JPH11120366A - Segment contiguity relation determining method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the time needed to estimate a segment more by effectively determining the contiguity relation between segments extracted from scanned video and storing the contiguity relation information in a memory. SOLUTION: A division part 20 extracts a pattern primitive like a segment by using segmenting and outline extracting algorithm. The extracted outline that the pattern primitive represents is mapped to, for example, an X-Y plane and converted into vector data. Then the segment data approximated to a segment is supplied to a segment estimation part 30. This estimation part 30 determines contiguity relation between segments along an X or Y axis for the selected segment and information on the determined contiguity relation is stored in a memory in the estimation part 30. Then this contiguity relation information is supplied to a body recognition part 50, which performs pattern classification according to retrieved syntax and its own pattern syntax rule to output a final pattern recognition result.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern recognition system and, more particularly, to a method for determining an adjacency between a plurality of lines representing contours of an object extracted from a scanned image in the pattern recognition system. .

[0002]

2. Description of the Related Art In video processing in telecommunications information systems, pattern recognition techniques are being actively studied. In pattern recognition techniques, pattern recognition is performed by comparing an input pattern with a stored list of pattern models. Other pattern recognition techniques include determining which pattern a given input video belongs to, and improved techniques include sub-patterns or pattern primitives. Configured based on identification, connectivity or syntax.

[0003] An input image is converted into binary data, and thereafter, an outline of an object is extracted. The contour line of the object is obtained as a trajectory formed by connecting pixels existing in the boundary portion with line segments. Thereafter, the line segments or curves constituting the contour of the object are approximated by a series of continuous line segments by a line segment approximation method. This straight-line approximation method is performed by a Hough transform method.

When an outline of an object in an input image can be represented by a plurality of line segments, a certain rule exists between the plurality of line segments constituting the object.

[0005] However, data obtained by approximating the input video data by the linear approximation method is only meaningless data. In the conventional method, all the line segments obtained by the straight line approximation method are applied to the line segments constituting the object to form the outer shape of the object.

Therefore, the conventional method has a problem that it takes a considerable amount of time to extract the contour of an object and is inefficient.

[0007]

SUMMARY OF THE INVENTION Accordingly, a main object of the present invention is to use a pattern recognition system, which effectively determines the adjacency between a plurality of line segments extracted from an input image, thereby achieving high speed and high speed. An object of the present invention is to provide a line segment adjacency determination method for efficiently recognizing a pattern.

[0008]

According to the present invention, there is provided, in accordance with the present invention, a contour line of an object used in a pattern recognition system, each of which is extracted from a scanned image and approximated by a straight line. Is a line segment adjacency relationship determining method for determining an adjacency relationship between a plurality of line segments, and the adjacency relationship is satisfied with a reference line segment sequentially selected from the plurality of line segments in the X-axis and Y-axis directions. An a step of obtaining a set of line segments, a b step of obtaining a gradient of the reference line segment, and a gradient of a plurality of straight lines connecting the reference line segment and the midpoint of each of the line segments satisfying the adjacent relationship; When the gradient of the reference line segment is smaller than the gradient of the straight line, it is determined that the reference line segment has an adjacent relationship on the left side of the straight line, and when the gradient of the reference line segment is larger than the gradient of the straight line, , The reference line segment has an adjacent relationship on the right side of the line segment. Determining that the gradient of the reference line segment is equal to the gradient of the straight line, and determining that the reference line segment does not have an adjacent relationship with the straight line. A line segment adjacency determination method is provided.

[0009]

Preferred embodiments of the present invention will be described below in detail with reference to the drawings.

Referring to FIG. 1, there is shown a schematic block diagram of a pattern recognition system according to the present invention, which comprises a scanner 10, a division unit 20, a line segment estimation unit 30, and an object recognition unit 40. Have been.

The scanner 10 is one of ordinary scanner machines, and plays a role of converting an input video signal into binary data. The image scanned by the scanner 10 includes an object constituted by line segments L ₁ , L ₂ , L ₃ ,... As shown in FIG.

The dividing unit 20 extracts a pattern primitive such as a line segment using a line differentiation and contour line extraction algorithm. The output extracted by the division unit 20 has an object composed of line segments L ₁ , L ₂ , L ₃ ,... As shown in FIG. The extracted contour represented by the pattern primitive is mapped to a variable space, for example, an XY plane, and is converted into vector data. In this vector data conversion process, a Hough transform method that approximates an object to a line segment is used. After that, the line segment data approximated to the line segment is supplied to the line segment estimation unit 30.

The line segment estimating unit 30 determines the adjacency of the selected line segment in the X-axis or Y-axis direction, that is,
Determine the right or left adjacency. The information on the adjacency determined in this manner is stored in the memory 40 (see FIG. 3) in the line segment estimating unit 30.

As shown in FIG. 3, the memory 40 comprises a line segment data storage area 42 and an adjacent relation information storage area 44. Line segment data storage area 42 stores line segment vector data, and adjacency information storage area 44 stores adjacency information for the line segment vector data, that is, right or left adjacency information. Thereafter, the adjacency relation information obtained by the line segment estimation unit 30 is
0 is supplied.

The object recognition unit 50 performs pattern classification based on the syntax to be searched and its own pattern syntax rule, and outputs a final pattern recognition result.

FIGS. 4 to 6 are flow charts for explaining the process of determining the adjacency of line segments performed by the line segment estimating unit 30 according to the present invention. 7 to 9 are schematic diagrams each showing an adjacent relationship between line segments.

Referring to FIGS. 4-6, initially, at step 110, a set of line segment vectors (eg, L ₁ ,
L _2, each line segment is represented by L _3, L ₄₎ is arranged on the X-Y plane. A set X is formed by aligning both end points of these line segments on the X coordinate. This set X is expressed as the following equation (1).

[0018]

X = ｛χ _i | χ _i <χ _{i + 1} , i = 1, 2,..., N} Equation (1)

The set X satisfying the above equation (1) is
{X ₁ , X ₂ , X ₃ ,..., X ₈ }.

In step 120, both end points of each line segment are mapped on the X axis to obtain a set S composed of two adjacent elements. This set S can be expressed as in the following equation (2).

[0021]

S = {(X, y)} _i ≤ {≤ { _{i + 1} ,-{y}, i = 1, 2,..., N-1} Equation (2)

Therefore, the set S satisfying the above equation (2)
Are ｛(X ₁ , X ₂ ), (X ₂ , X ₃ ), (X ₃ , X ₄ ),
.., (X ₇ , X ₈ )}.

In step 130, the first element of the set U composed of line segment elements, which is present in the section defined by the elements of the set S on the X coordinate, is obtained. here,
As shown in FIG. 7, each line segment of the first element of the set U is arranged in the ascending order of the Y coordinate value by mapping the midpoint of each line segment on the Y coordinate. For example, the first element of the set U corresponding to the section (X ₁ , X ₂ ) is (L ₄ , L ₁ ). Thereafter, in step 140, it is determined whether there is a section to be processed. Still, the remaining sections are (X ₂ , X ₃ ) more than (X ₂ , X ₃ )
Since there are up to X ₃ ), the process returns to step 130 and is repeated for the entire remaining interval. Similarly, the second element of the set U corresponding to the section (X ₁ , X ₂ ) is (L ₄ , L ₁ , L ₂ ), and the third element corresponding to the section (X ₃ , X ₄ ) is ( L ₄ , L ₁ , L ₂ ). When the processing for all the sections is completed in this way, step 1
At 50, the set U is ｛(L ₄ , L ₁ ), (L ₄ , L ₁ ,
L ₂ ), (L ₄ , L ₁ , L ₂ ), (L ₄ , L ₁ , L ₂ ),
(L ₄ , L ₂ ), (L ₄ , L ₃ ), (L ₃ )}.

[0024] In step 160 to 180, obtains the set M of each element consists of a set of line segments having adjacent relationship with each line segment L _i. Here, i is a positive integer, which is the same as the number of line segments constituting the contour, and is 4 in this embodiment. The first step 160, the line segment having the adjacency relationship between the line segment L ₁ is obtained as two line segments L _4, L _2. Thereafter, it is checked whether or not there is also a line segment having an adjacent relationship (step 170). The remaining line segments L ₂ , L ₃ , L
_Since there are _four , the process returns to step 160,
This is repeated for the remaining line segments. In this way, when the processing for all the line segments is completed, the set M becomes ｛(L ₂ , L ₄ ), (L ₁ , L ₄ ),
(L ₄ ), (L ₁ , L ₂ , L ₃ )}. Therefore, each set of line segments having an adjacent relationship with each of the line segments L _{1 to} L ₄ satisfies the adjacent relationship in the Y-axis direction.

Similarly to the process of obtaining the set of line segments satisfying the adjacent relationship in the Y-axis direction in steps 110 to 180 described above, steps 190 to 260 determine the set of line segments satisfying the adjacent relationship in the X-axis direction. It is done repeatedly to find out.

At step 190, the set Y is ｛y ₁ ,
y ₂ , y ₃ ,..., y ₈ } are required.

In step 200, a set T in which each element represents a section between both end points of each line segment is represented by ｛(y ₁ , y ₂ ), (y
₂ , y ₃ ),..., (Y ₇ , y ₈ )}.

As in steps 130 to 150, the step
Steps 210 to 230 are repeatedly performed, so that Y
Exists in the section defined by each element of set T on coordinates
The first element of the set V consisting of line segment elements
You. Here, as shown in FIG.
Each line segment is obtained by mapping the midpoint of each line segment on the X coordinate.
Sorted in ascending order of coordinate values. So asked
The set V obtained is ｛(L_Four, L_Three), (L₁, L_Four,
L_Three), (L₁, L_Four, L_Three), (L₁, L_Four, L_Three),
(L₁ , L_Three), (L₁, L_Two), (L_Two)｝

In steps 240 to 260, each element is
Line segment L_iA set O consisting of a set of line segments having an adjacent relationship with
｛(L_Two, L_Three, L_Four), (L₁), (L₁, L_Four),
(L₁, L_Three )｝ Required. Each line segment L₁~ L_FourAnd adjacent
The elements of each line segment that have a relationship satisfy the adjacent relationship in the X-axis direction.
Add.

In step 270, the union Ndi (L) between the set M and the set O is ｛(L ₂ , L ₃ ,
L ₄ ), (L ₁ , L ₄ ), (L ₁ , L ₄ ), (L ₁ , L ₂ ,
L ₃ ). Here, the union Ndi (L)
Is the sum of the components of the corresponding elements in both sets M and O.

FIG. 9 shows a line segment satisfying the adjacent relationship in the X and Y axis directions with respect to the line segments L ₁ , L ₂ , L ₃ , and L ₄ . In step 280, obtaining the gradient or slope m of the reference line, the slope m _i of the straight line connecting the midpoint of each line segment that satisfies the midpoint and the reference line segment and the adjacent relationship of the reference line segment. For example, the reference line segment is L _1, the slope of the reference line segment is m, the line segments that satisfy the adjacency and L ₁ is L _2, L ₃ and L _4. Slope of the line connecting the midpoint of each line segment L ₁ of the midpoint and L _2, L ₃ and L ₄ and satisfy the adjacency is determined as each m _2, m ₃ and m _4.

In step 290, a comparison is made between gradient m and gradient _mi (eg, m ₂ ). If the slope m is the slope m ₂ less than, determines a reference line L ₁ is disposed on the left side of the line segment L ₂ satisfying the adjacency (step 300), the slope m is than the slope m ₂ If large, the reference line segment L ₁ is determined as being located on the right side of the line segment L ₂ (step 310), if the slope m = slope m ₂ is
It is determined that it does not belong to any of both sides (step 30)
5).

At step 280, all line segments L ₁ ,
By checking L ₂ , L _3, and L ₄ , in this embodiment, it is determined that the line segment L ₁ has an adjacent relationship on the left side of the line segment L ₂ . Then step 3
The result data obtained in 00 to step 310 is stored in the memory 4
0 (step 320).

In step 330, it is determined whether or not there is a line segment to be processed. If so, the process returns to step 300, and the above-mentioned steps 280 to 33 are performed.
This operation is repeated 0. If all line segments have been processed and the resulting data is all stored in memory 40, the process ends.

While the preferred embodiment of the present invention has been described above, those skilled in the art will be able to make various modifications without departing from the scope of the present invention.

[0036]

According to the present invention, therefore, the adjacency between the line segments extracted from the scanned image is effectively determined, and the adjacency information is stored in the memory to estimate the line segments. The time required to do so can be further reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram of a normal pattern recognition system.

FIG. 2 is a schematic diagram showing an approximate contour image of an object composed of a plurality of line segments.

FIG. 3 is a schematic diagram showing a configuration of a memory in a line segment estimation unit in FIG. 1;

FIG. 4 is a flowchart illustrating a line segment adjacency determination method according to the present invention.

FIG. 5 is a flowchart illustrating a line segment adjacency determination method according to the present invention.

FIG. 6 is a flowchart illustrating a line segment adjacency determination method according to the present invention.

FIG. 7 is a schematic diagram showing an adjacency relationship between line segments.

FIG. 8 is a schematic diagram showing an adjacency relationship between line segments as in FIG. 7;

FIG. 9 is a schematic diagram showing an adjacency relationship between line segments as in FIGS. 7 and 8;

[Explanation of symbols]

 Reference Signs List 10 scanner 20 division unit 30 line segment estimation unit 40 memory 42 line segment data storage area 44 adjacency relation information storage area 50 object recognition unit

Claims (8)

[Claims] 1. A line segment adjacency determination for use in a pattern recognition system, wherein each line segment is extracted from a scanned image and determines the adjacency between a plurality of line segments representing the contour of an object approximated by a straight line. A method of obtaining a set of line segments satisfying an adjacent relationship with a reference line segment sequentially selected from the plurality of line segments in the X-axis direction and the Y-axis direction, a gradient of the reference line segment; B) determining a gradient of a plurality of straight lines connecting the reference line segment and the midpoint of each of the line segments satisfying the adjacency relationship; and if the gradient of the reference line segment is smaller than the gradient of the straight line, the reference line Is determined to have an adjacent relationship on the left side of the straight line, and if the gradient of the reference line segment is greater than the gradient of the straight line, it is determined that the reference line segment has an adjacent relationship on the right side of the line segment. And the gradient of the reference line segment is equal to the gradient of the straight line. And c. Determining that the reference line segment does not have an adjacent relationship with the straight line. 2. The method according to claim 1, further comprising the step of: d storing the result data obtained in the step c in a memory incorporated in the pattern recognition system.
3. The method for determining line segment adjacency according to item 1. 3. The a-step in which the a-step includes: an element mapping each end point of the plurality of line segments on an X coordinate and a Y coordinate to obtain two sets representing values on each coordinate; An a2 step in which each element maps both end points of the plurality of line segments on the X axis and the Y axis to obtain two sets of two adjacent values; and each set obtained in the a2 step. A3 step of obtaining two sets of line segment elements existing in an interval between the two adjacent values, and for each set obtained in the a3 step, Has an a4 step of obtaining two sets of line segments having an adjacent relationship with each of the line segments, and an a5 step of obtaining a union between the two sets obtained in the a4 step. 2. The method according to claim 1, wherein: 4. An a1 step in which each element maps each end point of the plurality of line segments on an X coordinate and a Y coordinate to obtain two sets representing values on each coordinate. An a2 step in which each element maps both end points of the plurality of line segments on the X-axis and the Y-axis to obtain two sets of two adjacent values, and each obtained in the a2 step A3 step of obtaining two sets of line segment elements, which are present in the section between the two adjacent values, for each set; and for each set obtained in the a3 step, An a4 step of obtaining two sets of line segments whose elements are adjacent to each of the line segments; and an a5 step of obtaining a union between the two sets obtained in the a4 step. 3. The method for determining a line segment adjacency relationship according to claim 2, wherein: 5. A line segment adjacency determination which is used in a pattern recognition system and which determines an adjacency between a plurality of line segments each extracted from a scanned image and representing a contour of an object approximated by a straight line. An apparatus for calculating a set of line segments satisfying an adjacency with a reference line segment sequentially selected from the plurality of line segments in the X-axis direction and the Y-axis direction; And gradient calculating means for determining the gradient of a plurality of straight lines connecting the reference line segment and the midpoint of each of the line segments satisfying the adjacency relationship. If the gradient of the reference line segment is smaller than the gradient of the straight line, It is determined that the reference line segment has an adjacent relationship on the left side of the straight line, and if the gradient of the reference line segment is larger than the gradient of the straight line, the reference line segment has an adjacent relationship on the right side of the line segment. And the gradient of the reference line segment is A line segment adjacency determination device, comprising: an adjacency determination unit that determines that the reference line segment does not have an adjacency with the straight line when the gradient is equal to the line gradient. 6. The line segment adjacency determination device according to claim 5, further comprising a storage unit for storing result data obtained by said adjacency determination unit. 7. A first set in which the line segment set calculating means maps each end point of each of the plurality of line segments on an X coordinate and a Y coordinate to obtain two sets representing values on each coordinate. Calculating means; and each element mapping both end points of the plurality of line segments on the X axis and the Y axis to obtain two sets of two adjacent values; and the second set A third set calculating means for obtaining, for each set obtained by the calculating means, two sets of line segment elements existing in an interval between the two adjacent values; and the third set calculating means A fourth set calculating means for obtaining, for each set obtained by the above, two sets each including a line segment having each element adjacent to each of the line segments; and two sets obtained by the fourth set calculating means. And a fifth set calculating means for obtaining a union between Line adjacency determination apparatus according to Motomeko 5. 8. A first set in which the line segment set calculating means maps each end point of each of the plurality of line segments on an X coordinate and a Y coordinate to obtain two sets representing values on each coordinate. A second set calculation means, wherein each element maps both end points of the plurality of line segments on an X axis and a Y axis to obtain two sets of two adjacent values, and the second set A third set calculating means for obtaining, for each set obtained by the calculating means, two sets of line segment elements existing in an interval between the two adjacent values; and the third set calculating means A fourth set calculating means for obtaining, for each set obtained by the above, two sets each including a line segment having each element adjacent to each of the line segments; and two sets obtained by the fourth set calculating means. And a fifth set calculating means for obtaining a union between Line adjacency determination apparatus according to Motomeko 6.