JPH0636036A - Holonic visual recognizing device - Google Patents

Abstract

PURPOSE:To provide the holonic visual recognizing device which recognizes a curved line figure by using line end information on segments in N-azimuths detected from the curved line figure as a meaning information. CONSTITUTION:This visual recognizing device consists of an information input part 101 which composes an image reception surface of pixels arranged in a simple square lattice, an information generation part 102 which has N (N: natural number) detection azimuths and generates the line end information corresponding to partial line information from the image information of the information input part, and a meaning storage part 103 which arranges relative positions in the detection directions and has plural couples of line end information in the same azimuth and opposite directions. The gradients of ridges of a regular inverted 2N-angled pyramid whose apexes are arranged in the detection azimuths are regarded as a weight function used to information interchanging between the information generation part 102 and meaning storage part 103.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a holonic system between semantic information groups created from input graphic information such as characters / images (still images, moving images) and stored semantic information groups of graphics. The present invention relates to a visual recognition device that performs pattern recognition by associating (that is, associating an information loop including top-down and bottom-up of semantic information).

[0002]

2. Description of the Related Art A model that captures the human visual system in three layers, the retina (and the lateral geniculate body), the primary visual cortex, and the memory function, and uses the non-linear oscillator pull-in phenomenon to process information is known. Holovision (Reference: Hiroshi Shimizu, Yoko Yamaguchi: "Cerebrum Information Principle and Its Application to Bio-Computers", Biological Science Vol.37, No.1, pp.26-40, 1986)
Is a visual recognition model known as.

A basic model of holographic vision is shown in FIG. As shown in FIG. 5, the structure of holographic vision is composed of three layers of an information input unit 301 (R plane), an information generation unit 302 (V layer), and a meaning storage unit 303 (S layer). Retina (and lateral geniculate body) of the visual system, first
Corresponds to the next visual cortex and memory function. Information input unit 301
Is an image receiving surface in which image receiving pixels are arranged in a simple square lattice (4 × 4 pixel is an example in the figure), and binary pixel information (also called raw information, black circles ● indicate signals) 300
The graphic information composed of is input to the pixel. The information generation unit 302 detects the azimuth of a line (the azimuth indicated by \, |, /,-in the figure) from the array of input elementary information and the S unit 312 group and the line end (in the figure,

[0004]

[Outer 1]

It is composed of an assembly of T unit 322 groups for detecting the line end displayed by. An aggregate 3 of units of the information generation unit 302 corresponding to one pixel of the information input unit 301
32 (hatched portion in the figure) is called a hyper column 332, and the basic column forming the hyper column 332 is shown in FIG.
S unit 312 for generating line azimuth information as shown in FIG.
And a T unit 322 composed of a TL unit 341 and a TR unit 342 for generating line end information in the arrow direction. Each unit forming the basic column is a non-linear oscillator. The hyper column 332 is a stack of basic columns having different orientations. Meaning storage unit 303
Means the meaning of the memory figure in the direction of the line end of each side (in the figure,

[0006]

[Outside 2]

A collection of M units 313 to be stored as relative positions (directions indicated by) and relative positions of line ends (relative positions indicated by L1, L2, L3, L4, L5, L6, L7, L8 in the figure). Consists of. The M unit is also a nonlinear oscillator. The meaning storage unit 303 can also correspond to the rotation of the figure by performing the conversion Li → Lj (i, j = 1 to 8; i ≠ j). The unit assembly 333 in different line end directions corresponding to one relative position of the meaning storage unit 303 (the hatched portion of L6 in the figure) is referred to as a memory column 333.

Next, the VS relay 304 in the figure has a function of converting the absolute position of the line end detected by the information generator 302 into a relative position. If the information input unit 301 is a holographic system having 4 × 4 pixels as an example, the operation of the VS relay is such that the weighting function shown in FIG. This is a process in which the maximum value is used as the signal at that array position. In this case, the weighting function corresponds to the slope of the ridge of an inverted regular octagonal pyramid whose apex coincides with the direction of line detection. Similarly, SV relay 3 in the figure
Reference numeral 05 has a function of converting the relative positions of the line ends stored in the meaning storage unit 303 into absolute positions. Information input section 3
If the 01 is a 4 × 4 pixel holvision, for example, the operation of the SV relay is to multiply the signal of the meaning storage unit by the weighting function as shown in FIG. 8, and to obtain the maximum value of the signal at the same array position. This is the process of setting the signal at the array position. The weighting function in this case also corresponds to the gradient of the ridgeline of an inverted regular octagonal pyramid whose apex coincides with the azimuth of line detection, but an interpolated gradient is used for the array position corresponding to the ridgeline. The position conversion function enables graphic recognition independent of position and size. As both of these relays operate at the same time, the information generation unit 302 and the meaning storage unit 303 are synchronized with respect to the line end information (the mutual pulling action of the collection of nonlinear oscillators) and desynchronized (of the nonlinear oscillators). Suppression of mutual aggregation) is performed, and the line end information group that represents the input figure (that is, the set of line end information that represents the relationship between the line ends) and the line edge information group that is stored as the figure meaning (That is, a set of line end information representing the relationship between the line ends) is associated with each other, and graphic recognition is established.

As described above, the set of line end information generated by the information generation unit from the line information of the graphic input to the information input unit is bottom-up to the meaning storage unit as the semantic information of the graphic (this is performed by the VS relay). ), And top-down the semantic information of the graphic stored in the meaning storage unit to the information generation unit (this is S
In the information loop formed by the V relay), the non-linear oscillator groups of the semantic storage unit and the information generation unit, which carry the semantic information, are in the pull-in state, that is, at the semantic information level of the input figure and the stored figure. With the configuration and principle of matching, it is possible to recognize figures regardless of position, size, or rotation.

Next, how the holographic vision recognizes a quadrangle will be described with reference to FIG. (1) First, a quadrangle having 3 dots on each side is input to the image receiving surface of the information input unit 301.

(2) In the information generating section 302, the S unit 312, which detects the direction of the line from the arrangement of the dots of the input information by the pulling action of the nonlinear oscillator, as shown in FIG.
Line information of the azimuths of and | is generated.

(3) Next, based on the line information generated by the S unit 312, the T unit 322 (specifically, the TR unit 342 and the TL unit 341 corresponding to the direction of the line end).
Generates line end information in the directions of ←, →, ↓, and ↑ by the pulling action of the nonlinear oscillator.

FIG. 10 shows a mechanism for generating line end information for horizontal line information. In the figure, the solid line represents mutual excitatory coupling, and the dotted line represents mutual inhibitory coupling. When line information indicated by diagonal lines is present on the image receiving surface 301, the shaded S unit 312 is excited, and the shaded TL unit 351 is excited to generate line end information.

(4) On the other hand, the quadrilateral meaning information stored in the M unit 313 of the meaning storage unit 303 is such that the arrows representing the relative position and direction of the line ends are set to the relative positions of L2, L4, L6 and L8. It is expressed by placing two of each. For example, L6 on the circumference of the figure is the memory column 333 of FIG.
That there is information in the direction of the arrow,
That is, it means that the nonlinear oscillator is excited.
The strength of information (that is, the amplitude strength of the nonlinear oscillator) is represented by the length of the arrow. Since this information set (that is, the semantic information represented by the eight arrows) is represented in the retracted state of the nonlinear oscillator, all the oscillator groups related to memory vibrate with the same phase.

(5) Semantic information group (that is, bottom-up semantic information group) generated from the line end defined by the absolute position generated in the above step (3), and generated in the above step (4) The semantic information group (that is, the top-down semantic information group) generated from the line end defined by the relative position is the SV relay and VS.
An information loop is formed by the relays, which are associated with each other by pulling in the non-linear oscillator, and figure recognition is established.

[0016]

However, in the above-mentioned horovision, the set of line end information obtained from the line information of the input figure is bottomed up as the semantic information of the figure, and the figure stored in the meaning storage section is stored. In the information loop formed by top-downing the semantic information, it is the principle of figure recognition by drawing in between the non-linear oscillator groups that carry the semantic information of both, so if it is as it is, the curves and A figure having a curved line (specifically, a circle, an ellipse, a semicircle, a crescent, etc.) cannot be recognized.

The object of the present invention is to pay attention to such a problem, and a curve or a figure having a curve (specifically, a circle, an ellipse, a semicircle, a crescent shape, etc.) to be inputted to the image receiving surface of the information input section. )
It is to provide a holonic visual recognition device that recognizes.

[0018]

In order to achieve the above-mentioned object, the present invention comprises an information input section forming an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions. An information generation unit that generates line end information corresponding to partial line information from the image information of the information input unit, and a plurality of sets of line end information in the same azimuth and opposite directions with relative positions arranged in the detection azimuth. A holonic visual recognition device comprising a meaning storage unit having, and a gradient of a ridgeline of an inverted regular 2N pyramid having vertices arranged in the detection direction as a weighting function used in information exchange between the information generation unit and the meaning storage unit, Alternatively, an information input unit that constitutes an image receiving surface with pixels arranged in a simple square lattice, and N
An information generation unit that has (N is a natural number) detection azimuths and generates line edge information corresponding to partial line information from the image information of the information input unit, and a relative position is arranged in the middle of the detection azimuths. The information generating unit includes a meaning storage unit having a plurality of pairs of line end information representing an angle smaller than 180 degrees, and a slope of an ridgeline of an inverted regular 2N pyramid having a vertex in the middle of the detection orientation. And a holonic visual recognition device that uses a weighting function for information exchange between the meaning storage unit and the above.

[0019]

With the above-described structure, the present invention makes it possible to recognize a curve or a figure having a curve (specifically, a circle, an ellipse, a semicircle, a crescent, etc.), and an apparatus for recognizing an image or a handwritten document. There are things with great practical value.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIGS. 1, 2, 3 and 4 are embodiments of the present invention for recognizing circles and ellipses. Hereinafter, description will be given with reference to the drawings. FIG. 1 is the same as the basic configuration of a horovision that detects four directions, and an information input unit 1 including an image receiving surface in which pixels are arranged in a simple square lattice (in the figure, an example of 4 × 4 pixels).
01 and the S unit 112 group for detecting the azimuth of the line (the azimuth displayed by \, |, /,-in the figure) from the arrangement of the input pixel information of the figure and the line end (in the figure,

[0021]

[Outside 3]

An information generating unit 102 composed of an assembly of T unit 122 groups for detecting the line ends displayed by, and the meaning of the memory graphic are the directions of the line ends of each side (in the figure,

[0023]

[Outside 4]

A collection of M units 113 to be stored as a relative position (direction indicated by) and a relative position of line ends (relative positions indicated by L1, L2, L3, L4, L5, L6, L7, L8 in the figure). The VS relay 104 multiplies the line end information generated by the information generation unit 102 by the gradient of the ridge of the inverted regular octagonal pyramid whose vertex matches the detected azimuth as a weighting function. The SV relay 105 is a first holonic visual recognition device that multiplies the line edge information of the meaning storage unit with the gradient of the ridgeline of the inverted regular octagonal pyramid whose vertex coincides with the detected azimuth as a weighting function and sends it to the information generation unit. .

In FIG. 2, a part of the basic structure of the holovision for detecting four directions is modified, and the information input section 10 is described above.
1 and the information generation unit 102 are the same, but the meaning storage unit 133 (this meaning storage unit 133 rotates the meaning storage unit 103 of the first holonic visual recognition device counterclockwise by [π / 4]). In this case, the VS relay 134 is also an aggregate of M units 143), and the VS relay 134 includes the ridgeline of the inverted regular octagonal pyramid in which the vertex is arranged in the middle of the detection direction in the line end information generated by the information generation unit 102. The gradient is multiplied as a weighting function and sent to the meaning storage unit 133, and the SV relay 1
Reference numeral 35 denotes a second holonic visual recognition device that multiplies the line edge information of the meaning storage unit with the gradient of the ridgeline of an inverted regular octagonal pyramid whose vertex is located in the middle of the detection direction as a weighting function and sends it to the information generation unit 102.

When the true circle shown in FIG. 3 (c-1) is input to the information input unit 101 of the holonic visual recognition device (not limited to the first and the second, it applies to both), the information is input. Since the image receiving surface of the input unit 101 is a simple square lattice array of pixels, line information that is coincident with the detection direction as shown in FIG. 3 (c-2) but is interrupted at the line end is the information generation unit. 102 is generated. In reality, since a part of the curve is detected as straight line information, the amplitude of the nonlinear oscillator of the line end information becomes small as shown in FIG. 3 (c-3), but no problem occurs.

When recognizing a circle with the first holonic visual recognition device shown in FIG. 1, as can be seen from FIG. 3 (c-3), the line end of the line segment detected from the circle is in the middle part of the detection direction. appear. Therefore, if the gradient of the ridge of an inverted regular octagonal pyramid whose apex coincides with the detection direction as shown in FIG. 3 (a) is used as a weighting function in the VS relay section, the line end information appearing in the middle section of the detection direction is It appears to the same degree in the detection direction, and
Line end information such as -2) is sent to the meaning storage unit 103.
In the line-end information of FIG. 3 (a-2), if one pays attention to the line-ends of the same direction but the same size and opposite directions corresponding to the tangent of the circumference, the line-end information is 8 sets in total (this is There are 2N sets when the detected azimuth is N), and this is the semantic information of a circular figure that never appears in a normal figure other than a curved figure. Therefore, the meaning storage unit 103 can store the meaning information of a circle as a set of 2N sets of line edges in the same direction, the same size, and opposite directions. As a result, if the gradient of the ridge of an inverted regular octagonal pyramid whose apex coincides with the detected azimuth is used as a weighting function in the SV relay unit, a set of 2N sets of the same azimuth, which are the same azimuth, which is the semantic information of the circle, and which have opposite directions Is given the possible absolute position information, the meaning storage unit 103 to the information generation unit 1
Sent to 02. A circular figure is recognized when the non-linear oscillator group of the information generation unit that carries the semantic information of the circle and the meaning storage unit is pulled in.

Also when recognizing a circle with the second holonic visual recognition device shown in FIG. 2, as can be seen from FIG. 3 (c-3),
The line ends detected from the circle appear in the middle of the detection direction. Therefore, as shown in Fig. 3 (b-1), the gradient of the ridge of an inverted regular octagonal pyramid whose vertex is in the middle of the detection direction is VS as the weighting function.
If used in the relay section, the line end information that appears in the middle of the detection direction will appear in the direction that the detection direction is rotated by [π / 4], and the line end information as shown in Fig. 3 (b-2) will appear in the meaning storage section. 133
Sent to. The line-end information in Fig. 3 (b-2) is a set of two that are convex outward and correspond to the vertex of a regular octagon inscribed in the circumference, are smaller than 180 degrees, and have the same size sandwiching the angle closest to 180 degrees. The line-end information (which will be referred to as "ku" -shaped line-end information for the sake of simplicity) is rotated by [π / N] degrees for each relative position while maintaining the outward convexity. There is. If we pay attention to this "ku" -shaped line end information, there will be a total of 8 sets of this line end information (this is 2N when the detection direction is N).
(Corresponding to a set), this is also the maximum regular N-gonal line edge information that can be detected when the detection direction is N on the image receiving surface of the simple square lattice. However, if N becomes large, it becomes difficult to distinguish between a regular N-gon and a circle, and it can be used as semantic information of the circle. Therefore, it becomes possible to store the meaning information of the circle in the meaning storage unit 133 as a set of 2N sets of "K" -shaped line end information. As a result, if the gradient of the ridge of an inverted regular octagonal pyramid whose apex is located in the middle of the detection direction in the SV relay part is used as a weighting function, a set of 2N sets of "ku" -shaped line end information, which is the semantic information of the circle, is used. Is provided with possible absolute position information and is sent from the meaning storage unit 133 to the information generation unit 102. A circular figure is recognized when the non-linear oscillator group of the information generation unit that carries the semantic information of the circle and the meaning storage unit is pulled in.

Next, when the ellipse shown in FIG. 4 (d-1) is input to the information input unit 101 of the holonic visual recognition device (which applies to both of the first and the second unless otherwise specified). Since the image receiving surface of the information input unit 101 is a simple square lattice array of pixels, line information which is coincident with the detection direction but is interrupted at the line end is information as shown in FIG. 4 (d-2). It is generated by the generation unit 102. Actually, since a part of the curve is detected as straight line information, the amplitude of the nonlinear oscillator of the line end information becomes small as shown in FIG. 4 (d-3), but no problem occurs.

When recognizing a circle with the first holonic visual recognition device shown in FIG. 1, as can be seen from FIG. 4 (d-3), the line end of the line segment detected from the circle is in the middle part of the detection direction. appear. Therefore, if the gradient of the ridge of the inverted regular octagonal pyramid whose apex coincides with the detection direction as shown in Fig. 4 (a-3) is used as the weighting function in the VS relay section, the line end information appearing in the middle section of the detection direction will be It appears to the same degree in the detection direction, and
The line end information such as -4) is sent to the meaning storage unit 103.
In the line-end information of FIG. 3 (a-2), if one pays attention to the line ends of the same azimuth and opposite directions corresponding to the tangents of the circumference, this line-end information has a total of 8 sets (the detection azimuth is N Then, it corresponds to 2N sets), and this becomes the semantic information of an elliptical figure that never appears in a normal figure other than a curved figure. Therefore, the meaning storage unit 103 can store the meaning information of an ellipse as a set of 2N sets of line edges in the same direction but in opposite directions. As a result, if the slope of the ridge of an inverted regular octagonal pyramid whose apex coincides with the detected azimuth is used as a weighting function in the SV relay part, a set of 2N sets of opposite azimuths, which is the semantic information of the ellipse, is possible. The position information is given and sent from the meaning storage unit 103 to the information generation unit 102. A circular figure is recognized when the non-linear oscillator group of the information generation unit that carries the semantic information of the circle and the meaning storage unit is pulled in.

Also when recognizing an ellipse by the second holonic visual recognition device shown in FIG. 2, as can be seen from FIG. 4 (d-3), the line end detected from the ellipse is the middle part of the detection direction. Appear in. Therefore, if the gradient of the ridge of an inverted regular octagonal pyramid whose vertex is in the middle of the detection direction as shown in Fig. 4 (b-3) is used as the weighting function in the VS relay section, the line end information appearing in the middle section of the detection direction can be detected. It appears in the direction in which the azimuth is rotated by [π / 4], and the line end information as shown in FIG. 4B-4 is sent to the meaning storage unit 133. The line end information in Fig. 4 (b-4) is a set of two pieces that are convex outward and correspond to the vertex of a regular octagon inscribed in the circumference, are smaller than 180 degrees, and have the same size sandwiching the angle closest to 180 degrees. The line end information (which will be referred to as the modified "ku" -shaped line end information for simplicity hereinafter) is rotated by [π / N] degrees for each relative position while maintaining the outward convexity. ing. Paying attention to this modified "ku" -shaped line end information, there are a total of 8 sets of this line end information (this corresponds to 2N sets when the detection direction is N), which is the image of a simple square lattice. It is also the line edge information of the maximum line symmetry N polygon which can be detected when the detection direction is N on the surface. However, if N becomes large, it becomes difficult to distinguish between an axisymmetric N-gon and an ellipse, and it can be used as semantic information of the ellipse. Therefore, it becomes possible to store the meaning information of the ellipse in the meaning storage unit 133 as a set of 2N sets of the deformed "ku" -shaped line end information. As a result, if the gradient of the ridgeline of the inverted regular octagonal pyramid whose apex is located in the middle of the detection direction in the SV relay section is used as a weighting function, the 2N sets of modified "ku" -shaped line end information, which is the semantic information of the ellipse, can be obtained. The set is given possible absolute position information and sent from the meaning storage unit 133 to the information generation unit 102. An ellipse figure is recognized when the non-linear oscillator group of the information generation unit and the meaning storage unit that carries the meaning information of the ellipse is drawn.

In the above, the graphic recognition of a circle or an ellipse has been described, but a curve can be regarded as a part of a circle or an ellipse, so that it can be easily expanded.

[0033]

As described above, according to the present invention, by detecting partial line information and corresponding line edge information from a curve or a figure composed of the curve, the semantic information of the curve or the curved figure is drawn. It becomes possible to express by a set of edge information, and line edge information corresponding to a tangent line can be used as memory information while making use of the conventional holographic structure, or the meaning memory part and weight function of conventional holographic vision can be [π / N] (N is a natural number of the detection direction) and using line edge information corresponding to the vertices of a regular 2N polygon, it is possible to recognize a curved figure such as a circle or an ellipse or a part of the curve. It becomes possible, and the practical effect of the present invention as a figure recognition device is extremely large.

[Brief description of drawings]

FIG. 1 is a basic configuration diagram of a holographic visual recognition device according to a first embodiment of the present invention.

FIG. 2 is a basic configuration diagram of a holographic visual recognition device according to a second embodiment of the present invention.

FIG. 3 (c-1) is a diagram showing how line information is generated for a circle (c-2) is a diagram showing how line information is generated for a circle (c-3) is generated line information for a circle Figure (a-1) shows that the gradient of the ridgeline of the inverted regular octagonal pyramid whose apex coincides with the detected azimuth becomes the weight function of the relay part. (A-2) shows the weight of (a-1). The figure (b-1) showing the state of the line edge information generated by the circle sent to the meaning storage unit by the VS relay of the first holographic visual recognition device using a function is an inverted inverted Masahachi with the vertex in the middle of the detection direction. The figure which shows that the gradient of the ridgeline of a pyramid becomes a weighting function of a relay part. (B-2) is a VS relay of the second holographic visual recognition device using the weighting function of (b-1) Diagram showing the state of line end information generated by the circle sent

FIG. 4 (d-1) is a diagram showing how line information is generated for an ellipse (d-2) is a diagram showing how line information is generated for an ellipse (d-3) is a diagram showing how line information is generated for an ellipse (A-3) shows that the gradient of the ridgeline of the inverted regular octagonal pyramid whose apex coincides with the detected azimuth becomes the weight function of the relay part. (A-4) shows the weight of (a-3). The figure (b-3) showing the state of the line edge information generated by the ellipse sent to the meaning storage unit in the VS relay of the first holographic visual recognition device using a function is an inverted position with the vertex in the middle of the detection direction. The figure which shows that the gradient of the ridgeline of a regular octagonal pyramid becomes a weighting function of a relay part. (B-4) is a VS relay of the second holographic visual recognition device using the weighting function of (b-3). Diagram showing the state of line end information generated by an ellipse sent to

[Fig. 5] Basic configuration diagram of holography in a conventional example

FIG. 6 is a diagram showing a main part of the conventional example.

FIG. 7 is a diagram showing a weighting function used in the conventional example.

FIG. 8 is a diagram showing a weighting function used in the conventional example.

FIG. 9 is a diagram showing how the conventional holographic vision recognizes a rectangle.

FIG. 10 is a diagram showing a mechanism in which the conventional holography generates line end information for horizontal line information.

[Explanation of symbols]

 101 information input unit 102 information generation unit 103, 133 meaning storage unit 104, 134 VS relay weighting function 105, 135 SV relay weighting function

Claims (6)

[Claims] 1. An information input section that constitutes an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions, and partial line information from the image information of the information input section. And a meaning storage unit having a plurality of pairs of line end information in the same azimuth and opposite directions arranged relative positions in the detection azimuth, and an apex in the detection azimuth. The holonic visual recognition device, wherein the gradient of the ridgeline of the inverted regular 2N pyramid in which is arranged is used as a weighting function used in information exchange between the information generation unit and the meaning storage unit. 2. An information input section which constitutes an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions, and partial line information from the image information of the information input section. An information generation unit that generates line edge information corresponding to the above, and a meaning storage unit that has two pairs of line edge information in the same azimuth and opposite directions in the same azimuth for each azimuth. A holonic visual recognition device characterized in that the gradient of the ridgeline of an inverted regular 2N pyramid having vertices arranged in the detection direction is used as a weighting function used in information exchange between the information generation unit and the meaning storage unit. 3. An information input section forming an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions, and partial line information from the image information of the information input section. An information generation unit that generates line edge information corresponding to the above, and a meaning storage unit that has a pair of line edge information in the same azimuth having opposite positions and having relative positions arranged in the same azimuth and opposite directions. Inverted 2 with apex placed in the detection direction
A holonic visual recognition device, characterized in that the slope of the ridge of an N-pyramid is used as a weighting function used in information exchange between the information generation unit and the meaning storage unit. 4. An information input section that constitutes an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions, and partial line information from the image information of the information input section. And a meaning storage unit having a plurality of pairs of line end information each having a relative position arranged in the middle of the detection azimuth and representing an angle smaller than 180 degrees. The holonic visual recognition device is characterized in that the gradient of the ridgeline of an inverted regular 2N pyramid having a vertex arranged in the middle of the detection direction is used as a weighting function used in information exchange between the information generation unit and the meaning storage unit. 5. An information input section forming an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions, and partial line information from the image information of the information input section. And an information generation unit that generates line end information corresponding to the above, and a pair of line end information that has a relative position arranged in the middle of the detection azimuth, is convex outward, is smaller than 180 degrees, and represents the angle closest to 180 degrees. And a meaning storage unit having 2N sets of line end information rotated by (180 / N) degrees for each of the relative positions, and the information about the slope of the ridgeline of the inverted regular 2N pyramid having the apex in the detection direction. A holonic visual recognition device, which is a weighting function used in information exchange between a generation unit and the meaning storage unit. 6. An information input section forming an image receiving surface with pixels arranged in a simple square lattice, and N (N is a natural number) detection directions, and partial line information from the image information of the information input section. And an information generating unit that generates line edge information corresponding to the above, and a relative position is arranged in the middle of the detection azimuth, is convex outward, is smaller than 180 degrees, and is equal in size at an angle closest to 180 degrees.
It consists of a set of 2N sets of line end information rotated by (180 / N) degrees at each of the relative positions while maintaining the convexity of the individual line end information to the outside. A holonic visual recognition device, wherein the gradient of the ridgeline of an inverted regular 2N pyramid having vertices is used as a weighting function used in information exchange between the information generation unit and the meaning storage unit.