JP2573153B2 - Apparatus and method for detecting certainty between stereo images - 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 detecting certainty between correspondences between stereo images, and more particularly, to such a stereo that can determine correspondence between feature points on a stereo image having a convergence angle. The present invention relates to an apparatus and method for detecting certainty between images.

[0002]

2. Description of the Related Art In three-dimensional measurement using a stereo image, there is a method using an image by a sensor of a center projection system. The general method is a method that does not use an epipolar line described later. Therefore, the epipolar line will be described first with reference to FIG.

[0003] With reference to FIG. 9, X _r points on the other image R corresponding to the point X _l on the image L is the projection center XO _l of the two images, XO _r and the point X _l make plane 101 and image R
Are on the straight line where they intersect. This straight line 103 is called an epipolar line. If the images L and R are rearranged along the epipolar line, the line of the corresponding point search matches the line of the image, so that the search may be a one-dimensional search and the algorithm is simplified. Therefore, it is important that the image be rearranged along the epipolar line.

FIG. 10 shows a rearrangement along an epipolar line.
It is a figure for explaining a column. Referring to FIG.
Projection center X of L_lIs the origin, and x 'axis and y' axis are
Each u_lAxis, v_lA coordinate system (x ′, y ′,
z '). This coordinate system is called a model coordinate system. This
In the model coordinate system, the position of the projection center of the sensor of the image L
Is (0,0,0) and the posture is (0,0,0). Ma
The position of the projection center of the sensor of the image R is (B_x,
B_y, B_Z), And the posture is (ω ′, φ ′, κ ′).
B_xIs the unit length, and b_y= B_y/ B_x, B_Z= B_Z
/ B_xThen the model coordinate system and each image coordinate
The coordinate conversion equation between the target system and the coordinate system is as shown in Equation (1).
The unknown in the equation (1) is related to the image R by (b
_y, B _Z), (Ω ′, φ ′, κ ′). this
The five unknowns satisfy the coplanar condition expression shown in Expression (2).
Is determined as follows.

[0005] Therefore, it is necessary to first select five or more points that can be recognized as the same on an image and measure the image coordinates. Then, five unknowns are determined using the measured image coordinates.

Therefore, in order to detect the degree of correspondence between images using epipolar lines, at least five points must be 2 points.
Must be recognized as identical on one image. Therefore, if it is not possible to select five or more points that can be recognized as the same,
For example, if five or more points cannot be selected automatically,
In this manner, the correspondence between stereo images using epipolar lines is not performed.

On the other hand, there is a method in which a corresponding point is searched for without rearrangement along the epipolar line as described above. In this method, assuming that a plane image coordinate system is (u, v) and an object space coordinate system is (x, y, z) as basic expressions in a center projection type sensor image, Expression (3) is obtained. Can be Here, (x ₀ , y ₀ , z ₀ ) is the position of the projection center of the sensor, (ω, φ, κ) is the attitude of the sensor, and (u ₀ , v ₀ ) is the position of the viewpoint. Position and c
Is the scale factor in the transformation to the image coordinate system. Further, the position (x ₀ , y ₀ ,
z ₀ ) and posture (ω, φ, κ) and viewpoint position (u
₀ , v ₀ ) is a function of time t.

As can be seen from equation (3), for each of the stereo images, the position (x ₀ , y ₀ , z ₀ ) of the projection center of the sensor, which is an unknown number, and the attitude (ω, φ,
κ), the viewpoint position (u ₀ , v ₀ ) and the scale factor c in the transformation of the image coordinate system must be determined. Because of these unknowns, it is difficult to make correspondence between images using a method that does not use epipolar lines.

Therefore, the present invention solves the above-mentioned problem and provides a stereo image correspondence certainty detecting apparatus and method capable of easily associating feature points between stereo images. It is to be.

[0010]

(Equation 1)

[0011]

(Equation 2)

[0012]

According to a first aspect of the present invention, there is provided an apparatus for detecting the degree of certainty in associating a stereo image with a first image feature point on a first image in a left visual field from a first viewpoint. A stereo-image associative certainty detecting apparatus for detecting a certainty regarding a correspondence relationship with a second image feature point on a second image in a right visual field from a second viewpoint, comprising: A first straight line calculating means for calculating a first straight line passing through the viewpoint and the first image feature point; and a second straight line calculating means for calculating a second straight line passing through the second viewpoint and the second image feature point. Straight line calculation means, distance calculation means for calculating the distance between the first straight line and the second straight line,
There is provided certainty determining means for determining the certainty regarding the correspondence between the first image feature point and the second image feature point in accordance with the calculated distance.

According to a second aspect, the certainty determining means of the first aspect determines the certainty based on a function representing the certainty, which is inversely proportional to the calculated distance.

According to a third aspect, the function of the second aspect has a constant value below a predetermined value. According to the fourth aspect of the present invention, there is provided a stereoscopic image association certainty degree detecting method, wherein a first image feature point on a first image in a left view from a first viewpoint and a right image in a right view from a second viewpoint are provided. A stereoscopic image association certainty detecting method for detecting a certainty regarding a correspondence relationship with a second image feature point on a second image, comprising: a first viewpoint and a first viewpoint;
A first step of calculating a first straight line passing through the second image feature point; a second step of calculating a second straight line passing through the second viewpoint and the second image feature point; A third step of calculating a distance between the straight line and the second straight line, and determining, based on the calculated distance, a degree of certainty with respect to the correspondence between the first image feature point and the second image feature point. And a fourth step of determining.

According to a fifth aspect, the fourth step of the fourth aspect includes a step of determining the certainty based on a function representing the certainty, which is inversely proportional to the calculated distance. .

According to claim 6, the function of claim 5 has a constant value below a predetermined value.

[0017]

According to the first aspect of the present invention, there is provided a stereo image associative certainty degree detecting apparatus for forming a first straight line passing through a first viewpoint and a first image feature point on a first image in a left visual field. Calculating, calculating a second straight line passing through the second viewpoint and the second image feature point on the second image in the right visual field, and calculating the first and second images according to the calculated distance. The certainty of the correspondence between the feature points can be determined.

According to the stereoscopic image association certainty detecting device according to the second aspect of the present invention, the certainty can be determined based on a function inversely proportional to the calculated distance.

According to a third aspect of the present invention, there is provided a stereo image correspondence certainty degree detecting apparatus which uses a function which becomes a constant value when the distance is equal to or less than a predetermined value, and which fixes the certainty degree when the distance is equal to or less than the predetermined value. Can be.

According to a fourth aspect of the present invention, there is provided the stereoscopic image association certainty degree detecting method, wherein a first straight line passing through a first viewpoint and a first image feature point on a first image in a left visual field is formed. Calculating a second straight line between the second straight line and the second image feature point on the second image in the right field of view, calculating a distance between the first straight line and the second straight line, According to the calculated distance, the certainty factor of the correspondence between the first and second image feature points can be determined.

According to the fifth aspect of the present invention, the method of detecting the certainty factor between stereo images can determine the certainty factor based on a function inversely proportional to the calculated distance.

According to a sixth aspect of the present invention, there is provided a stereoscopic image association certainty degree detecting method using a function which becomes a constant value when the distance is equal to or less than a predetermined value. Can be.

[0023]

FIG. 1 is a diagram for explaining the principle of a stereo image correspondence certainty detecting device according to an embodiment of the present invention. FIG. 2 is a diagram showing the correspondence between stereo images according to an embodiment of the present invention. It is the schematic block diagram which showed the structure of the attachment certainty detection apparatus.

Referring to FIG. 1, the three-dimensional position of the feature point is
It is well known that triangulation can be determined from binocular images
Have been. First viewpoint XO_lImage on image 61 for
The coordinates of the tip of a two-dimensional finger that is an image feature point are x_l(I) ↑
(↑ represents the following vector.) = ((X_l(I),
y_l(I))^TIs defined as Similarly, the second perspective
XO_rTwo-dimensional image feature points on the image 62
The coordinates of the tip of the finger are x _r(J) ↑ = (x_r(J), y_r
(J))^TIs defined as I-th image 61
Finger tip x_l(I) ↑ is the tip x of the j-th finger in the right image
_r(J) Know the camera parameters corresponding to ↑
The three-dimensional finger tip position X (i, j) ↑ = (X
(I, j), Y ((i, j), Z (i, j))^TIs straight
Line L_l(I) and straight line L_rFind the intersection with (j)
Is determined by This straight line L_l(I) and straight line L_r
(J) respectively show the first viewpoint XO_lAnd image feature point x_l
(I) A straight line connecting ↑ and a second viewpoint XO_rAnd images
Feature point x_r(J) A straight line connecting ↑.

However, the image feature point x_l(I) and
Image feature point x_r(J) Do not know the correspondence with ↑
If no intersections occur. That is, as shown in FIG.
_l(I) and straight line L_rWhen (j) has a torsion relationship
There is. In that case, they are separated by a distance d (i, j).
Then, considering the influence of noise, the image feature point x_l(I) and
x_r(J) Certainty function C indicating the degree of correspondence with ↑
_d(I, j) is defined as in equation (4). here,
d (i, j) is a straight line L_l(I) and straight line L_r(J) and
Distance and d₀Is a constant. This constant d₀Is
What is necessary is just to determine according to the required certainty factor. No.
Equation (4) shows only a one-dimensional limitation for each image.
However, not all correspondences are determined. Ma
Equation (4) indicates that the magnitude of the certainty factor is inversely proportional to the distance.
And the distance becomes d ₀When
Has a certainty factor (1 / d₀)
I have. Furthermore, only equation (4) determines the certainty factor
If the distance d (i, j) becomes large instead of the function of
And the distance d (i, j) becomes smaller,
Any function can be used as long as the function shows a proper value.

The stereo image correspondence certainty degree detecting device 39 based on such a principle includes a first straight line calculating section 63 for calculating a straight line L _l (i) and a second straight line calculating section 63 for calculating a straight line L _r (j). , A distance calculating unit 65 for calculating the distance d (i, j), and a certainty determining unit 67 for determining the certainty using a predetermined function shown in the equation (4).

[0027]

(Equation 3)

Next, a description will be given of a hand posture detecting apparatus to which the stereo image correspondence certainty degree detecting apparatus 39 is applied. This hand posture detection device is a non-wearable device, and when an object having a thickness that can be regarded as a planar shape such as a hand is viewed from a viewpoint on the same plane, as an extreme example, it is represented as a single straight line. The posture of the hand can be detected even when the phenomenon of occlusion which appears to occur. for that reason,
The hand posture detection device detects the posture and position of an object using the position of the tip of a finger, the center of gravity of the hand, and the direction of the hand as feature amounts indicating the characteristics of the hand. FIG. 3 is a schematic block diagram of the feature extracting unit 1 that can extract the center of gravity of the two-dimensional hand, the position of the tip of the two-dimensional finger, and the direction of the two-dimensional hand obtained in each image.

Referring to FIG. 3, feature extracting unit 1 includes color space converting unit 3, hand region probability distribution unit 5 using colors, Sobel filter 7, inter-frame difference calculating unit 9, multiplier 1
1, a low-pass filter 13, a level discriminator 15,
It includes statistical operation units 17 and 19, an affine transformation unit 21, a run-length encoding unit 23, an edge detection unit 25, and a group processing unit 27.

The color space conversion unit 3 converts the input RGB image into an HSV image, inputs the H image and the S image to the hand region probability distribution unit 5 using colors, and converts the V image into a Sobel filter 7 and It is input to the inter-frame difference calculator 9. The background image is also input to the inter-frame difference calculator 9, which performs a difference calculation between frames of the input image and outputs the result to the multiplier 11. The H and S images obtained by the hand region probability distribution unit 5 using colors are also input to the multiplier 11, and an image representing a result of the multiplication is supplied to the low-pass filter 13.

The low-pass filter 13 filters the applied image and supplies the filtered image to the level discriminator 15. The level discrimination unit 15 binarizes the image with a predetermined threshold value and outputs the binarized binary image to the statistical operation units 17 and 19. The statistical operation unit 17 performs an arithmetic operation on the given binary image with a statistical operator and outputs data relating to the two-dimensional center of gravity of the hand. This data is also provided to the affine transformation unit 21.

On the other hand, the statistical operation unit 19 performs an arithmetic operation on the V image filtered by the Sobel filter 7 and the binary image binarized by the level discrimination unit 15 by using a statistical operator, and calculates a two-dimensional hand direction. Output data about This data is also provided to the affine transformation unit 21. The affine transformation unit 21 performs affine transformation using an affine operator having parameters set by the center of gravity of the two-dimensional hand and the direction of the two-dimensional hand. The affine-transformed image subjected to the affine transformation is provided to the run-length encoding unit 23. The run-length encoding unit 23 encodes the affine-transformed image into a run-length encoded image and supplies the encoded image to the edge detection unit 25. The edge detection section 25 performs edge detection at high speed. The edge detected at high speed is given to the group processing unit 27,
After being grouped into groups from the top to the bottom of the images, the images corresponding to the fingers form a group,
By removing the group, a point at the tip of the finger is obtained.

Here, all processing that does not include edge detection and grouping can be pipelined.
Also, all processes can be performed in parallel on binocular images. That is, the features of the hand from the binocular image are effectively extracted using the parallel pipeline algorithm.
Hereinafter, a hand posture detection device that can detect a three-dimensional hand posture even when occlusion has occurred based on the extracted two-dimensional hand characteristics will be described.

FIG. 4 is a schematic block diagram of the hand posture detecting device. Referring to FIG. 4, the center of gravity of the two-dimensional hand 2
9. The hand posture detecting device capable of detecting the hand posture from the two-dimensional finger tip position 31 and the hand direction 33 is the stereo image correspondence certainty degree detecting device (C _{d in the} drawing _).
(Represented by (i, j)) 39. Further, the hand posture detection device includes a three-dimensional feature point extraction unit 45, multipliers 47 and 48, and a certainty detection unit (C _s (i, j) in the drawing).
) 49 and a correspondence detection unit (P in the drawing)
_{m, k} (i, j) 51) and a posture detector 53.

In particular, important configurations of the hand posture detecting device include a stereo image correspondence certainty detecting device 39, a certainty detecting portion 49, a correspondence detecting portion 51,
The posture detection unit 53. The posture detection unit 53 includes an erroneous correspondence removal table creation unit 55, a determination unit 57, an optimal erroneous correspondence removal table creation unit 59, and a multiplier 60. Then, the posture detection unit 53 estimates the Euler angle Φ58.

On the other hand, the three-dimensional feature point extraction unit 45 includes a three-dimensional hand center-of-gravity detection unit 35, a coordinate conversion unit 37, a conversion unit 43, and an Euler angle detection unit 41.

The operation of the configuration shown in FIG. 2 excluding the stereo image correspondence certainty detecting device will be described in detail including the principle.

By the way, the stereo image associative certainty detector 39 based on the principle shown in FIG. 1 can indicate only one-dimensional limitation of distance, and does not determine all correspondences. Therefore, next, the three-dimensional feature point extraction unit 45 that detects the three-dimensional finger tip point X (i, j)} that is the three-dimensional feature point shown in FIG. 1 will be described.

[0039] FIG. 5 is a diagram showing the relationship between X _h Y _h Z _h coordinate system and the XYZ world coordinate system fixed to the hand the center of gravity of the hand as the origin.

Referring to FIG. 5, the center of gravity of the three-dimensional hand O =
(X_o, Y_O, Z_O)^TAnd three-dimensional hand direction vector V
↑ = (X_V, Y_V, Z_V)^TIs the two-dimensional center of gravity of the hand
Dimensional hand direction vector or v_l↑ = (X_vl,
Y_vl)^TAnd v_r↑ = (X_vr, Y _vr)^TFrom each
It is easily determined. These decisions are made in three dimensions as shown in FIG.
The hand's center of gravity detection unit 35 and the hand direction detection unit (not shown) are determined.
I have decided. Here, as shown in FIG.
Defined X_hY_hZ_hIn the coordinate system, Euler angles Φ,
Θ, Ψ, the point X_h(I, j) ↑ = (X_h(I, j), Y
_h(I, j), Z_h(I, j))^TIs given by equation (5).
Is converted to a point X (i, j). Where M_x, M_yYou
And M_zIs X_h, Y_hAnd Z_hRotation around an axis
It is a matrix. These operations are performed by the coordinate conversion unit 37.
I do.

Further, since Θ and Ψ are easily derived from the three-dimensional hand direction vector V ↑, the Euler angle detector 41 performs this. The three-dimensional hand center of gravity O detected by the three-dimensional hand center-of-gravity detecting unit 35, the point X (i, j) ↑ transformed by the coordinate transformation unit 37, and the Euler angles Θ, 、 form the ( 6) Equation holds. This calculation is performed by the conversion unit 43.

However, the point X 'converted by the conversion unit 43
(I, j)} indicates a three-dimensional tip point in a certain sense. The three-dimensional tip point includes a point called an erroneous correspondence point which is generated due to an incorrect correspondence. Then, an erroneous corresponding point due to an erroneous correspondence relationship is removed, and X ′ (i,
j) From ↑, the remaining Euler angle Φ needs to be finally predicted.

Next, the certainty detection unit 49 will be described. Based on the assumption that the position of the tip of the extended finger is located at a certain distance from the center of gravity of the hand, a point X ′ (i, j) ↑ spatially exists as shown in Expression (7). A confidence factor C _s (i, j) is defined. Here, a, r ₀ and r ₁ are constants. In particular, a is a value obtained by the stereo image association certainty degree detecting device 39,
r ₀ and r ₁ are constants determined by the size of the hand and the like. The calculation shown in the expression (7) is performed by the multiplier 47. Therefore, the multiplier 47 may be provided in the certainty detection unit 49, but the calculation is clarified. For this reason, the multiplier 47 is shown in the figure.

What can be obtained by the expression (7) is limited to the distance between the center of gravity of the hand and the position of the tip of the finger.
Not all miscorrespondence points have yet been removed.

[0045]

(Equation 4)

FIG. 6 is a diagram for explaining the principle of the operation of the correspondence detection unit. Referring to FIG. 6, erroneous corresponding points are removed by one-dimensional limitation. If the tips of the fingers are on a three-dimensional surface, there are two types of projecting them on the binocular image. That is, as in the plane S1, a first aspect XO ₁ and a second aspect X opposite the one surface
And if the O _r are opposed, facing the first viewpoint XO ₁ is one surface as a plane S2, a second aspect X on the other side
_Or is opposed. If the plane S1 is the tip mutual relationship of the finger in the image 61 and image _{62, n (x 1 (i)} ↑ T) ↑ · n (v 1 ↑ T) ↑ and n
^{_{(X r (j) ↑ T}} ) ↑ · n (v r ↑ T) ↑ and there is a need to scale of the order consisting. On the other hand, in the case of the plane S2 is a _{n (x 1 (i) ↑} T) ↑ · n (v 1 ↑ T) ↑, -
_{n (x r (j) ↑} T) ↑ · n (v r ↑ T) ↑ and it is necessary to become the order from. Where n ( ^* ) is a vector ^*
Indicates a normalized vector. For example, five and four finger tips are extracted from each image, and n (x ₁ (i)
{ ^T ) ↑ · n (v ₁ ↑ ^T )} and n ( _xr (j)
^{↑ T) ↑ · n (v} r ↑ T) ↑ of i according to the size, the table is created that corresponds to j, it as shown in Table 1 2
Two typical types of elimination tables P _{m, 1} (i, j) and P
_{m, 2} (i, j) is defined. The correspondence detection section 51 creates these removal tables.

[0047]

[Table 1]

Then, the posture detecting section 53 compares the created removal table P _{m, k} (i, j), the confidence C _d (i, j) and the confidence C _s (i, j) with the (8) ) To define the k-th erroneous correspondence removal table P _{, k} (i, j). here,
The multiplier 48 is illustrated in the same manner as the multiplier 47. Furthermore, the optimal erroneous correspondence removal table P _{m, opt} (i, j) is the (9)
Since it is obtained as in the equation, it is created by the optimum mis-correspondence removal table creating unit 59. Here, c is a constant, and w ′ is a vector predicted by the AR / MA model, and w ′ on this frame reflects the posture of the previous hand.
Is specified. Further, assuming that the time factor is t, Expression (10) is obtained. W on the next frame is the first (1
Expected by equation 1). Where a _l (l = 1,...,
n), b _l (l = 0,..., r) and G are constants.

The final posture w is the first (1) having the maximum eigenvalue.
2) It is obtained by M eigenvectors defined by the equation.

[0050]

(Equation 5)

The block diagram shown in FIG. 4 also corresponds to a data flow diagram relating to these methods.
As described above, the multipliers 47 and 48 are illustrated.

Next, a two-dimensional hand feature is extracted from the input image when the two objects are translated and the hands are rotated about the axis of the arm, and the hand predicted based on the feature is extracted. The experimental results of collecting the postures of the subjects will be described.

FIGS. 7 and 8 are diagrams for explaining the experimental results. In particular, FIG. 7 is a diagram illustrating the input left image, the features on the left image, the features on the right image, and the posture of the hand. Further, in FIG. 7, the left image, the feature on the left image, the feature on the right image, and the posture of the hand are arranged for each frame in order from the left. FIG. 8 (a)
FIG. 8B is a diagram showing the rotation trajectory of the actual Euler angles Φ, Ψ, Θ, and FIG. 8B shows the measured Euler angles Φ ′,
FIG. 8C is a diagram showing the rotation orbits of Ψ ′ and Θ ′, and FIG.
Is a diagram showing the trajectory of the position (X _O , Y _O , Z _O ) of the center of gravity of the hand. 8A and 8B, the horizontal axis represents the number corresponding to the frame, and the vertical axis represents the Euler angle d.
It is displayed in egree units. In FIG. 8C, the horizontal axis indicates the frame number, and the vertical axis indicates the XYZ coordinate system in cm.

Referring to FIGS. 7 and 8, the input left image arranged on the left side is detected as a left feature amount arranged on the right side. The left feature amount indicates the position of the tip of the finger, the direction of the hand, the center of gravity of the hand, and the like. Similarly,
Although not shown, the right feature amount is obtained from the right image.
Finally, the posture of the hands arranged on the rightmost side is detected.

[0055] of the hand posture, X'Y'Z axis and the X _h and Y
Visual perception is facilitated by overlapping the wings created by the _h- axis. The wing shows a hand that rotates about the _Xh axis from 0 to 180 ° and restores by reverse rotation. 7th, 10th, 18th
In the 2nd and 21st frames, the hand posture prediction is stable despite significant occlusion. However, in the eighteenth frame, the appearance of the tip of the finger is erroneously detected, and it is clear that the occlusion assumption adversely affects the detected hand posture.

Next, with reference to FIGS. 8A to 8C, the actual Euler angles Θ and が have an average angle of 0. Throughout the rotation gesture around the X _h-axis, the standard deviation is 6.2 and 6.8 °, the maximum error, were respectively 9.7 and 14.9 °. Φ changed its value slowly even if occlusion occurred. The predicted angles show a similar trend. Furthermore, the continuity is improved. Therefore,
By accurately detecting the expected angle in this way,
The utility of this device in systems has been proven.

Next, regarding the movement of the center of gravity of the three-dimensional hand, the maximum error was 2.2 cm throughout this operation. It has been confirmed that the maximum three-dimensional feature point detection error is a maximum of 1.2 cm. Therefore, it is considered that the present apparatus can sufficiently remove erroneous corresponding points.

The hand posture detecting device guaranteed for such experimental results is particularly useful for recognizing hand gestures. The feature is that even if occlusion occurs, the posture of the hand is predicted by the erroneous correspondence correspondence removal table using the three-dimensional prediction model. Further, this device is useful for constructing a virtual environment.

Although the posture detecting device has been described as an application example of the stereo image associative certainty detecting device shown in the embodiment, the present invention is not limited to this, but is applied to the field of shape recognition using image processing. On the other hand, the associating certainty detecting device is useful.

[0060]

As described above, according to the present invention, the first
A first image feature point on the first image in the left field of view from the first viewpoint, a first straight line passing through the first viewpoint, and a second line on the second image in the right field of view from the second viewpoint. The first image feature point and the second image feature point can be associated with each other by the distance between the second image feature point and the second straight line passing through the second viewpoint. It is possible to easily perform correspondence in a situation where rearrangement cannot be performed.

[Brief description of the drawings]

FIG. 1 is a diagram for describing a principle required for a stereo image correspondence certainty degree detection device according to an embodiment of the present invention.

FIG. 2 is a schematic block diagram of a stereoscopic image association certainty detecting device according to an embodiment of the present invention;

FIG. 3 is a diagram for explaining a principle required for a hand posture detection device to which a stereo image association certainty detection device according to an embodiment of the present invention is applied;

FIG. 4 is a schematic block diagram of a hand posture detection device to which the stereo image correspondence certainty degree detection device according to one embodiment of the present invention is applied;

FIG. 5: X _h Y _h Z _h coordinate system fixed to center of gravity of hand and XY
FIG. 3 is a diagram illustrating a relationship with a Z world coordinate system.

FIG. 6 is a diagram for explaining the principle of operation of the correspondence detection unit.

FIG. 7 is a first diagram for explaining an experimental result.

FIG. 8 is a second diagram for explaining an experimental result.

FIG. 9 is a diagram for explaining an epipolar line.

FIG. 10 is a diagram for explaining rearrangement along an epipolar line.

[Explanation of symbols]

 39 Stereo image association certainty degree detection device 63 First straight line calculation part 64 Second straight line calculation part 65 Distance calculation part 67 Confidence degree determination part

Continued on the front page (72) Haruo Takemura Inventor Haruo-cho, Soraku-gun, Kyoto 5th Sanraya, Inaya small-sized character

Claims (6)

(57) [Claims] 1. A first image feature point on a first image in a left field of view from a first viewpoint and a second image feature point on a second image in a right field of view from a second viewpoint. A stereo image associating certainty degree detecting device for detecting a certainty degree with respect to the correspondence relationship, wherein a first straight line passing through the first viewpoint and the first image feature point is calculated. A straight line calculation means, a second straight line calculation means for calculating a second straight line passing through the second viewpoint and the second image feature point, and a first straight line and the second straight line Distance calculation means for calculating a distance; and certainty determination for determining the certainty regarding the correspondence between the first image feature point and the second image feature point according to the calculated distance. And a means for ascertaining the degree of association between stereo images. 2. The method according to claim 1, wherein the certainty determining unit determines the certainty based on a function representing the certainty, the function being inversely proportional to the calculated distance. The stereoscopic image association certainty detection device according to claim 1. 3. The stereo image associating certainty degree detecting apparatus according to claim 2, wherein the function is constant below a predetermined value. 4. A first image feature point on a first image in a left field of view from a first viewpoint and a second image feature point on a second image in a right field of view from a second viewpoint. A stereo image associative certainty level detecting method for detecting a certainty factor with respect to the correspondence relationship, wherein a first straight line passing through the first viewpoint and the first image feature point is calculated. And a second step of calculating a second straight line passing through the second viewpoint and the second image feature point; and calculating a distance between the first straight line and the second straight line. A third step of determining the degree of certainty with respect to the correspondence between the first image feature point and the second image feature point according to the calculated distance. And a stereo image correspondence certainty degree detection method. 5. The method according to claim 4, wherein the fourth step includes a step of determining the certainty factor based on a function representing the certainty factor, the function being inversely proportional to the calculated distance. Method for detecting the degree of certainty between stereo images. 6. The stereo image associating certainty detecting method according to claim 5, wherein the function is constant below a predetermined value.