Moving body recognition apparatus

• US 5,414,474 A
• Filed: 03/04/1993
• Issued: 05/09/1995
• Est. Priority Date: 03/04/1992
• Status: Expired due to Term

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Yanagishita, Hideki, Shiohara, Morito, Kamada, Hiroshi, Ohta, Yoshiyuki
• Primary Examiner(s): Britton, Howard W.
• Assistant Examiner(s): Le, Vu
• Application Number:
  US08/026,685
• Time in Patent Office: 796 Days
• Field of Search: 348/169, 348/170, 348/171, 348/172, 348/135, 348/136, 348/141, 348/113, 348/116, 348/50, 348/26, 348/155, 348/700, 348/699, 382/44, 382/1
• US Class Current: 348/700
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments

Abstract

Apparatus which recognizes a shape and movement of a moving body based on position data of feature points in an image of the moving body in an orthogonal projection image of the moving body. The moving body recognition apparatus includes a feature point position normalization unit for providing normalized coordinates of a feature point by determining an image capture plane of the moving body as the XY plane, a direction in which an orthogonal projection image of a rotation axis of the moving body formed on the image capture plane as the Y axis and a direction of an image capture operation as a Z axis perpendicular to the XY plane and a rotation axis calculation unit for calculating the angle between the axis of rotation of the moving body and the Y axis.

Claims

1. A moving body recognition apparatus, comprising image input means for receiving an image of a moving body, feature point extraction means for extracting feature points from said image and feature point storage means for storing said extracted feature points, for recognizing a shape and movement of a moving body based on position data of the feature points in an orthogonal projection image of the moving body, comprising:

feature point position normalization means for providing normalized coordinates of the feature points in the XYZ coordinate system stored in the feature point storage means by determining an image plane of the moving body in the image input means as the XY plane, the direction of the orthogonal projection image of a rotation axis of the moving body formed on the image plane as the Y axis and a direction of an image capture operation by the image input means as the Z axis perpendicular to said XY plane; and

rotation axis calculation means for calculating an angle between the rotation axis of the moving body and the Y axis as a depression angle PSI based on the output of said feature point position normalization means.

2. The moving body recognition apparatus according to claim 1 further comprising known data input means for inputting known data relating to the shape and movement of the moving body, and wherein said rotation axis calculation means determines a result which is not inconsistent with the known data as the result of the calculation of said depression angle PSI.

3. The moving body recognition apparatus according to claim 1 wherein said feature point position normalization means selects a first feature point from among a plurality of said feature points of the moving body in the image plane as the origin of the XYZ coordinate system and obtains the coordinate values of a second of said feature points and the following said feature points relative to the origin as the normalized coordinates.

4. The moving body recognition apparatus according to claim 3 wherein said rotation axis calculating means comprises:

Y coordinate determination means for comparing normalized Y coordinates of one of said feature points output from said feature point position normalization means at a plurality of instants in time;

first PSI calculation process means for calculating the depression angle PSI when said Y coordinate determination means judges that, with regard to a second feature point of two of said feature points at a second instant in time out of three instants in time, the Y coordinate is the same and the absolute value of the X coordinate is not the same as the second feature point at a first instant in time; and

second PSI calculation process means for calculating a value of sin PSI and the sign of cos PSI for the angle PSI and obtaining a value of said angle PSI by using the value of said sin PSI and the sign of said cos PSI when said Y coordinate determination means judges that the Y coordinates of the second feature point are all different at respective three instants in time.

5. The moving body recognition apparatus according to claim 4 further comprising recognition impossible determination process means for making recognition of the angle PSI impossible, thereby performing an impossible recognition determination process, when said Y coordinate determination means judges that the absolute value of the X coordinate and the Y coordinate of the second feature point selected from among the two feature points are the same at least at two instants in time out of three instants in time.

6. The moving body recognition apparatus according to claim 4 wherein said first PSI calculation process means makes said depression angle PSI zero when said Y coordinates are not negative and makes the depression angle PSI 180°

when said Y coordinates are negative.

7. The moving body recognition apparatus according to claim 4 wherein said second PSI calculation process means obtains the value of sin PSI and the sign of said cos PSI by using the coordinate (x_i, y_i) (time point i=1, 2, or 3) of the second feature point at respective three time points and expressing as h wherein h is the component in the rotation axis direction of the position vector from the origin to the second feature point, based on the following equations:

space="preserve" listing-type="equation">sin PSI=±

[(y.sub.1 -y.sub.2)(y.sub.2 -y.sub.3)(y.sub.3 -y.sub.1)/{(x.sub.2.sup.2 -x.sub.1.sup.2)(y.sub.3 -y.sub.1)

space="preserve" listing-type="equation"> -(x.sub.3.sup.2 -x.sub.1.sup.2)(y.sub.2 -y.sub.1)}].sup.1/2 ;

or

space="preserve" listing-type="equation">sin PSI=±

[(y.sub.1 -y.sub.2)(y.sub.2 -y.sub.1)(y.sub.3 -y.sub.1)/{x.sub.1.sup.2 (y.sub.2 -y.sub.3)

space="preserve" listing-type="equation"> +x.sub.2.sup.2 (y.sub.3 -y.sub.1)+x.sub.3.sup.2 (y.sub.1 -y.sub.2)}].sup.1/2

and

space="preserve" listing-type="equation">h cos PSI={(x.sub.i.sup.2 -x.sub.1.sup.2) sin.sup.2 PSI+(y.sub.i.sup.2 -y.sub.1.sup.2)}/{2(y.sub.i -y.sub.1){(i=2,

     3).

8. The moving body recognition apparatus according to claim 7, wherein:

(1) cos PSI=0 when sin² PSI-1

(2) the sign of h cos PSI is determined as the sign of cos PSI when h cos PSI±

0

(3) the sign of cos PSI is not determined when sin² PSI±

1 and h cos PSI=0,

thereby providing PSI and PSI+π

as the solution in the case of (3).

9. The moving body recognition apparatus according to claim 4 wherein said second PSI calculation process means obtains the value of sin PSI and the sign of cos PSI by using the coordinates (x_i, y_i) (time point i=1, 2, or 3) of the second feature point at respective three time points and expressing as h wherein h is the component in the rotation axis direction of the position vector from the origin to the second feature point, based on the following equations:

space="preserve" listing-type="equation">sin PSI=±

[(y.sub.1 -y.sub.2)(y.sub.2 -y.sub.3)(y.sub.3 -y.sub.1)/{(x.sub.2.sup.2 -x.sub.1.sup.2)(y.sub.3 -y.sub.1)

space="preserve" listing-type="equation"> -(x.sub.3.sup.2 -x.sub.1.sup.2)(y.sub.2 -y.sub.1)}].sup.1/2 ;

or

space="preserve" listing-type="equation">sin PSI=±

[(y.sub.1 -y.sub.2)(y.sub.2 -y.sub.3)(y.sub.3 -y.sub.1)/{x.sub.1.sup.2 (y.sub.2 -y.sub.3)

space="preserve" listing-type="equation"> +x.sub.2.sup.2 (y.sub.3 -y.sub.1)+x.sub.3.sup.2 (y.sub.1 -y.sub.2)}].sup.1/2 ;

and

space="preserve" listing-type="equation">h cos PSI={(x.sub.2.sup.2 -x.sub.1.sup.2)(y.sub.3.sup.2 -y.sub.1.sup.2)-(x.sub.3.sup.2 -x.sub.1.sup.2)(y.sub.2.sup.2 -y.sub.1.sup.2)}/

space="preserve" listing-type="equation"> 2{(x.sub.2.sup.2 -x.sub.1.sup.2)(y.sub.3 -y.sub.1)-(x.sub.3.sup.2 -x.sub.1.sup.2)(y.sub.2 -y.sub.1)}

or

space="preserve" listing-type="equation">h cos PSI={x.sub.1.sup.2 (y.sub.2.sup.2 -y.sub.3.sup.2)+x.sub.2.sup.2 (y.sub.3.sup.2 -y.sub.1.sup.2)+x.sub.3.sup.2 (y.sub.1.sup.2 -y.sub.2.sup.2)}/

space="preserve" listing-type="equation"> 2{x.sub.1.sup.2 (y.sub.2 -y.sub.3)+x.sub.2.sup.2 (y.sub.3 -y.sub.1)+x.sub.3.sup.2 (y.sub.1 -y.sub.2)}.

10. The moving body recognition apparatus according to claim 9, wherein:

(1) cos PSI=0 when sin² PSI=1

(2) the sign of h cos PSI is determined as the sign of cos PSI when h cos PSI±

0

(3) the sign of cos PSI is not determined when sin² PSI±

1 and h cos PSI=0,

thereby providing PSI and PSI+π

as the solution in the case of (3).

11. The moving body recognition apparatus according to claim 4 wherein said second PSI calculation process means obtains the value of sin PSI and the sign of cos PSI by using the coordinates (x_i, y_i) (time point i=l, 2 or 3) of the second feature point at respective three instants in time and expressing as h wherein h is the component in the rotation axis direction of the position vector from the origin to the second feature point, based on the following equations:

space="preserve" listing-type="equation">h cos PSI={(x.sub.2.sup.2 -x.sub.1.sup.2)(y.sub.3.sup.2 -y.sub.1.sup.2)-(x.sub.3.sup.2 -x.sub.1.sup.2)(y.sub.2.sup.2 -y.sub.1.sup.2)}/

space="preserve" listing-type="equation"> 2{(x.sub.2.sup.2 -x.sub.1.sup.2)(y.sub.3 -y.sub.1)-(x.sub.3.sup.2 -x.sub.1.sup.2)(y.sub.2 -y.sub.1)};

or

space="preserve" listing-type="equation">h cos PSI={x.sub.1.sup.2 (y.sub.2.sup.2 -y.sub.3.sup.2)+x.sub.2.sup.2 (y.sub.3.sup.2 -y.sub.1.sup.2)+x.sub.3.sup.2 (y.sub.1.sup.2 -y.sub.2.sup.2)}/

space="preserve" listing-type="equation"> 2{x.sub.1.sup.2 (y.sub.2 -y.sub.3)+x.sub.2.sup.2 (y.sub.3 -y.sub.1)+x.sub.3.sup.2 (y.sub.1 -y.sub.2)}

and

space="preserve" listing-type="equation">sin PSI=±

[2(y.sub.k -y.sub.1)h cos PSI-(y.sub.k -y.sub.1).sup.2 }/{x.sub.k.sup.2 -x.sub.1.sup.2 }].sup.1/2 (k=2,3).

12. The moving body recognition apparatus according to claim 11, wherein:

(1) cos PSI=0 when sin² PSI=1

(2) the sign of h cos PSI is determined as the sign of cos PSI when h cos PSI≠

0

(3) the sign of cos PSI is not determined when sin² PSI≠

1 and h cos PSI=0, thereby providing PSI and PSI+π

as the solution in the case of (3).

13. A moving body recognition apparatus, comprising image input means for receiving an image of a moving body, feature point extraction means for extracting feature points from said image and feature point storage means for storing said extracted feature points, for recognizing the shape and movement of a moving body based on position data of the feature points in an orthogonal projection image of the moving body, comprising:

feature point position normalization means for providing normalized coordinates of the feature points in the XYZ coordinate system stored in the feature point storage means by determining an image plane of the moving body in the image input means as the XY plane, the direction of the orthogonal projection image of a rotation axis of the moving body formed on the image plane as the Y axis and a direction of an image capture operation by the image input means as the Z axis perpendicular to said XY plane;

shape and movement calculation means for obtaining the normalized Z coordinate of the feature points of the moving body as the shape of the moving body, and a rotation amount θ

around the rotation axis of the moving body as the movement, based on a depression angle PSI between the rotation axis of the moving body and the Y axis and the output of said feature point position normalization means.

14. The moving body recognition apparatus according to claim 13 wherein said moving body recognition apparatus comprises known data input means for receiving known data relating to the shape and movement of the moving body;

and

said shape and movement calculating means outputs a result which is not inconsistent with the known data input to the known data input means as the calculation result of the shape and movement of the moving body.

15. The moving body recognition apparatus according to claim 13 wherein said feature point position normalization means provides the first feature point out of four of said feature points of the moving body on the image plane as the origin of the XYZ coordinate system and obtains the coordinate values of the second to fourth feature points relative to the origin as the normalized coordinates with regard to two instants in time;

said shape and movement calculating means calculates a matrix R representing the rotation of tile moving body from the first instant in time to the second instant in time based on the following matrix R, when the value of the matrix with the normalized coordinates of the second to fourth feature points as its elements is not zero;

##EQU21## where ##EQU22## represents the coordinate value of the j-th feature point at the i-th instant in time (i=1, 2;

j=2, 3,

     4).

16. A moving body recognition apparatus according to claim 13 wherein said feature point position normalization means provides a first feature point of a plurality of said feature points of the moving body in the image plane as the origin of said XYZ coordinate system and obtains the coordinates of a second said feature point and following said feature points relative to the origin as the normalized coordinates.

17. A moving body recognition apparatus according to claim 16 wherein said moving body recognition apparatus comprises:

known data input means for inputting known data relating to shape and movement of said moving body; and

PSI determination means for determining the value of said depression angel PSI wherein said shape and movement calculating means calculates the shape and movement of the moving body based on the output of said feature point position normalization means in accordance with the recognition mode corresponding to the known data relating to the shape and movement of the moving body which is inputted into said data input means.

18. A moving body recognition apparatus according to claim 17 wherein when the content of the known data inputted into known data input means designates that the moving body moves on a plane, rotating at a constant rotation speed, said shape and movement calculation means comprises a constant calculation means for calculating the shape and movement of the moving body based on the output from the feature point normalizing means of two feature points of the moving body at three equally-spaced instants in time in accordance with the recognition mode for the content of the known data.

19. A moving body recognition apparatus according to claim 17 wherein when the content of the known data inputted into known data input means designates that the moving body has an edge containing a right angle h as its shape data, said shape and movement calculating means comprises the right-angle shape calculating means for calculating shape and movement of the moving body based on the output of the feature point position normalizing means for three feature points which form a right angle on the moving body at two instants in time.

20. A moving body recognition apparatus according to claim 17 wherein said shape and movement calculating means comprises shape and general calculating means for calculating the shape and movement of the moving body based on the output of the feature point normalization means for three of said feature points of the moving body at three instants in time, when the content of the known data input to the known data input means designates that the moving body moves on a plane, rotating at a known constant rotation speed and that the shape does not have an edge including a right angle.

21. A moving body recognition apparatus according to claim 16 further comprising known data input means for inputting known data relating to a depression angle PSI determination means and wherein said PSI determination means determines the value of PSI by using the output of said known data input means and the output of said feature point position normalization means and outputs the determination result PSI to said shape and movement calculating means.

22. A moving body recognition apparatus according to claim 24 wherein said shape and movement calculating means obtains the rotation amount θ

of the moving body with regard to a rotation around said rotation axis when the value of x_i² is not 0, by using normalized coordinate value (x_i y_i) (time point i=1,2) at two instants in time of the second feature point, based on the following equations;

space="preserve" listing-type="equation">cos θ

=(x.sub.1 x.sub.2 +y.sub.1 y.sub.2)/(x.sub.i.sup.2 +y.sub.i.sup.2)(i=1 or

     2)

space="preserve" listing-type="equation">sin θ

=(x.sub.1 y.sub.2 -x.sub.2 y.sub.1)/(x.sub.i.sup.2 +y.sub.i.sup.2)(i=1 or

     2).

23. A moving body recognition apparatus according to claim 21 wherein, when said PSI determination means determines that the value of PSI is neither 0°

nor 90°

, said shape and movement calculating means calculates the component h in the direction of the rotation axis of the position vector from the origin to the second feature point when a provided Y coordinate y₁ and y₂ are not equal, by using normalized coordinates (x_i, y_i) (time point i=1,

     2) of the second feature point at two instants in time which is selected from two of said feature points outputted from said feature point position normalization means and a normalized Z coordinate z_i of the second feature point, based on the following equations;

space="preserve" listing-type="equation">h={(x.sub.2.sup.2 -x.sub.1.sup.2) sin.sup.2 PSI+(y.sub.2.sup.2 -y.sub.1.sup.2)}/2 cos PSI(y.sub.2 -y.sub.1),

space="preserve" listing-type="equation"> [z.sub.1 (h-y.sub.1 cos P)/]z.sub.i =(h-y.sub.i cos PSI)/sin PSI(i=1 or

     2).

24. A moving body recognition apparatus according to claim 23 wherein said shape and movement calculating means comprising recognition impossible determination process means determines that the recognition of the shape of said moving body is impossible when a normalized Y coordinates y₁ and y₂ of the second feature point at two instants in time are equal, thereby performing the recognition impossible determination process.

25. A moving body recognition apparatus according to claim 23 wherein said shape and movement calculating means obtains the rotation amount θ

of the second feature point between two time points by using normalized X coordinate x_i (time point i=1,2) and normalized Z coordinate z_i (i=1,

     2) of the second feature point at two instants in time, based on the following equations;

space="preserve" listing-type="equation">z.sub.i '"'"'=-y.sub.i sin PSI+z.sub.i cos PSI,

space="preserve" listing-type="equation">cos θ

=((X.sub.1 X.sub.2 +Z.sub.1 '"'"'Z.sub.2 '"'"')/(x.sub.1 x.sub.2 +z.sub.1 '"'"'z.sub.2 '"'"'))/(x.sub.i.sup.2 +z.sub.i '"'"'.sup.2)

space="preserve" listing-type="equation">(i=1 or

     2)

space="preserve" listing-type="equation">sin θ

=((X.sub.1 Z.sub.2 '"'"'+X.sub.2 Z.sub.1 '"'"')/(x.sub.1 z.sub.2 '"'"'+x.sub.2 z.sub.1 '"'"'))/(x.sub.i.sup.2 +z.sub.i '"'"'.sup.2)

space="preserve" listing-type="equation">(i=1 or

     2).

26. A moving body recognition apparatus according to claim 23 wherein said shape and movement calculating means obtains a distance d between the second feature point and the rotation axis of the moving body by using the normalized coordinates (x_i, y_i) at two instants in time of the second feature point, a component h in the direction of said rotation axis of the position vector and the depression angle PSI:

space="preserve" listing-type="equation">d={x.sub.i.sup.2 +(y.sub.i -h cos PSI).sup.2 /sin.sup.2 PSI}.sup.178 (i=1,

     2).

27. A moving body recognition apparatus, comprising image input means for receiving an image of a moving body, feature point extraction means for extracting feature points from said image and feature point storage means for storing said extracted feature points, for recognizing a shape and movement of a moving body based on position data of the feature points in an orthogonal projection image of the moving body, comprising:

feature point position normalization means for providing normalized coordinates of the feature points in the XYZ coordinate system stored in the feature point storage means by determining an image plane of the moving body in the image input means as the XY plane, the direction of the orthogonal projection image of a rotation axis of the moving body formed on the image plane as the Y axis and a direction of an image capture operation by the image input means as the Z axis perpendicular to said XY plane;

shape and movement calculating means for calculating the shape and movement of the moving body in a virtual rotation state of the moving body when the rotation axis of the moving body is made in parallel to the YZ plane to permit the rotation axis of the moving body to match with the Y axis based on the output from the feature point position normalization means and a depression angle PSI between the rotation axis of the moving body and the Y axis, thereby obtaining the shape and movement of the moving body in the actual rotation state of the moving body based on said calculation result.

28. The moving body recognition apparatus according to claim 27 wherein said feature point position normalization means provides a first feature point out of four of the feature points of the moving body at the image plane to the origin of the XYZ coordinate system and obtains the coordinate of the second to fourth feature points for the origin as the normalized coordinate with regard to two instants in time;

shape and movement calculating means the matrix R representing the rotation of the moving body from the first instant in time to the two instants in time when the value of the matrix having the normalized coordinate of the second to fourth feature points as their elements is not zero, based on the following matrix R;

##EQU23## where ##EQU24## represents the coordinate value of a j-th feature point at an i-th instant in time (i=1, 2;

j=2, 3,

     4).

29. The moving body recognition apparatus according to claim 27 further comprises a known data input means for receiving the known data relating to the shape and movement of the moving body;

and

said shape and movement calculating means calculates shape and movement of the moving body based on the result which is not inconsistent with the known data input to the known data input means.

30. A moving body recognition apparatus according to claim 27 wherein said feature point position normalization means provides a first feature point out of a plurality of said feature points of the moving body on the image plane as the origin of the XYZ coordinate system and obtains the coordinates of a second of said feature points and the following said feature points relative to the origin as said normalized coordinate.

31. The moving body recognition apparatus according to claim 30 further comprising known data input means for inputting known data relating to the depression angle PSI, depression angle PSI determination means for determining a value of the depression angle PSI, wherein

said PSI determination means determines the value of the depression angle PSI based on the output of the known data input means and the output of the feature point position normalization means and outputs the determination result of the depression angle PSI to the shape and movement calculating means.

32. The moving body recognition apparatus according to claim 31 wherein, when said PSI determination means determines that the value of the depression angle PSI is not π

/2+nπ

(n is an integer), said shape and movement calculating means obtains the normalized Z coordinate z of the feature point in an actual rotation state by using z'"'"' as the Z coordinate calculating result of one of said feature points of the moving body in the virtual rotation state, normalized Y coordinate y of the one feature point and the value of the depression angle PSI, based on the following equation;

space="preserve" listing-type="equation">z=(y sin PSI+z'"'"')/ cos PSI.

33. The moving body recognition apparatus according to claim 31 wherein, when said PSI determination means determines that the value of the depression angle PSI is not π

/2+nπ

(n is integer), said shape and movement calculating means obtains a rotation amount θ

of the moving body from the first instant in time to the second instant in time when the

space="preserve" listing-type="equation">(x.sub.i.sup.2 +z.sub.i '"'"'.sup.2 (i=1 or

     2)

is not zero by using

space="preserve" listing-type="equation">cos θ

=((X.sub.1 X.sub.2 +Z.sub.1 '"'"'Z.sub.2 '"'"')/(x.sub.1 x.sub.2 +z.sub.1 '"'"'z.sub.2 '"'"'))/(X.sub.i.sup.2 +Z.sub.i '"'"'.sup.2)/(x.sub.i.sup.2 +z.sub.i '"'"'.sup.2))

space="preserve" listing-type="equation">(i=1 or

     2)

space="preserve" listing-type="equation">sin θ

=((X.sub.1 Z.sub.2 '"'"'-X.sub.2 Z.sub.1 '"'"')/(x.sub.1 z.sub.2 '"'"'-x.sub.2 z.sub.1 '"'"'))/((X.sub.i.sup.2 +Z.sub.i.sup.2)/(x.sub.i.sup.2 +z.sub.i '"'"'.sup.2))(i=1 or

     2)

and using z₁ '"'"', z₂ '"'"' as the z coordinate calculation result of the second feature point out of two of said feature points at two instants in time with regard to the moving body in the virtual rotation state and the two normalized values of the X coordinate x₁, x₂ of the second feature point.

34. The moving body recognition apparatus according to claim 31 wherein, when said PSI determination means determines that the depression angle PSI is π

/2+nπ

(n is an integer), said shape and movement calculating means obtains a rotation amount θ

of the moving body around said rotation axis by using the following equations;

space="preserve" listing-type="equation">cos θ

=((X.sub.1 X.sub.2 +Y.sub.1 Y.sub.2)/(x.sub.1 x.sub.2 +y.sub.1 y.sub.2))/((X.sub.i.sup.2 +Y.sub.i '"'"'.sup.2)/(x.sub.i.sup.2 +y.sub.i.sup.2))

space="preserve" listing-type="equation">(i=1 or

     2)

space="preserve" listing-type="equation">sin θ

=((X.sub.1 Y.sub.2 -X.sub.2 Y.sub.1)/(x.sub.1 x.sub.2 -x.sub.2 y.sub.1))/((X.sub.i.sup.2 +Y.sub.i.sup.2)/(x.sub.i.sup.2 +y.sub.i.sup.2))

space="preserve" listing-type="equation">(i=1 or

     2)

when the value of x_i² +y_i² is not zero, by using the normalized coordinate values (x_i, y_i) (i=1,

     2) of the second feature point at the two instants in time.

35. The moving body recognition apparatus according to claim 30 further comprising PSI determination means for determining the value of said depression angle PSI and known data input means for inputting the known data relating to the shape and movement of the moving body and wherein said shape and movement calculating means comprises calculation method determination means for determining a calculation method of the shape and movement of the moving body by using the output of the feature point position normalizing means, in accordance with the content of the known data output from the known data input means, when the value of the PSI is not equal to π

/2+nπ

(n is an integer).

36. The moving body recognition apparatus according to claim 35 wherein said shape and movement calculating means comprises a movement constant calculating means for calculation which is deviated when said calculating determination means judges that the content of said known data represents that the moving body moves on the plane rotating at a constant rotation and for calculating the shape and movement of the moving body based on the feature point position normalization means for two of said feature points of the moving body at three instants in time with equal time intervals.

37. The moving body recognition apparatus according to claim 35 wherein said shape and movement calculating means comprises the shape rectangular which is activated when said calculating determination method means determines that the content of the known data represents that the moving body has a rectangular h as its shape and moves on the plane while rotating and for calculating the shape and movement of the moving body based on the output of the feature point position normalization means for three of said feature points of the rectangular shape of the moving body at two instants in time.

38. The moving body recognition apparatus according to claim 35 wherein said shape and movement calculating means comprises shape generic calculating means which is activated when the calculation method determination method means determines that the content of the known data represents that the moving body moves around a plane by rotating at a known constant rotating speed and the moving body does not have a rectangular shape and for calculating the shape and the movement of the moving body based on the output of the feature point position normalization means for three of said feature points of the moving body at three instants in time.

39. A moving body recognition apparatus for recognizing a movement of a moving body based on positional data in an image of feature points of the moving body generated by an image input means, comprising;

depression angle recognizing means for obtaining an angle between a rotation axis of the moving body and the Y axis when an image taking plane of the moving body of the image input means is determined as an XY plane and an image taking direction of the image input means is determined as a Z axis; and

movement recognition means for recognizing the movement of the moving body based on a value of the angle obtained by said depression angle recognition means.