JP2010282437A - Object recognition device, object recognition method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image recognition device for performing calculation, with data amount that can be stored in a primary cache or a secondary cache of a CPU, necessary for image recognition. <P>SOLUTION: The object recognition device includes: an evaluation value calculating unit for associating respective second feature points of a recognition target object with first feature points of a registered image and acquiring COS and SIN values by angle differences θ of normal vectors of the first and second feature points for each first feature point; a candidate coordinate point computing unit for acquiring candidate coordinate points from rotated displacement vectors and the second feature points; a COS values integrating unit for projecting each of the candidate coordinate points to coordinate axes, corresponding to the projected axis coordinate points, and integrating the COS values as a COS integration value; a SIN value integrating unit for projecting each of the candidate coordinate points to a plurality of coordinate axes, corresponding to the projected axis coordinate points, and integrating the SIN values as a SIN integration value; an integration value adding unit for acquiring a total value of the COS and SIN integration values of respective axis coordinate points in combination of respective axis coordinate points; and a recognition target object determining unit for determining the agreement between the recognition target object and the registered image when the total value exceeds a threshold value. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an object recognition apparatus that detects a position, posture (rotation angle), and scaling factor of a reference object registered in advance in an input image.

As a detection method for detecting a position of a reference object registered in advance (position coordinates (x, y) in a two-dimensional coordinate system), a rotation angle θ, a scaling factor S, and the like from an input image input from an imaging device or the like, correlation The method, the Fourier descriptor method, the moment method, the Hough transform method, etc. are well known.
In any of the detection methods described above, the feature points of the registered image of the registered reference object are extracted and registered in advance.
Then, using the feature points of the recognition target object extracted from the input image and the feature points of the registered image registered, the position of the image corresponding to the registered image in the input image, and the attitude to the registered image And a scaling factor.

In the detection method using the feature points described above, the Hough transform method is based on a simple majority rule called voting.
Therefore, it has robust performance in the recognition target object recognition process against partial occlusion due to contact or overlap of parts of the recognition target objects in the input image, image noise, and the like.
In this Hough transform method, there is a generalized Hough transform method as a method for detecting a recognition target object having an arbitrary shape at high speed (for example, see Non-Patent Document 1).

In the above generalized Hough transform method, in the feature point (pixel unit) extracted from the reference object, the tangent at the feature point or the angle of the normal line, and the position of the reference point serving as the basis of the feature point and the object Prepare the relationship as a reference table. This reference point is an arrangement reference point indicating the arrangement position of the reference object on the two-dimensional coordinates.
That is, as shown in the flowchart of FIG. 9 for generating the reference table of the registered image as a model, the image of the registered image is input (step S1), and the feature point (P) of the reference object as shown in FIG. Is extracted by the second derivative (Laplacian) method or the Moravec method (step S2). That is, a feature point is a point on a contour line where a difference in value (pixel value) from surrounding pixels is large or a point where the surface state changes greatly.

  Then, binarization processing of feature points in the registered image is performed using a preset threshold value, and binarized data is output (step S3). As the binarization method performed here, the threshold when the inter-class variance becomes maximum when the histogram is divided into two classes with a certain threshold, including a simple binarization method with a preset threshold, is set to 2. Otsu method as a threshold value ("Automatic threshold selection method based on discrimination and least-squares criteria" (Otsu), IEICE Transactions, Vol. J63-D, No. 4. pp. 349-356, 1980), or a binarization method for setting a threshold value according to local density for an image having gradation.

Next, the storage area of the reference table that stores data of each feature point of the registered image is initialized (step S4), and the reference point of the reference object is obtained as an arbitrary position, for example, the barycentric position of the reference object (step S5). ).
Processing for obtaining feature point data consisting of a displacement vector from a feature point (start point) to the reference point (end point) and a unit normal vector (normalized normal vector) at the feature point is performed on all binarized data. It is determined whether or not it has been performed (step S6). When the feature point data acquisition process for all binarized data is completed, the process ends, and the binarization without the process for acquiring the feature point data is performed. If there is data, the process proceeds to step S7.

Unprocessed binarized data is read (step S7), and whether or not the binarized data is a feature point is determined based on the threshold value (step S8). If it is not a feature point, the process proceeds to step S6. If it is a feature point, the process proceeds to step S9.
Then, a tangent or unit normal vector at the feature point (simply referred to as a point in the figure) is obtained, and the process proceeds to step S6 (step S9).
For example, in step S9, the normal vector N1 at the point P1 is obtained, the normal vector at the point P2 is obtained, and the normal vectors are similarly obtained at other points. These normal vectors are standardized as unit normal vectors so that the lengths thereof are the same, and as shown in FIG. 11, x corresponding to the feature point numbers and constituting the unit normal vectors at the feature points A reference table of the component NXn (an integer of 1 ≦ n ≦ M, M is the number of feature points), the y component NYn, and the x component DXn and the y component DYn constituting the displacement vector Dn at the feature point is generated.

The image recognition process according to Non-Patent Document 1 using the reference table is performed by the operation of the flowchart shown in FIG.
An input image is input from an imaging device or the like (step S11), and as shown in FIG. 13, a feature point (Q) of the recognition target object in the input image is extracted by the same method as the reference image (step S12). In FIG. 13, the feature point of the registered image is superimposed on the feature point of the recognition target object, the angle of the normal vector at the feature point of the registered image is matched, and the position of the reference point indicated by the displacement vector of the registered image is adjusted. It is a conceptual diagram which shows performing voting.
Then, the binarization processing of the feature points in the input image is performed using a preset threshold value similarly to the processing of the reference image, and binarized data is output (step S13).

Next, the storage area of the voting space (two-dimensional plane showing the coordinates of the reference point for each rotation angle in the conceptual diagram of the voting space in FIG. 14) for voting on the reference point candidates is initialized (step S14).
It is determined whether or not the processing for obtaining the feature point data including the displacement vector from the reference point to the feature point and the unit normal vector at the feature point has been performed for all the binarized data (step S15). ) When the feature point data acquisition process for all binarized data has been completed, the process ends. When there is binarized data for which feature point data acquisition processing has not been performed, the process proceeds to step S16. .

Unprocessed binarized data is read (step S16), and whether or not the binarized data is a feature point is determined based on the threshold value (step S17). If it is not a feature point, that is, smaller than the threshold value. The process proceeds to step S15, and if it is a feature point, that is, if it is equal to or greater than the threshold value, the process proceeds to step S18.
And the unit normal vector of a feature point is calculated | required and a process is advanced to step 19 (step S18).
It is determined whether or not the reference point voting process corresponding to the selected feature point of the recognition target object has been performed using all the feature points of the registered image (step S19), and the selected feature of the recognition target object is determined. If the voting process using all the feature points of the registered image is completed for the point, the process proceeds to step S15. On the other hand, if there is a feature point of the registered image that is not used for the voting process, the process is performed. Proceed to step S20.

  The feature point of the registered image is superimposed on the feature point of the recognition target object, and the angle difference (rotation angle θ) between the unit normal vector at the feature point of the recognition target object and the unit normal vector at the feature point of the registration image Is obtained (step S20).

Next, the displacement vector of the feature point of the registered image is rotated by the following equation about the feature point by the rotation angle θ (step S21).
A COS value and a SIN value are obtained from the unit normal vector at the feature point of the recognition target object and the unit normal vector at the feature point of the registered image. For example, when it is assumed that the point P1 in FIG. 10 is the point Q2 in FIG. 13, it can be obtained by the following equation (1).
COS = Nx1 x nx2 + Ny1 x ny2
SIN = Nx1 × ny2−Ny1 × nx2 (1)
Here, the unit normal vector of the registered image is (Nx1, Ny1), and the unit normal vector of the recognition target object is (nx2, ny2).
Here, since the displacement vector is also rotated with respect to the feature point by the rotation angle θ, the components dx and dy of the rotated displacement vector can be obtained by the following equation (2).
dx = COS × DX−SIN × DY
dy = SIN × DX + COS × DY (2)
Here, the displacement vector of the registered image is (DX, DY).

The candidate position of the reference point is obtained by adding the rotated displacement vector to the position of the feature point of the recognition target object in the input image as shown in the following equation (step S22).
If the coordinates of the point Q2 are (Qx, Qy) according to the above-described equation, the reference point candidate of the recognition target object is obtained by the following equation (3).
Rx = Qx + dx
Ry = Qy + dy (3)

Then, “1” is added to the position (Rx, Ry) of the reference point candidate with respect to the coordinates of the two-dimensional plane corresponding to the rotation angle θ, that is, voting is performed (step S23).
As described above, in the method of Non-Patent Document 1, a displacement vector indicating the angle of each normal line of a registered image and a recognition target object at each feature point (the rotation angle at this time becomes a posture) is shown. Vote for the location of the reference point.
As a result, when the superimposed feature points correspond to the registered image and the recognition target object, when the normal vector angle is adjusted, the displacement vector of the registered image points in the direction of the same point, Since the reference points overlap, it is possible to detect the position where the number of votes is the peak as the position of the reference point of the recognition target object, and to obtain the rotation angle at that time as the posture with respect to the registered image. By this process, the recognition process between the registered image and the recognition target object in the input image is performed very efficiently.

However, considering the dimensions of this voting space, even if the scaling factor S is not taken into consideration, as shown in FIG. 14, the data structure is vertical position (vertical coordinate), horizontal position (horizontal coordinate), and rotation angle. 3D. In addition, when the size change due to the scaling factor is also taken into consideration in the object detection, one dimension is added and becomes four dimensions.
For this reason, when applied to actual object recognition, a very large amount of memory is required to construct three-dimensional and four-dimensional voting spaces. In the following description, evaluation is performed on a three-dimensional voting space excluding the dimension of the scaling factor.

In the case of this three-dimensional voting space, the number of bits “NX × NY × NA × NV” is required in the method of Non-Patent Document 1. Here, when the position of the recognition target object in the input image in the two-dimensional coordinates and the detection result of the rotation angle with respect to the registered image are shown, the number of divisions of the detection position of the x-axis (horizontal axis) of the two-dimensional coordinates (corresponding to the resolution) The number of coordinate points NX, the number of coordinate points NY that is the number of divisions of the detection position of the y-axis (vertical axis), the angle division NA that is the number of divisions of the detection angle of the rotation angle, and the number of bits NV of the memory that accumulates the number of votes (Normally 2 bytes or 4 bytes).
That is, in Non-Patent Document 1, a large amount of memory capacity is required because each vote corresponds to each position and angle of a recognition target object candidate.
In order to reduce this large memory capacity, there is a method of simply integrating the COS value and the SIN value corresponding to the rotation angle θ of the object detection candidate for each position of the object detection candidate (see, for example, Patent Document 1).

Image recognition processing using Patent Document 1 is performed by the operation of the flowchart shown in FIG.
An input image is input from an imaging device or the like (step S31), and as shown in FIG. 13, feature points of the recognition target object in the input image are extracted by the same method as the reference image (step S32).
Then, the binarization processing of the feature points in the input image is performed using a preset threshold value similarly to the processing of the reference image, and binarized data is output (step S33).

Next, the storage area of the voting space for voting on the reference point candidates (each two-dimensional plane for accumulating COS values and SIN values described later) is initialized (step S34).
It is determined whether or not the process of integrating the COS value and SIN value at each feature point with respect to the position of the reference point candidate is performed for all binarized data (step S35). When the COS value and SIN value integration process for the binarized data is completed, the process ends. When there is binarized data that has not been subjected to the COS value and SIN value integration process, the process proceeds to step S36. Proceed to

Unprocessed binarized data is read (step S36), and whether or not the binarized data is a feature point is determined based on the threshold value (step S37). If it is not a feature point, the process proceeds to step S35. If it is a feature point, the process proceeds to step S38.
And the unit normal vector of a feature point is calculated | required and a process is advanced to step 39 (step S18).
Using all the feature points of the registered image, it is determined whether or not the COS value and the SIN value have been added to the reference point corresponding to the selected feature point of the recognition target object (step S39). When the integration process using all the feature points of the registered image is completed for the selected feature point of the object, the process proceeds to step S35, while there are feature points of the registered image that are not used for the integration process. If it exists, the process proceeds to step S40.

Based on the unit normal vector (nx, ny) of the feature point of the recognition target object in the input image and the unit normal vector (Nx, Ny) of the feature point of the registered image, A COS value and a SIN value are obtained (step S40).
COS = Nx × nx + Ny × ny
SIN = Nx × ny−Ny × nx (4)

The displacement vector (dx, dy) rotated by the rotation angle θ from the COS value and SIN value obtained by the equation (4) and the displacement vector (DX, DY) of the feature point of the registered image that has been superimposed is expressed as follows: Obtained by equation (5).
dx = COS × DX−SIN × DY
dy = SIN × DX + COS × DY (5)

Then, from the feature point (Qx, Qy) of the recognition target object in the input image and the rotated displacement vector (dx, dy), the position of the reference point position candidate is obtained by the following equation (6).
Rx = Qx + dx
Ry = Qy + dy (6)
With respect to the candidate position (Rx, Ry) of the obtained reference point position, COS (θ) and SIN (θ) are respectively represented as COS (θ) voting space and SIN as shown in FIG. It accumulates in the voting space of (θ).
As a result, the memory capacity as the voting space is obtained by “NX × NY × 2 × (number of bits for integration)”. Since the number of bits to be integrated is 2 to 4 in both cases, when the angle division number NA is 360 and the resolution is 1 degree, the method of Patent Document 1 is 1 / It can be realized with a memory capacity of 180. Here, 2 indicates two two-dimensional coordinates for integrating the COS value and the SIN value, respectively.

By this voting process, when the candidate point of the reference point of the recognition target object is a true reference point having a rotation angle θ with respect to the registered image, from all of the feature points of the registered image with respect to the candidate point. It can be considered that the majority of the votes of the rotation angle θ are. For this reason, each of the integrated values of the COS value and the SIN value at the point that is the true reference point is determined by the vote of the rotation angle θ of the recognition target object, and the COS corresponding to the rotation angle θ. Value, an integer multiple of the SIN value, and the number of votes. As a result, a peak can be obtained in one or both of the COS integrated value and the SIN integrated value.
Therefore, the peak position of the COS integrated value and the SIN integrated value is the reference point of the recognition target object in the input image, and the ratio between the integrated value of the COS value integrated to the reference point and the integrated value of the SIN value, that is, tan ^{− The} rotation angle θ can be obtained from ¹ (integrated value of SIN value / integrated value of COS value).

On the other hand, if the reference point candidate point is not a true reference point, the vote at the feature point of the recognition target object may be a false vote or may be caused by noise, and the voted rotation angle θ is uniformly random. It is assumed to have a distribution.
As a result, the integration of the COS value and the SIN value at a point that is not the true reference point is an integral multiple of the COS function and the SIN function, each of which is simply integrated from 0 degrees to 360 degrees with a uniform weight. It is expected that the integrated value will approach zero.

JP 2007-128374 A

D.H. Ballard, “Generalizing The Hough Transform To Detect Arbitrary Shapes”, Pattern Recongition, Vol. 13, no. 2, pp. 111-122, 1981

However, in the object recognition method shown in Patent Document 1, the memory capacity used in the calculation process is reduced by two or more digits compared to Non-Patent Document 1, but the memory capacity is still large when performing high-speed calculations. It has many drawbacks.
The calculation speed of the CPU (central processing unit) is extremely high, and the processing capability is improved, so that the speed difference from the memory element of the main storage device that stores data is large.
In order to shorten the access time to the memory of the main storage device, recent CPUs incorporate a high-speed memory such as a primary cache or a secondary cache.
CPU calculation is performed efficiently when the data and instructions stored in the cache are valid (when a cache hit occurs).
On the other hand, if the cache does not hit, that is, if the data and instructions stored in the cache are not valid, they need to be read from the memory of the main storage device, which is relatively slow relative to the cache, and the CPU will perform at its original performance. In contrast, the operation speed becomes inferior.

The capacity of the cache memory is several tens of kilobytes in the primary cache, several hundred kilobytes in the secondary cache, and several megabytes in the tertiary cache.
In Non-Patent Document 1, when the coordinate point of a two-dimensional coordinate is 1024 × 1024, the resolution of the rotation angle θ is 1 degree, and the number of bytes necessary for voting is 4, it is “1024 × 1024 × 360 × 4”. A memory capacity of about 1.5 Gbytes is required.
Further, in Patent Document 1, when the two-dimensional coordinate point is 1024 × 1024, the COS value and the SIN value are 2, and the number of bytes to be integrated is 4 bytes, “1024 × 1024 × 2 × 4” is obtained. It will be about 8M bytes.
This latter memory capacity is a capacity that cannot be stored in the primary cache and the secondary cache and can be stored in the tertiary cache. Further, if the image size is increased, the coordinate points need to be increased, and it becomes impossible to store them in the tertiary cache.

  The present invention has been made in view of such circumstances, and an object of the present invention is to provide an image that can perform calculations necessary for image recognition with a data amount that can be stored in a primary cache or a secondary cache of a CPU. To provide a recognition device.

The present invention has been made to solve the above-described problem, and an object recognition apparatus according to the present invention includes a plurality of first feature points extracted from a registered image and a first normal line at the first feature points. A feature point data storage unit for storing feature point data including a vector and a displacement vector from the first feature point to the reference point of the registered image; and a second that constitutes the feature of the recognition target object in the input image A feature point extracting unit for extracting feature points; a normal vector computing unit for obtaining a second normal vector at the second feature point; and a first feature point superimposed on the second feature point, An evaluation value calculation unit for obtaining COS (θ) and SIN (θ) values based on the angle difference θ between the normal vector and the second normal vector, and rotating the displacement vector from the angle difference to rotate the displacement vector And the coordinates of the second feature point, A candidate coordinate point calculation unit for obtaining a candidate coordinate point that is a complement, and projecting the candidate coordinate point onto each of the plurality of first coordinate axes, and corresponding to the projected axis coordinate point, COS (θ ) As a COS integrated value, and the candidate coordinate points are projected onto a plurality of second coordinate axes, respectively, and SIN (θ) at the candidate coordinate points corresponding to the projected axis coordinate points. SIN value integration unit for integrating the SIN integration value, and the COS integration value and SIN integration value of the axis coordinate points corresponding to the candidate coordinate points in the first coordinate axis and the second coordinate axis are added to obtain a total value. An integrated value addition unit to be obtained;
A recognition target object determination unit that determines that a recognition target object that matches the registered image is arranged with the candidate coordinate point as a reference point when the total value exceeds a preset threshold value; And
According to the object recognition method of the present invention, the registered image data generation unit uses a plurality of first feature points extracted from the registered image, a first normal vector at the first feature points, and a first feature point. A process of storing feature point data including a displacement vector to a reference point of the registered image in a feature point data storage unit, and a second feature point in which the feature point extraction unit constitutes a feature of the recognition target object in the input image , A process in which the normal vector calculation unit obtains a second normal vector at the second feature point, and an evaluation value calculation unit calculates the first registered image for each of the second feature points. A process of obtaining COS (θ) and SIN (θ) values for each first feature point by matching all of the feature points of the first and second normal vectors with an angle difference θ between the first normal vector and the second normal vector; The candidate coordinate point calculation unit rotates the displacement vector from the angle difference. A process for obtaining candidate coordinate points that are candidates for the reference point from the rotated displacement vector and the coordinates of the second feature point, and the COS value integrating unit respectively projects the candidate coordinate points onto a plurality of coordinate axes, A process of integrating COS (θ) at the candidate coordinate points as COS integrated values corresponding to the projected axis coordinate points of each coordinate axis, and a SIN value integrating unit respectively adding the candidate coordinate points to the plurality of coordinate axes. A process of projecting and integrating SIN (θ) at the candidate coordinate points as SIN integrated values corresponding to the projected coordinate points of the coordinate axes, and an integrated value adding unit for each coordinate coordinate point on each of the coordinate axes In the combination, the process of obtaining the total value by adding the COS integrated value and the SIN integrated value at each axis coordinate point, and the recognition target object determination unit, when the total value exceeds a preset threshold, the registration One with the image Recognition target object that is characterized by having a process determining as being arranged the candidate coordinate point as the reference point.
The program according to the present invention includes a plurality of first feature points extracted from a registered image by a registered image data generation unit, a first normal vector at the first feature points, and the registration from the first feature points. Processing for storing feature point data consisting of a displacement vector to a reference point of an image in a feature point data storage unit, and a feature point extraction unit extracting a second feature point that constitutes a feature of the recognition target object in the input image A process for calculating a second normal vector at the second feature point, and an evaluation value calculator for the first feature of the registered image for each of the second feature points. Processing for associating all of the points and obtaining COS (θ) and SIN (θ) values for each first feature point by the angle difference θ between the first normal vector and the second normal vector, and candidate coordinates The point calculator rotates the displacement vector from the angle difference. , A process for obtaining candidate coordinate points that are candidates for the reference point from the rotated displacement vector and the coordinates of the second feature point, and a COS value integrating unit that projects the candidate coordinate points onto a plurality of coordinate axes, Corresponding to the coordinate point of each coordinate axis, a process of integrating COS (θ) at the candidate coordinate point as a COS integrated value, and a SIN value integrating unit respectively projecting the candidate coordinate point onto the plurality of coordinate axes Then, corresponding to the projected axis coordinate point of each coordinate axis, a process of integrating SIN (θ) at the candidate coordinate point as a SIN integrated value, and an integrated value adding unit, for each axis coordinate point on each of the coordinate axes, In the combination, a process for obtaining the total value by adding the COS integrated value and the SIN integrated value at each axis coordinate point, and the recognition target object determination unit, when the total value exceeds a preset threshold, the registered image Matches That the recognition target object is a program for causing a computer to function as the object recognition device having and a process for determining is arranged the candidate coordinate point as the reference point.
According to the present invention, instead of voting for the COS (θ) and SIN (θ) voting space for each candidate coordinate point that is a candidate for the reference point of the arrangement in the input image of the recognition target object, the candidate coordinate point is represented as a coordinate axis. A memory for storing the integrated value in order to project COS (θ) and SIN (θ) for each projected axial coordinate point and add the accumulated results at the candidate coordinate points corresponding to each axial coordinate point. By changing the part from a two-dimensional array to a one-dimensional array, the memory capacity can be reduced compared to the conventional one. As a result, the cache memory can be used effectively and the recognition process can be speeded up. Can do.

In the object recognition apparatus of the present invention, the coordinate axes are an x-axis and a y-axis in a two-dimensional plane.
As a result, a total of four one-dimensional arrays of storage units for storing the integrated values of COS (θ) and SIN (θ) corresponding to the axis coordinate points, two in the x-axis direction and two in the y-axis direction. Therefore, the memory capacity can be reduced as compared with the conventional example in which a storage unit is required for each coordinate point of the two-dimensional coordinates.

In the object recognition apparatus of the present invention, the coordinate axes are three axes, that is, an x-axis and a y-axis on a two-dimensional plane, and a linear axis composed of a straight line represented by an x-axis coordinate point and a y-axis coordinate point. Features.
As a result, the one-dimensional array of storage units for storing the integrated values of COS (θ) and SIN (θ) corresponding to the axis coordinate points is two in the x-axis direction, two in the y-axis direction, and the linear axis direction. Therefore, the memory capacity can be reduced as compared with the conventional example in which a storage unit is required for each coordinate point of two-dimensional coordinates.
In addition, when there are a plurality of reference points, an artifact may occur. However, by using three axes for projecting two-dimensional coordinate points, the position identification ability is improved compared to the case of two. It is possible to improve the above-described artifacts, to eliminate the artifacts, and to improve the object recognition processing efficiency.

The object recognition apparatus according to the present invention is characterized in that the integrated value adding unit squares and adds the COS integrated value and the SIN integrated value at the axial coordinate point of each coordinate axis to obtain the total value.
Since the integrated values corresponding to the axis coordinates of a plurality of axes are added, the S / N ratio at the time of peak value detection can be improved, and the accuracy of reference point detection can be improved.

In the object recognition device of the present invention, the integrated value adding unit adds a numerical value obtained by squaring the added value of the COS integrated value at the axis coordinate point of each coordinate axis and a numerical value obtained by squaring the added value of the SIN integrated value. The total value is obtained.
Since the integrated values corresponding to the axis coordinates of a plurality of axes are added, the S / N ratio at the time of peak value detection can be improved, and the accuracy of reference point detection can be improved.
Further, since the integrated value is averaged when noise is present, it is possible to further improve the S / N ratio compared to the case where the above-described COS integrated value and SIN integrated value are squared and added. Become.

It is a block diagram which shows the structural example of the object recognition apparatus by 1st Embodiment and 2nd Embodiment of this invention. It is a conceptual diagram which shows a response | compatibility with an input image and the accumulation | storage memory | storage part row | line | column provided in the x-axis and y-axis of this input image, and demonstrates operation | movement of 1st Embodiment. FIG. 4 is a conceptual diagram showing an arrangement configuration of an x-axis x-axis COS value accumulation storage unit, an x-axis SIN value accumulation storage unit, a y-axis y-axis COS value accumulation storage unit, and a y-axis SIN value accumulation storage unit. It is a flowchart which shows the operation example of the object recognition of the object recognition apparatus by 1st Embodiment. It is a conceptual diagram explaining the operation | movement of the object recognition by 1st Embodiment. It is a conceptual diagram explaining the operation | movement of the object recognition by 1st Embodiment. It is a conceptual diagram explaining the operation | movement of the object recognition by 2nd Embodiment. x-axis x-axis COS value integration storage unit, x-axis SIN value integration storage unit, y-axis y-axis COS value integration storage unit, y-axis SIN value integration storage unit, A-axis A-axis COS value integration storage unit, A It is a conceptual diagram which shows the arrangement | sequence structure of an axis | shaft SIN value integrating | accumulating memory | storage part. It is a flowchart explaining operation | movement of extraction of the 1st feature point of a registration image, and extraction of the feature point data in this 1st feature point. It is a conceptual diagram explaining extraction of feature point data. It is a conceptual diagram which shows the table structure in which feature point data are memorize | stored. 10 is a flowchart showing an operation of image recognition processing using generalized Hough transform of Non-Patent Document 1. FIG. 13 is a conceptual diagram illustrating reference point voting processing in FIG. 12. It is a conceptual diagram explaining the memory capacity required in the case of the image recognition in the generalized Hough conversion of a nonpatent literature 1. 10 is a flowchart for explaining the operation of image recognition processing in Patent Document 1. It is a conceptual diagram explaining the integration | accumulation process of the COS value and SIN value in patent document 1. FIG.

<First Embodiment>
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a schematic block diagram showing a configuration example of an object recognition apparatus according to the first embodiment of the present invention.
In this figure, the object recognition apparatus according to the present embodiment has a registered image data generation unit 1, a feature point data storage unit 2, and an object recognition unit 3.
The registered image data generation unit 1 extracts feature points extracted from the registered image as an array of first feature points, sets a reference point of the registered image, obtains a displacement vector from the first feature point to the reference point, The first unit normal vector at the first feature point is obtained, and the coordinates of the first feature point, the displacement vector, and the first unit normal vector are used as the first feature point data. Corresponding to the feature point number which is the identification information of the feature point, each first feature point is written and stored in the feature point data storage unit 2. The reference point is an arrangement reference point indicating the arrangement position of the registered image and the reference object on the two-dimensional coordinates. The reference point may be an arbitrary point on the two-dimensional coordinates where the registered image is arranged. For example, the center of gravity of the registered image may be used as the reference point.
The feature point data storage unit 2 stores the first feature point data of each registered image corresponding to the feature point number of the first feature point according to the table configuration shown in FIG.
The object recognizing unit 3 obtains feature points extracted from the recognition target object in the input image as an array of second feature points, obtains a first unit normal vector at the first feature points, and extracts from the feature point data storage unit 2 The first feature point data of the registered image is sequentially read out, the first feature point is sequentially associated with the second feature point, and the similarity between the recognition target object and the registered image is determined.

The registered image data generation unit 1 includes a registered image acquisition unit 11, a feature point extraction unit 12, a feature point binarization unit 13, and a normal vector extraction unit 14. The operation of the registered image data generation unit 1 is the same as the operation of the flowchart of FIG. 9 described in Patent Document 1.
The registered image acquisition unit 11 inputs image data of a registered image used for recognizing a recognition target object later from the imaging device or CAD.
The feature point extraction unit 12 extracts the first feature point (pixel unit) of the registered image using the second-order differential (Laplacian) method or the Morabek method described in the feature extraction in Non-Patent Document 1. .

  The feature point binarization unit 13 performs binarization based on the gradation of pixels of the first feature point of the registered image and pixels other than the first feature point. For example, the feature point binarization unit 13 sets the gradation level of the pixel of the first feature point to “1” and the gradation level of the pixels other than the first feature point to “0” based on a threshold value of the gradation level set in advance. And Here, as the preset threshold value, a threshold value for performing the binarization processing described in the generalized Hough transform method is used.

The normal vector extraction unit 14 performs vertical and horizontal differential processing on the gradation of the pixels of the original registered image corresponding to each obtained first feature point, and performs horizontal processing based on the gradation change. A normal vector having the x-axis component in the direction and the y-axis component in the vertical direction as vectors is obtained. This differentiation process is performed by a first-order differential filter such as a Prewitt filter or a Sobel filter. Multiplying the filter coefficients by the gradation levels of nine pixels (up, down, left, and right) centered on the pixel of the feature point of interest, the results are summed to obtain an x-axis component or a y-axis component.
In addition, the normal vector extraction unit 14 normalizes the obtained normal vector composed of the x-axis component and the y-axis component so that the length becomes 1, and outputs the normal vector as a first unit normal vector.
Further, the normal vector extraction unit 14 extracts a displacement vector from each first feature point to the reference point, and uses the coordinates of the first feature point, the displacement vector, and the first unit normal vector as the first feature. As point data, it is written and stored in the feature point data storage unit 2.

Next, the object recognition unit 3 includes an input image acquisition unit 31, a feature point extraction unit 32, a feature point binarization unit 33, a normal vector extraction unit 34, an evaluation value calculation unit 35, a candidate coordinate point calculation unit 36, a COS. A value integration unit 37, a SIN value integration unit 38, an integration value addition unit 39, and a recognition target object determination unit 40 are provided.
The input image acquisition unit 31 inputs image data of an input image including a recognition target object from an imaging device such as a camera.
The feature point extraction unit 32 extracts the coordinate point of the second feature point (pixel unit) of the recognition target object in the input image using the same method as the feature point extraction unit 12.

The feature point binarization unit 33 performs binarization based on the gradation of pixels of the second feature point of the recognition target object and pixels other than the second feature point. For example, the feature point binarization unit 33 sets the gradation levels of the pixels of the second feature point to “1” and the gradation levels of the pixels other than the second feature point to “0” based on a threshold value of the gradation level set in advance. And Here, a numerical value similar to that of the feature point binarization unit 13 is used as the preset threshold value.
The normal vector extraction unit 34 uses the same method as the normal vector extraction unit 14 to calculate the horizontal x-axis component and the vertical y-axis component corresponding to the obtained second feature points. Find the normal vector as a vector.

The evaluation value calculation unit 35 associates the first feature point with the second feature point, the first unit normal vector (nx, ny) of the first feature point, and the second feature point second. COS (θ) and SIN (θ) based on the angular difference (rotation angle) θ with respect to the unit normal vector (Nx, Ny) are determined as evaluation values by the equation (4) already described.
The candidate coordinate point calculation unit 36 uses the equation (5) to set the first normal vector to the first normal vector so that the direction of the first normal vector is the same as the second normal vector. Rotate around a coordinate point.
Then, the candidate coordinate point calculation unit 36 uses the rotated displacement vector and the coordinates (Px, Py) of the second feature point, using the equation (6), the candidate position (Rx) of the reference point of the recognition target object. , Ry).

As shown in FIG. 2, the COS value integration unit 37 includes a number of x-axis COS value integration storage units (ΣCOS (θ)) corresponding to the number of pixels (image width) in the x-axis direction of the input image, and the input image. The number of y-axis COS value integration storage units (ΣCOS (θ)) corresponding to the number of pixels in the y-axis direction (image width) is provided. That is, for each axis coordinate corresponding to each pixel, an x-axis COS value integration storage unit and a y-axis COS value integration storage unit are provided. FIG. 2 is a conceptual diagram showing the correspondence between the candidate position (Rx, Ry) and the axis coordinates on which the candidate position (Rx, Ry) is projected.
Further, as shown in FIG. 3, the COS value integrating unit 37 projects the reference point candidate position (Rx, Ry) for which the COS (θ) value is calculated on the x-axis, and this projected axis coordinate point ( The COS (θ) is integrated (ΣCOS (θ)) in the x-axis COS value integration storage unit corresponding to Rx). Also, the COS value integrating unit 37 projects the reference point candidate position (Rx, Ry) from which the COS (θ) value is calculated onto the y axis, and the y axis corresponding to the projected axis coordinate point (Ry). The COS value accumulation storage unit accumulates the COS (θ) (ΣCOS (θ)). That is, as shown in FIG. 3, a total of four one-dimensional integrated value storage units are provided, two corresponding to the x-axis and two corresponding to the y-axis.

As shown in FIG. 2, the SIN value integration unit 38 includes a number of x-axis SIN value integration storage units (ΣSIN (θ)) corresponding to the number of pixels in the x-axis direction (image width) of the input image, and the input image. The number of y-axis SIN value integration storage units (ΣSIN (θ)) corresponding to the number of pixels in the y-axis direction (image width) is provided. That is, for each axis coordinate corresponding to each pixel, an x-axis SIN value integration storage unit and a y-axis SIN value integration storage unit are provided. FIG. 2 is a conceptual diagram showing the correspondence between the candidate position (Rx, Ry) and the axis coordinates on which the candidate position (Rx, Ry) is projected.
Further, as shown in FIG. 3, the SIN value integrating unit 38 projects the reference point candidate position (Rx, Ry) for which the SIN (θ) value is calculated on the x-axis, and this projected axis coordinate point ( The SIN (θ) is integrated (ΣSIN (θ)) in the x-axis SIN value integration storage unit corresponding to Rx). Also, the SIN value integrating unit 38 projects the reference point candidate position (Rx, Ry) for which the SIN (θ) value is calculated on the y axis, and the y axis corresponding to the projected axis coordinate point (Ry). The SIN (θ) is integrated into the SIN value integration storage unit (ΣSIN (θ)).

The integrated value adder 39 stores the square of the COS integrated value (ΣCOS (θ)) stored in the x-axis COS value integrated storage unit at the same axis coordinate on the x-axis and the x-axis SIN value integrated storage unit. The square of the SIN integrated value (ΣSIN (θ)) is added to calculate an x-axis total value for each axis coordinate of the x-axis.
Further, the integrated value adding unit 39 stores the square of the COS integrated value (ΣCOS (θ)) stored in the y-axis COS value integrated storage unit at the same axis coordinate on the y-axis and the y-axis SIN value integrated storage unit. The sum of squared SIN integrated values (ΣSIN (θ)) is added to calculate a y-axis total value for each axis coordinate of the y-axis.
Further, the integrated value adding unit 39 adds the x-axis total value and the y-axis total value for each axis coordinate, and calculates the total value at each coordinate of the two-dimensional coordinates.
That is, by adding the x-axis total value of the axis coordinate Rx and the y-axis total value of the axis coordinate Ry, the calculation process of the total value of (Rx, Ry) of the two-dimensional coordinates is performed on the axis corresponding to the x-axis pixel. The total value at each coordinate point of the two-dimensional coordinates of all the pixels in the input image is calculated by performing all the combinations of the coordinates and the axis coordinates corresponding to the y-axis pixels (processing of back projection on the two-dimensional plane).
The total value obtained here is obtained by the following equation.
Total value = (x-axis (ΣCOS (θ)) ² ) + (x-axis (ΣSIN (θ)) ² )
+ ((Y-axis (ΣCOS (θ)) ² ) + (y-axis (ΣSIN (θ)) ² )

The recognition target object determination unit 40 extracts a coordinate point where the total value exceeds a preset peak value as a reference point on which the registered image is arranged, and as a reference point on which the recognition target object that matches the registered image is arranged. Output.
The ratio of the integrated value of the integrated value and SIN (theta) of the recognition target object determination unit 40 has been credited to the reference point COS (theta), i.e. ^{tan -1 (ΣSIN (θ) /} ΣCOS (θ)) Thus, the rotation angle θ can be obtained. ΣSIN (θ) and ΣCOS (θ) used here are stored in, for example, the x-axis COS value integration storage unit and the x-axis SIN value integration storage unit corresponding to the axial coordinates of the position where the reference point is projected on the x-axis. COS integrated value and SIN integrated value.
This peak value is actually measured by experiment, and is set from a value that statistically takes the value of the total value of the reference points when the recognition target object matches the registered image.
Further, when it is necessary to know in detail the position of the recognition target object, the recognition detail detection unit 41 connected to the object recognition unit 3 uses the reference point and the rotation angle θ extracted by the recognition target object determination unit 40 as follows. A process of extracting in detail the position where the recognition target object is arranged is performed. Therefore, if it is not necessary to extract the detailed arrangement position of the recognition target object, it is not necessary to provide the detail detection unit 41.

Further, the integrated value adding unit 39 described above may be configured as follows.
The integrated value adding unit 39 is a COS integrated value of the x-axis COS value integrated storage unit of each axis coordinate (Rx) in the x-axis and a COS integrated of the y-axis COS value integrated storage unit of each axis coordinate (Ry) in the y-axis. The COS total value in the two-dimensional coordinates (Rx, Ry) is obtained by performing back projection on the two-dimensional plane for all combinations of the x-axis coordinate and the y-axis coordinate.
Similarly, the integrated value adding unit 39 includes the SIN integrated value of the x-axis SIN value integrated storage unit for each axis coordinate (Rx) on the x-axis and the y-axis SIN value integrated storage unit for each axis coordinate (Ry) on the y-axis. The total SIN value of the two dimensional coordinates (Rx, Ry) is added to the SIN total value, and the back projection onto the two-dimensional plane is performed for all combinations of the x-axis coordinate and the y-axis coordinate. Ask and do.
Further, the integrated value adding unit 39 adds the squared value of the COS total value and the squared value of the SIN total value in the corresponding two-dimensional coordinates (Rx, Ry), and adds the value at each coordinate of the two-dimensional coordinates. Calculate the total value.
The total value obtained here is obtained by the following equation.
Total value = ((x-axis (ΣCOS (θ))) + (y-axis (ΣCOS (θ)))) ²
+ ((X-axis (ΣSIN (θ))) + (y-axis (ΣSIN (θ)))) ²

Next, the recognition process between the recognition target object and the registered image in this embodiment will be described with reference to FIG. FIG. 4 is a flowchart for explaining an operation example of the recognition processing.
The input image acquisition unit 31 inputs an input image from an imaging device (not shown), and outputs it to the feature point extraction unit 32 and the normal vector extraction unit 34 as image data expressed by pixels, for example (step S51).
When the input image is input, the feature point extraction unit 12 extracts the second feature point of the recognition target object in the input image, and uses the extracted input image (the image made the second feature point) as the feature point 2. It outputs to the value conversion part 33 (step S52).
Next, the feature point binarization unit 33 binarizes each pixel of the input image to be input into a second feature point pixel and a pixel other than the second feature point. (Step S53).

The COS value integration unit 37 initializes the x-axis COS value integration storage unit and the y-axis COS value integration storage unit, that is, sets the integration value to “0”.
Similarly, the SIN value integration unit 38 initializes the x-axis SIN value integration storage unit and the y-axis SIN value integration storage unit, that is, sets the integration value to “0” (step S54).
Next, the evaluation value calculation unit 35 determines whether or not the COS value and the SIN value as evaluation values at the feature points have been calculated for the binarized data of all pixels of the input image (step S55) When the evaluation values at the feature points are calculated for the binarized data of all the pixels, the recognition process is terminated, while there are pixels for which the evaluation values at the feature points are not calculated. The process is output to step S56.
Here, the detection of the pixel for which the evaluation value is not calculated at the feature point is performed by, for example, fixing the y-axis axis coordinates and sequentially processing the pixels corresponding to the x-axis axis coordinates. When the processing on the x-axis axis coordinates corresponding to the axis coordinates is completed, the y-axis axis coordinates are changed to sequentially process the pixels corresponding to the x-axis axis coordinates. Thus, the end point can be detected by forming a loop of the combination of the x-axis coordinate and the y-axis coordinate according to the number of pixels on the x-axis and the y-axis.

Next, the evaluation value calculation unit 35 reads the unprocessed pixel binarization data from the feature point binarization unit 33 (step S56), and the read binarization data exceeds a preset threshold value. It is determined whether or not this pixel is the second feature point (step S57).
At this time, if the read pixel is the second feature point, the evaluation value calculation unit 35 advances the process to step S58. If the pixel is not the second feature point, the evaluation value calculation unit 35 advances the process to step S55.
When the pixel is the second feature point, the normal vector extraction unit 34 obtains a second unit normal vector at the second feature point, and advances the process to step S59 (step S58).
Then, the evaluation value calculation unit 35 calculates the evaluation value from the first unit normal vector of the second feature point and the second normal vector of the registered image, and performs the calculation of the evaluation value for all the second images of the registered image. It is determined whether or not it has been performed between the feature points (step S59). At this time, if the evaluation value is calculated between all the second feature points of the registered image, the evaluation value calculating unit 35 advances the process to step S55, and the evaluation value is calculated for the registered image. If not performed with all the second feature points, the process proceeds to step S60.

Next, the evaluation value calculation unit 35 reads the first unit normal vector of the first feature point that is not used for calculating the evaluation value from the feature point data, and the second unit method of the second feature point. From the line vector (nx, ny) and the first unit normal vector of the first feature point, the evaluation values of COS (θ) and SIN (θ) are calculated using equation (4) (step S60). ).
Then, the candidate coordinate point calculation unit 36 uses the equation (5) to rotate the displacement vector (Dx, Dy) at the rotation angle θ, and calculates the rotated displacement vector (dx, dy) (step S61). ).
Further, the candidate coordinate point calculation unit 36 calculates the coordinates (Rx) of the reference point candidate from the above-described displacement vector (dx, dy) and the coordinates (Px, Py) of the first feature point by the equation (6). , Ry) is calculated (step S62).

Next, the COS value integration unit 37 projects the coordinates (Rx, Ry) onto the x axis, and stores the COS (θ) in the x axis COS value integration storage unit corresponding to the axis coordinate point (Rx) of the x axis. In addition, the coordinates (Rx, Ry) are projected onto the y-axis, and the COS (θ) is accumulated in the y-axis COS value accumulation storage unit corresponding to the y-axis axis coordinate point (Ry).
Similarly, the SIN value integrating unit 38 projects the coordinates (Rx, Ry) on the x axis, and stores the SIN (θ) in the x axis SIN value integrating storage unit corresponding to the x coordinate point (Rx) of the x axis. At the same time, the coordinates (Rx, Ry) are projected onto the y-axis, and the SIN (θ) is accumulated in the y-axis SIN value accumulation storage unit corresponding to the y-axis axis coordinate point (Ry) (step S63). .

Through the processing described above, the x-axis COS value integration storage unit corresponding to the x-axis axis coordinate, the y-axis COS value integration storage unit corresponding to the y-axis axis coordinate, and the x-axis SIN value integration corresponding to the x-axis axis coordinate For the storage unit and the y-axis SIN value integration storage unit corresponding to the axis coordinate of the y-axis, the second unit normal vector of each second feature point of the recognition target object and all the first units of the registered image The evaluation value with the normal vector is integrated.
The integrated value adding unit 39 calculates an x-axis total value for each axis coordinate of the x-axis, calculates a y-axis total value for each axis coordinate of the y-axis, and calculates an x-axis total value and a y-axis total. The value is added for each axis coordinate, and the total value at each coordinate of the two-dimensional coordinate is calculated.
The recognition target object determination unit 40 extracts a coordinate point where the total value exceeds a preset peak value as a reference point on which the registered image is arranged, and as a reference point on which the recognition target object that matches the registered image is arranged. , And the rotation angle θ obtained from the COS integrated value and the SIN integrated value.

As described above, in this embodiment, the x-axis COS value integration storage unit and the x-axis COS value integration storage unit corresponding to the one-dimensional axis coordinate on the x-axis, and the one-dimensional axis coordinate on the y-axis are used. Evaluation values including the rotation angle θ are stored in the corresponding y-axis COS value integration storage unit and y-axis COS value integration storage unit. As already described, evaluation values at one axis coordinate are integrated in each integrated value storage unit.
Here, when the memory capacity is estimated under the same conditions as in Patent Document 1, the calculation formula is “1024 × 4 × 4”, and the memory capacity is 16 kbytes. Compared to about 8 Mbytes in Patent Document 1, this memory capacity is reduced by two digits. As a result, the CPU can be accommodated in the primary cache of the CPU, and the CPU can perform arithmetic processing without causing a reduction in processing speed due to occurrence of a cache miss.
In the above calculation formula, “1024” is the number of pixels in the x-axis direction and the y-axis direction (corresponding to the number of axis coordinates), the next “4” is the capacity of each integrated value storage unit, and the next “4” "Is the number of integrated value storage units arranged one-dimensionally.

<Second Embodiment>
Next, an object recognition apparatus according to a second embodiment of the present invention will be described. The object identification device according to the second embodiment has the same configuration as that of the first embodiment. Only the different processing points will be described below.
First, in the first embodiment, a description will be given of erroneous detection when there are a plurality of reference points (hereinafter, described as two points). In the first embodiment, only one peak in the total value, that is, when a reference point exists, or as shown in FIG. 5, two reference points are on any axis (x-axis or y-axis). If they exist in parallel, the reference point and each axis coordinate have a one-to-one correspondence, so that no erroneous detection occurs.

  However, as shown in FIG. 6, when the reference points are not arranged parallel to any axis, the integrated value is stored in two integrated value storage units on the x axis as an integrated value that exceeds the peak value. And the integrated values stored in the two integrated value storage units exist on the y-axis. As a combination of the two axis coordinates of the x axis and the two axis coordinates of the y axis, when the back projection onto the two-dimensional plane is performed to obtain the coordinates of the reference point, the coordinates indicated by the black circle where the reference point actually exists are The coordinates indicated by white circles of false reference points called artifacts are mixed. In this case, although the artifact can be deleted by the detail detection unit 41, since the calculation process is required for the artifact detection and deletion process, the detection time becomes long.

Therefore, in the second embodiment, as shown in FIG. 7, in addition to the arrangement of the one-dimensional integration storage units in the x-axis direction and the y-axis direction, the A-axis in the direction of the angle φ different from the x-axis and y-axis A one-dimensional integration storage unit (A-axis COS value integration storage unit, A-axis SIN value integration storage unit) for the direction is provided in the direction of the angle φ. The A-axis is provided to face the x-axis and the y-axis in the quadrant of the two-dimensional plane where the recognition target object is arranged. That is, the A axis is provided at an arbitrary position on a two-dimensional plane as a projection axis for projecting coordinates (Rx, Ry) in addition to the x axis and the y axis, as shown in FIG. The axis coordinates on the A axis are obtained from “Z = Rx × COS (φ) + Ry × SIN (φ)”. An A-axis COS value integration storage unit that integrates COS (θ) and an A-axis SIN value integration storage unit that integrates SIN (θ) corresponding to the axis coordinate Z on which the two-dimensional coordinates (Rx, Ry) are projected. Will be provided. That is, as shown in FIG. 8, an array of six one-dimensional integrated value storage units is provided, two for the x axis, two for the y axis, and two for the A axis. It has been.
The difference is that the COS value integrating unit 37 and the SIN value integrating unit 38 project not only the x axis and the y axis but also the A axis as described above, and the projected axis coordinate is projected. The evaluation value is integrated in the corresponding integration storage unit. Hereinafter, this operation will be described.

The COS value integrating unit 37 projects the coordinates (Rx, Ry) on the x axis, integrates COS (θ) in the x axis COS value integrating storage unit corresponding to the axis coordinate point (Rx) of the x axis, and The coordinates (Rx, Ry) are projected onto the y-axis, the COS (θ) is accumulated in the y-axis COS value accumulation storage unit corresponding to the y-axis axis coordinate point (Ry), and the coordinates (Rx, Ry) Is projected onto the A axis, and the COS (θ) is accumulated in the A axis COS value accumulation storage unit corresponding to the axis coordinate point (Z) of the A axis.
Similarly, the SIN value integrating unit 38 projects coordinates (Rx, Ry) onto the x axis, and integrates SIN (θ) into the x axis SIN value integrating storage unit corresponding to the x coordinate axis (Rx). At the same time, the coordinates (Rx, Ry) are projected onto the y-axis, the SIN (θ) is accumulated in the y-axis SIN value accumulation storage unit corresponding to the y-axis axis coordinate point (Ry), and the coordinates (Rx , Ry) is projected onto the A axis, and the above SIN (θ) is accumulated in the A axis SIN value accumulation storage unit corresponding to the axis coordinate point (Z) of the A axis.

Then, the integrated value adding unit 39 calculates an x-axis total value for each axis coordinate of the x-axis, calculates a y-axis total value for each axis coordinate of the y-axis, and A of each axis coordinate of the A-axis The total axis value is calculated, the total x-axis value, the total y-axis value, and the total Z-axis value are added for each axis coordinate, and the total value at each coordinate of the two-dimensional coordinates is calculated. That is, for each two-dimensional coordinate of a combination of the x-axis coordinate and the y-axis coordinate, the x-axis total value of the corresponding x-axis axis coordinate, the y-axis total value of the y-axis axis coordinate, and the A-axis Are added together to calculate the total value of the coordinate points of each two-dimensional coordinate.
The recognition target object determination unit 40 extracts a coordinate point where the total value exceeds a preset peak value as a reference point on which the registered image is arranged, and as a reference point on which the recognition target object that matches the registered image is arranged. , And the rotation angle θ obtained from the COS integrated value and the SIN integrated value.
Then, the recognition target object determination unit 40 extracts a coordinate point where the total value exceeds the preset peak value as a reference point on which the registered image is arranged, and a reference on which the recognition target object that matches the registered image is arranged. Output as a point.
As described above, by providing the A-axis different from the x-axis and the y-axis, the total value is calculated from the sum of the three-axis integrated values, so that the coordinates of the artifact can be eliminated.

In the present embodiment, the x-axis COS value integration storage unit and the x-axis COS value integration storage unit corresponding to the one-dimensional axis coordinate on the x-axis, and the y corresponding to the one-dimensional axis coordinate on the y-axis. A rotation angle θ with respect to the axis COS value integration storage unit and the y-axis COS value integration storage unit, and the A-axis COS value integration storage unit and the A-axis COS value integration storage unit corresponding to the one-dimensional axis coordinate on the A-axis. The evaluation value including is stored. As already described, evaluation values at one axis coordinate are integrated in each integrated value storage unit.
Here, when the memory capacity is estimated under the same conditions as in Patent Document 1, the calculation formula is “1024 × 4 × n”, and the memory capacity is 4 × nk bytes. Compared to about 8 Mbytes in Patent Document 1, this memory capacity is reduced by two digits as in the first embodiment. As a result, the CPU can be accommodated in the primary cache of the CPU, and the CPU can perform arithmetic processing without causing a reduction in processing speed due to occurrence of a cache miss.
In the above calculation formula, “1024” is the number of pixels in the x-axis direction and the y-axis direction (corresponding to the number of axis coordinates), the next “4” is the capacity of each integrated value storage unit, and the next “n” "Is the number of integrated value storage units arranged one-dimensionally.

  In addition, the program for realizing the function of the object recognition apparatus in FIG. 1 is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed to execute object recognition. Processing may be performed. Here, the “computer system” includes an OS and hardware such as peripheral devices.

Further, the “computer system” includes a homepage providing environment (or display environment) if the WWW system is used.
The “computer-readable recording medium” refers to a storage device such as a flexible disk, a magneto-optical disk, a portable medium such as a ROM and a CD-ROM, and a hard disk incorporated in a computer system. Furthermore, a “computer-readable recording medium” dynamically holds a program for a short time, like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. It also includes those that hold programs for a certain period of time, such as volatile memory inside computer systems that serve as servers and clients in that case. The program may be a program for realizing a part of the above-described functions, or may be a program that can realize the above-described functions in combination with a program already recorded in a computer system.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design and the like within a scope not departing from the gist of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Registered image data generation part 2 ... Feature point data storage part 3 ... Object recognition part 11 ... Registered image acquisition part 12, 32 ... Feature point extraction part 13, 33 ... Feature point binarization part 14, 34 ... Normal vector Extraction unit 31 ... Input image acquisition unit 35 ... Evaluation value calculation unit 36 ... Candidate coordinate point calculation unit 37 ... COS value integration unit 38 ... SIN value integration unit 39 ... Integration value addition unit 40 ... Recognition target object determination unit 41 ... Detailed detection Part

Claims (7)

Feature point comprising a plurality of first feature points extracted from the registered image, a first normal vector at the first feature point, and a displacement vector from the first feature point to the reference point of the registered image A feature point data storage unit for storing data; and
A feature point extraction unit for extracting a second feature point constituting the feature of the recognition target object in the input image;
A normal vector computing unit for obtaining a second normal vector at the second feature point;
All the first feature points of the registered image are made to correspond to each of the second feature points, and COS (θ) and SIN (θ based on the angle difference θ between the first normal vector and the second normal vector ) An evaluation value calculation unit for obtaining a value for each first feature point;
A candidate coordinate point calculation unit that performs a rotation of a displacement vector from the angle difference and obtains a candidate coordinate point that is a candidate of a reference point from the rotated displacement vector and the coordinates of the second feature point;
A COS value integrating unit that projects the candidate coordinate points onto a plurality of coordinate axes, and integrates COS (θ) at the candidate coordinate points as COS integrated values corresponding to the projected coordinate points of the coordinate axes;
A SIN value integrating unit that projects the candidate coordinate points onto the plurality of coordinate axes, and integrates SIN (θ) at the candidate coordinate points as a SIN integrated value corresponding to the projected coordinate points of the coordinate axes;
An integrated value adding unit for adding a COS integrated value and a SIN integrated value at each axis coordinate point to obtain a total value in a combination of each axis coordinate point in each of the coordinate axes;
A recognition target object determination unit that determines that a recognition target object that matches the registered image is arranged with the candidate coordinate point as a reference point when the total value exceeds a preset threshold value. An object recognition device.   The object recognition apparatus according to claim 1, wherein the coordinate axes are an x-axis and a y-axis in a two-dimensional plane.   2. The coordinate system according to claim 1, wherein the coordinate axes are three axes of an x-axis and a y-axis in a two-dimensional plane, and a linear axis composed of a straight line represented by an x-axis coordinate point and a y-axis coordinate point. Object recognition device. The integrated value adding unit is
The object recognition apparatus according to any one of claims 1 to 3, wherein the integrated value is obtained by squaring and adding a COS integrated value and a SIN integrated value at an axial coordinate point of each coordinate axis. The integrated value adding unit is
The total value is obtained by adding a numerical value obtained by squaring the addition value of the COS integrated value at the axial coordinate point of each coordinate axis and a numerical value obtained by squaring the added value of the SIN integrated value. The object recognition apparatus according to claim 3. A registered image data generation unit extracts a plurality of first feature points extracted from the registered image, a first normal vector at the first feature points, and a first feature point to a reference point of the registered image. Storing feature point data consisting of displacement vectors in a feature point data storage unit;
A process of extracting a second feature point constituting a feature of the recognition target object in the input image by the feature point extraction unit;
A process in which the normal vector computing unit obtains a second normal vector at the second feature point;
The evaluation value calculation unit associates all of the first feature points of the registered image with each of the second feature points, and determines the COS (COS by the angle difference θ between the first normal vector and the second normal vector). obtaining θ) and SIN (θ) values for each first feature point;
A candidate coordinate point calculation unit rotates a displacement vector from the angle difference, and obtains a candidate coordinate point that is a reference point candidate from the rotated displacement vector and the coordinates of the second feature point;
A process in which the COS value integrating unit projects the candidate coordinate points onto a plurality of coordinate axes, and integrates COS (θ) at the candidate coordinate points as COS integrated values corresponding to the projected axis coordinate points of the coordinate axes. When,
The SIN value integrating unit projects the candidate coordinate points onto the plurality of coordinate axes, and integrates SIN (θ) at the candidate coordinate points as SIN integrated values corresponding to the projected axis coordinate points of the coordinate axes. Process,
An integrated value adding unit adding a COS integrated value and a SIN integrated value at each axis coordinate point to obtain a total value in a combination of each axis coordinate point in each of the coordinate axes;
A process in which a recognition target object determination unit determines that a recognition target object that matches the registered image is arranged with the candidate coordinate point as a reference point when the total value exceeds a preset threshold value. An object recognition method characterized by the above. A registered image data generation unit extracts a plurality of first feature points extracted from the registered image, a first normal vector at the first feature points, and a first feature point to a reference point of the registered image. Processing for storing feature point data consisting of displacement vectors in a feature point data storage unit;
A process of extracting a second feature point constituting a feature of the recognition target object in the input image by the feature point extraction unit;
A process in which the normal vector computing unit obtains a second normal vector at the second feature point;
The evaluation value calculation unit associates all of the first feature points of the registered image with each of the second feature points, and determines the COS (COS by the angle difference θ between the first normal vector and the second normal vector). θ) and SIN (θ) values for each first feature point;
A candidate coordinate point calculation unit rotates a displacement vector from the angle difference, and obtains a candidate coordinate point that is a reference point candidate from the rotated displacement vector and the coordinates of the second feature point;
A process in which the COS value integrating unit projects the candidate coordinate points onto a plurality of coordinate axes, and integrates COS (θ) at the candidate coordinate points as COS integrated values corresponding to the projected axis coordinate points of the coordinate axes. When,
The SIN value integrating unit projects the candidate coordinate points onto the plurality of coordinate axes, and integrates SIN (θ) at the candidate coordinate points as SIN integrated values corresponding to the projected axis coordinate points of the coordinate axes. Processing,
An integrated value adding unit that calculates a total value by adding the COS integrated value and the SIN integrated value at each axis coordinate point in a combination of each axis coordinate point in each of the coordinate axes;
A recognition target object determination unit that determines that a recognition target object that matches the registered image is arranged with the candidate coordinate point as a reference point when the total value exceeds a preset threshold value. A program for causing a computer to function as an object recognition device.

Images