JP2001014466A - Camera arranging method in moving image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To stably detect the recognition of symmetry and the variation of a situation based on the movement of a recognizing object by arranging a camera so as to set the optical axis direction of a camera at a maximum angle within a settable range with respect to the direction of a plane regulated in the moving direction of the recognizing object. SOLUTION: In the case of initializing a coordinate form so as to set the optical axis of a camera to be a Z axis, an image pickup surface to be an x-y plane and an origin to be the center of the optical axis, as the result of center-projecting motion vectors (P(t), Q(t), R(t)) in the three-dimensional space of a material point P (X, Y, Z), a motion vector (optical flow) becomes (u(t), v(t)). A condition for maximizing the motion vector on the picture under a condition fixing a distance between the camera and the material point and the magnitude of the moving of the recognizing object in the three-dimensional space is R(t)=0. Tin this case, the moving direction of moving component toward the camera is matched with a Z-axis and an image pickup surface direction is reset so as to be in parallel with a plane fixed by two moving components orthogonal with this.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention analyzes a series of images to monitor, for example, the measurement of traffic flow, the occurrence of disasters such as traffic accidents, the appearance of illegal intruders, and the change of physiological conditions. In this case, the present invention relates to a camera arranging method for stably detecting target recognition and a change in situation.

[0002]

2. Description of the Related Art In recent years, with the spread of TV cameras due to their commercialization, the spread of PCs has led to the miniaturization and speeding up. Image processing application functions that were conventionally realized using dedicated hardware are based on PCs. An environment that can be realized inexpensively with a system that has been developed has been established. On the other hand, crimes in the absence of traffic accidents, burglars, and the like are on an increasing trend, and the demand for surveillance systems using images is increasing. In such a surveillance / monitor system, a product in which one or more TV cameras are installed in a surveillance area, and a human observes the image and monitors a change in situation is generally used.

In order to automate the situation change detection, an image in a standard state is recorded in advance, and the movement of the recognition target and the change in the situation are detected by performing a difference from an image obtained every moment. There has been proposed a method of detecting a change in a motion or a situation of a recognition target by detecting a change in brightness between successive images based on a difference.

Further, a method has been proposed in which a local motion vector (optical flow) is detected from an image to detect a motion of a recognition target or a change in a situation.

[0005]

In the conventional method for detecting the movement of a recognition target or a change in a situation, there is no specific method for arranging a camera used for observation. There is a problem in operation such that the attitude parameters of the camera must be set on the spot by trial and error, and it takes time to install and adjust.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a simple camera arrangement method for stably detecting an object based on the movement of the object and detecting a change in situation.

[0007]

According to the present invention, there is provided a moving image processing apparatus for recognizing an object or detecting a change in a situation by processing a series of images. A camera arrangement method for a moving image processing apparatus, characterized in that the camera is arranged so that the optical axis direction of the camera is set at an angle as large as possible within a settable range with respect to the direction of.

In the present invention, when there are a plurality of objects that move independently, a plane that minimizes the sum of the distances from the trajectories of the plurality of objects is estimated, and within a settable range, The optical axis direction of the camera is set in a direction in which the difference from the normal direction of the plane is the smallest.

According to the present invention, the moving directions of the object to be recognized and the directions of the planes defined by these moving directions are estimated.

In the present invention, the optical axis direction of the camera is set so as to form an angle as large as possible within a settable range with respect to a plane direction defined by the moving direction of the recognition target.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A configuration of a moving image processing apparatus to which a camera arrangement method according to an embodiment of the present invention is applied will be described with reference to FIG.

As shown in FIG. 1, the moving image processing apparatus includes an optical flow estimating unit 11, a correspondence candidate selecting unit 12,
The three-dimensional position calculator 13 is configured.

The optical flow estimating unit 11 estimates a motion vector (optical flow) on the screen for each of a plurality of cameras arranged in the monitoring area. The correspondence candidate selection unit 12 uses the estimation result of the optical flow estimation unit 11 to select a candidate point group from which a three-dimensional position is to be obtained. The three-dimensional position calculation unit 13 associates the candidate point group for each camera with the recognition target and calculates a three-dimensional position.

Next, a description will be given of a camera arrangement method in the above-described moving image processing apparatus.

In FIG. 2, when the coordinate system is initially set such that the optical axis of the camera is the Z axis, the imaging plane is the xy plane, and the origin is the center of the optical axis, the mass point P (X, Y, Z) The result of central projection of the motion vector (p (t), Q (t), R (t)) in the three-dimensional space is a motion vector (optical flow) (u (t ), V (t))
Becomes

(U (t), v (t)) ≡ (Z (t) p
(T) -X (t) R (t) / Z (t) (Z (t) + R
(T)), Z (t) Q (t) -Y (t) R (t) / Z
(T) (Z (t) + R (t))) The size of the motion vector on this screen under the condition that the distance between the camera and the mass point and the size of the motion of the recognition target in the three-dimensional space are constant. Is maximum from the above equation, R (t) =
In the case of 0, that is, the direction of the motion component toward the camera is made coincident with the Z axis, and the direction of the imaging surface is reset so as to be parallel to a plane defined by two motion components orthogonal to the Z axis. Due to physical restrictions on the arrangement, it is common that the distance from the camera to the target must be more than a certain distance, but in many cases, the direction of the optical axis of the camera can be set relatively freely. By setting as above, the largest motion can be observed for the same object, and stable motion detection can be realized.

Even when there are a plurality of mass points constituting an object, for example, when the object is a rigid body that moves in the same direction as in an automobile in traffic flow measurement, or in a three-dimensional manner such as a reciprocating spring pendulum. In the case of a target whose direction of movement in space does not change but whose direction changes periodically, installing a camera in the same manner as described above makes it possible to observe the largest movement stably.

On the other hand, in the case of observing a rotating object such as a top from above, the moving direction changes for each material point constituting the object. Therefore, the camera direction must be set by the same means as described above. Can not. However, in this case, since the movement vector of each mass point in the three-dimensional space is limited to a single plane orthogonal to the rotation direction, the optical axis direction of the camera is set to be parallel to the rotation direction, If the imaging plane direction is set so as to be parallel to a plane where movement is limited, a large movement can be similarly stably observed. Also, even if it is not a target that rotates, for example, when monitoring a group of people or cars moving on the ground,
When it can be considered that the movement of the recognition target is limited within the same plane (the road surface), the imaging plane direction may be set so as to be parallel to the plane where the movement is limited. This can be realized by setting the optical axis direction of the camera so as to be orthogonal to the direction of the plane on which the movement is limited.

If there is a restriction that the camera direction and the imaging plane direction cannot be set in such a direction, the optimum direction defined by the direction of the plane where the movement is limited is set within the settable range. By setting the imaging plane direction in the direction in which the difference is the smallest, that is, by setting the optical axis direction of the camera in the direction in which the difference from the above-mentioned optimum direction is the largest, it becomes possible to observe the movement stably.

When there are a plurality of objects that move independently,
Even in the case where the movement of the recognition target cannot be regarded as being limited to the same plane, for example, estimating a plane in which the sum of the distances lowered from the trajectories of the plurality of objects is minimized,
Within the settable range of the camera direction and the imaging plane direction, setting the imaging plane direction in the direction having the largest difference from the normal direction of this plane, that is, setting the camera in the direction in which the difference from this normal direction is the smallest. By setting the direction of the optical axis, it is possible to observe the movement stably in the same manner.

[0021]

By using the present invention, in various situations such as occurrence of disasters such as fires, appearance of illegal intruders, changes in physiological conditions, etc., recognition of an object based on the movement of the recognition object and changes in the situation. Can provide a simple camera placement method for stable detection of
The practical effects are enormous.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a moving image processing apparatus to which a camera arrangement method of the present invention is applied.

FIG. 2 is a diagram illustrating projection of a motion of a mass point onto an image for explaining a camera arrangement method in a moving image processing apparatus according to an embodiment of the present invention.

[Explanation of symbols]

 11 optical flow estimating unit 12 correspondence candidate selecting unit 13 three-dimensional position calculating unit

 ──────────────────────────────────────────────────続 き Continued on the front page F-term (reference)

Claims (5)

[Claims] 1. A camera arrangement method in a moving image processing apparatus for recognizing an object or detecting a change in a situation by processing a series of images, the method comprising the steps of: And arranging the camera such that the optical axis direction of the camera is set at an angle as large as possible within a settable range. 2. When there are a plurality of independently moving objects, a plane that minimizes the sum of the distances from the movement trajectories of the plurality of objects is estimated, and the method of this plane is set within a settable range. 2. The camera arrangement method according to claim 1, wherein the optical axis direction of the camera is set in a direction having the smallest difference from the line direction. 3. The camera arrangement according to claim 1, wherein the direction of the motion component toward the camera is made coincident with the optical axis of the camera, and the imaging plane is set parallel to a plane defined by two motion components orthogonal to the optical axis. Method. 4. The camera according to claim 1, wherein an optical axis direction of the camera is set to be parallel to a rotation direction of the recognition target, and an imaging surface is set to be parallel to a plane where movement is limited. Placement method. 5. The camera arrangement method according to claim 1, wherein an optical axis direction of the camera is set in a direction having a largest difference from an optimum direction defined by a direction of a plane whose movement is limited within the settable range. .