of the United States is used. The position measuring means 1 measures the position P of the sensor 11 and outputs it to the area estimating means 3. The image input means 2 outputs a camera image 23 obtained by photographing with the camera 21 to the area estimating means 3.
The area estimating means 3 perspectively projects the position P of the sensor 11 obtained from the position measuring means 1 onto the virtual screen 22 of the camera 21 which has been measured and obtained in advance, and obtains a projection position Q in the virtual screen 22. This projection position Q is equal to the position R of the sensor image 12 in the camera image 23. Normally, a human face is shown with the head up in the image. Therefore, if the position of the top of the head in the camera image 23 can be calculated, it can be estimated that the face image 42 is below the top of the top. On the other hand, since the sensor 11 is attached to the top of the head 41, the position of the top in the camera image 23 is equal to the position R of the sensor image 12. Therefore, camera image 2
The position of the face image 42 in 3 can be estimated to be below the position R of the sensor image 12.

Further, by calculating the distance between the position P of the sensor 11 measured by the position measuring means 1 and the camera 21, the distance between the head 41 and the camera 21 is obtained. If you know the size of
The size of the face image 42 in the camera image 23 can also be estimated. Specifically, it is estimated as follows. D is the distance between the head 41 and the camera 21, d is the distance between the virtual screen 22 and the camera 21, which is determined in advance,
Also, the width and height of the head 41 are W and H, respectively, and the face image 4
If the width and height of 2 are w and h,

[0010]

The following relationship holds. From the above, the position and size of the face image 42 in the camera image 23 can be estimated.

[0012]

As described above, according to the present invention,
The position of the target three-dimensional object is measured, the three-dimensional object is photographed, an image thereof is generated, and the position of the three-dimensional object in the image is determined using the position information. The estimation eliminates the need for a dedicated apparatus for image processing as in the related art. Further, since the calculation performed by the area estimating means is only perspective transformation, high-speed processing can be performed. Further, since the position information of the object is used, even if a plurality of objects similar to the object appear in the video, they are not confused with other objects.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a diagram for explaining the operation of this embodiment.

[Explanation of symbols]

 Reference Signs List 1 position measuring means 2 image input means 3 area estimating means 11 sensor 12 sensor image 21 camera 22 virtual screen 23 camera image 41 head 42 face image P position of sensor 11 Q projection position R position of sensor image 12

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields investigated (Int. Cl. ⁶ , DB name) G06T 1/00 H04N 5/232 H04N 7/15 H04N 7/18 G08B 13/196 G01S 13/86 JICST file ( JOIS)

Claims (2)

(57) [Claims] 1. A position measuring means for measuring three-dimensional position information of a target three-dimensional object, an image input means for photographing the three-dimensional object and generating an image thereof, and the position measuring means comprises: The measured position information and the known size of the three-dimensional object
Using the image data to copy the image generated by the image input means.
The position of the three-dimensional object in the image,
A three-dimensional object region estimating device, comprising: a region estimating means for estimating a size . 2. Measuring three-dimensional position information of a target three-dimensional object, photographing the three-dimensional object, generating an image of the three-dimensional object, and calculating the position information and a known size of the three-dimensional object.
And the three-dimensional object projected on the image
A three-dimensional object region estimating method characterized by estimating a position and a size in the image .