GB2329738A - Determining relationship between line segments in pattern recognition - Google Patents
Determining relationship between line segments in pattern recognition Download PDFInfo
- Publication number
- GB2329738A GB2329738A GB9807618A GB9807618A GB2329738A GB 2329738 A GB2329738 A GB 2329738A GB 9807618 A GB9807618 A GB 9807618A GB 9807618 A GB9807618 A GB 9807618A GB 2329738 A GB2329738 A GB 2329738A
- Authority
- GB
- United Kingdom
- Prior art keywords
- line segment
- obtaining
- line segments
- sets
- line
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/757—Matching configurations of points or features
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
A method for determining an adjoining relationship between line segments in a pattern recognition system, comprises the steps of: a) obtaining sets of line segments L2,L3,L4 satisfying the adjoining relationship with a reference line segment L1 in an X-direction and a Y-direction, respectively; b) obtaining a gradient m of the reference line segment L1, and gradients m2-m4 of straight lines connecting the center point of the reference line segment L1 to the centers of the line segments L2-L4; and c) deciding that the reference line segment L1 has a left adjoining relationship with a straight line L2 if the gradient m of the reference line segment L1 is smaller than a gradient m2 of the straight line, and deciding that the reference line segment L1 has a right adjoining relationship with a straight line if the gradient m of the reference line segment L1 is larger than the gradient of the straight line, and further deciding the reference line segment L1 has no side adjoining relationship with a straight line if the gradient m of the reference line segment L1 is equal to the gradient of the straight line.
Description
METHOD FOR DETERMINING RELATIONSHIP BETWEEN LINE SEGMENTS
FOR USE IN A PATTERN RECOGNITION SYSTEM
The present invention relates to a pattern recognition apparatus and method and, more particularly, to a method for determining relationship between line segments which are extracted from an image for use in a pattern recognition system.
Pattern recognition techniques are increasingly being used or proposed in image processing for telecommunication and information systems. In some pattern recognition technique, patterns are identified by comparing the input pattern to a list of stored pattern representations. Another pattern recognition technique involves a decision which class a give input image belongs to. In a more advanced technique, pattern representation is made on the basis of the identification and connectivity or syntax of the subpatterns or pattern primitives.
An image input is transformed to binary data. Then, a contour of an object is extracted therefrom. The contour of an object can be a trajectory obtained by manipulating pixels near boundaries of the object. Lines, straight or curved, making up the contour of the object is then approximated by a set of consecutive straight line segments by using a straight line approximation method. The straight line approximation method is carried out with the help of Hough
Transform.
In the case when the contour of the object in the image input can be represented with line segments, one can find some regularities in the relationship between the line segments that form the object.
Yet, data itself obtained after applying the straightline-approximation to the image input is absolutely meaningless data. A conventional method has suggested that all the line segments obtained after applying the straightline-approximation be accessed for extracting line segments that constitute the contour, and in turn, form the outline of the object.
The conventional method, therefore, takes up too much time in extracting the contour, and thus, recognizing the object.
It is, therefore, a primary object of the present invention to provide a method for effectively determining relationship between line segments which are extracted from an image for use in a pattern recognition system. Fast and efficient pattern recognition can be accomplished.
In accordance with the present invention, there is provided a method for determining an adjoining relationship and a side adjoining relationship between line segments for use in a pattern recognition system, wherein the line segments are extracted from a scanned image and represent a straight line-approximated contour of an object, the method comprising the steps of: a) obtaining sets of line segments satisfying the adjoining relationship with a reference line segment in an X-direction and a Y-direction, respectively, wherein the reference line segment refers to a line segment selected in turn from the line segments; b) obtaining a gradient of the reference line segment, and gradients of straight lines connecting the center point of the reference line segment to the center of each of the line segments satisfying the adjoining relationship with the reference line segment; and c) determining a side adjoining relationship between the line segments by deciding that the reference line segment has a left adjoining relationship with a straight line if the gradient of the reference line segment is smaller than a gradient of the straight line, and by deciding that the reference line segment has a left adjoining relationship with a straight line if the gradient of the reference line segment is larger than the gradient of the straight line, and further by deciding the reference line segment has no side adjoining relationship with a straight line if the gradient of the reference line segment is equal to the gradient of the straight line.
The above and other objects and features of the present invention will become apparent from the following description of preferred embodiments given with reference to the accompanying drawings, in which:
Fig. 1 illustrates a typical pattern recognition system to which the present invention applies;
Fig. 2 presents an exemplary approximated contour image of an object represented with plural line segments;
Fig. 3 describes a makeup of a memory in a line estimation unit;
Figs. 4A to 4C depict a flow chart describing an inventive method; and
Figs. 5A to 5C illustrate a process of obtaining relationship between line segments.
The preferred embodiment in accordance with the present invention will now be described in detail with reference to
Figs. 1 to 5.
Referring to Fig. 1, there is illustrated a typical pattern recognition system to which the present invention applies.
A scanner 10 is one of conventional scanning devices.
The scanner 10 converts an image input into binary data. The image input scanned by the scanner 10 includes an image of an object represented by the line segments (L1, L2, L3, -) shown in Fig. 2.
A segmentation unit 20 extracts pattern primitives, e.g., straight line segments, by using a segmentation and a contour extraction algorithm. The output from the segmentation unit 20 includes an image of an object represented with line segments (L1, Lz, L3, . . .) shown in Fig. 2. The extracted contour represented with pattern primitives are mapped onto variable space, e.g., X-Y plane, so as to be transformed into data. Hough Transform is used for this transform. data about the straight-line-approximated line segments are passed to a line estimation unit 30.
The line estimation unit 30 finally determines a right or left adjoining relationship of the line segments with respect to a selected line segment in both the X and Ydirection. Information on the right or left adjoining relationship is stored in a memory of the line estimation unit 30.
The memory 40 shown in Fig. 3 is bisected into a line segment storage area 42 and an adjoining relationship storage area 44. The data of line segments are stored in the line segment storage area 42 while data on line segments having a certain adjoining relationship with each line segment are stored in the adjoining relationship storage area 44. For example, as shown in Fig. 3, data on line segment L1 is stored in the line segment storage area 42, while data on line segments L2 and L3, the line segments having the certain adjoining relationship with line segment L1, are stored in the adjoining relationship storage area 44. The information on the adjoining relationship determined at the line estimation unit 30 is passed to an object recognition unit 50.
The object recognition unit 50 makes a final classification of a pattern based on an observed syntax and on its own set of pattern syntax rules, and finally yields the result of the recognition.
Figs. 4A to 4C depict a flow chart describing a method, carried out in the line estimation unit 30, for determining the adjoining relationship in accordance with the present invention. The inventive method will be discussed later in detail in the description of the preferred embodiment.
Figs. 5A to 5C illustrate a process of obtaining the adjoining relationship between line segments.
Referring to Fig. 4A, at step 110, each line segment expressed by a set of line segments, e.g., (on, Lz, L3, L4), is arranged in an X-Y plane. Arrangement of the X-coordinate values of the ending points of the line segments form a set
X. The set X can be presented as mathematical formula 1.
[Mathematical Formula 1] X= I xi xi < xi+1, i = 1,2 ,...,n } The set X satisfying the above mathematical formula 1 can be expressed as (x1, X2, X3, . , Xs) At step 120, a set S that includes elements therein consisting of two neighboring values of projection on the Xaxis of each ending point of the line segments is obtained.
The set S can be expressed by using mathematical formula 2.
[Mathematical Formula 2]
S = ( (x,y) f xi # x # xi-1 -# < y < cc, i = 1,2,...,n-1 )
The set S satisfying the above mathematical formula 2 is ( (x1, x2), (X2TX3), (X3 x4) . . . (X7, x8) }
At following step 130, a first element of a set U is obtained, wherein the set U includes elements, each of which consists of line segments whose X-coordinate value fall within an interval as represented by each element incorporated in the set S. Herein, the line segments in the first element in the set U are arranged in an ascending order according to a Ycoordinate value of a center point of each line segment. For instance, a first element of the set U, that is, corresponding to the interval (x1, x2) , is (L4, L). Then, at step 140, it is checked if there are more intervals. Since the remaining intervals are (X2, X3) to (X2, X3), the routine goes back to step 130. The above-described routines of steps 130 and 140 for obtaining the set U is repeated with respect to all the intervals. Likewise, a second element of the set U, corresponding to the interval (xz, X3), is (L4, L1, L2); and a third element of the set Ui, corresponding to the interval (X3, X4), is (L4, Lq, L2) . Completion of steps 130 and 140 with respect to all the intervals produces the elements of (L4,
L1), (L4, L1, Lz) , (L4, L1, L2) , (L4, L1, L2) , (L4 L2) , (L L) (L3) . At step 150, the set U is obtained as ((L4, L1) , (L4, L1, L2) , (L4, L1 L2), (L4, L1, L2) , (L4, L2), (L4, L3) , (L3) ) At steps 160 to 180, a set M is obtained, wherein the set
M consists of elements, each element being a set of line segments having the adjoining relationship with each line segment Li, wherein i is a positive integer identical to the number of straight-line-approximated line segments that constitute the contour and is 4 in the preferred embodiment.
At step 160, first, with respect to the line segment La, line segments holding the adjoining relationship with the line segment L1 are obtained as the line segments L4 and L2. Then at step 170, it is checked if there are more segments with respect to which the adjoining relationship is to be obtained.
Since the remaining segments are L2, L3 and L4, the routine goes back to step 160. The line segment L2 has the adjoining relationship with line segments L4 and L. These routines 160 and 170 are repeated with respect to all the line segments. Therefore, at step 180, the set M becomes ((L2, L4) , (Ll, L4), (L4), (L1, L2, L3)) The set of line segments satisfying the adjoining relationship with each line segment,
L1 to L4, satisfies the adjoining relationship in the Ydirection.
Likewise, steps 190 to 260 are carried out for obtaining the adjoining relationship in the X-direction in a similar way as steps 110 to 180 have been executed for obtaining the adjoining relationship in the Y-direction. These routines will be briefly described below:
A set Y obtained at step 190 is (y1, y2, y3, ..., y8}.
At step 200, a set T having elements representing intervals between the ending points of the line segments is obtained. The set T satisfying the above condition yields {(y1, y2), (y2,y3), (y3, y4) , ... (y7, y8)}.
By executing steps 210 to 230, in a similar way as executing steps 130 to 150 for the set U, a set V is obtained, wherein the set V includes elements, each element consisting of line segments whose Y-coordinate value fall within an interval as listed in the set T. Herein, the line segments in the set V are arranged in an ascending order according to a X-coordinate value of a center point of each line segment.
Finally, the set V is {(L4, L3)j (L1, L4, L3), (L1, L4, L3) , (L1,
L4, L3) , (L1, L3) , (L1, L2) , (L2) } At steps 240 to 260, a set 0 is obtained, wherein the set
O consists of elements, each element being a set of line segments showing the adjoining relationship with each line segment Li. The set 0 becomes ( (L2, L3, L4) , (L1), (L1, L4) (Lr, L3)}. The set of line segments satisfying the adjoining relationship with each line segment, L1 to L4, satisfies the adjoining relationship in the X-direction.
At step 270, Nej(L), a union of the sets M and 0, is obtained, which yields a set Nei(L) as f(L2, L3, L4), (lit, L4), (L, L4), (Lr, L2, L3)}, wherein components in each element of the set Ne; (L) are union of components of the corresponding elements of the set M and the set 0.
Then, line segments satisfying the adjoining relationship in the X-direction and the Y-direction, respectively, with each of the line segments Lr, L2, L3, L4, are arranged as shown in Fig. 5C. At step 280, an m, a gradient of a reference line segment, is obtained. Subsequently, gradients of straight lines connecting the center point of the reference line segment to the center of each of the line segments satisfying the adjoining relationship with the reference line segment are calculated. For example, assuming the reference line segment is Lr, and the gradient of L1 is m, the line segments satisfying the adjoining relationship with L7 is Lz, L3 and L4, respectively. The gradients ofthe straight lines connecting the center point of L to the center of each of the line segments that satisfy the adjoining relationship with L2, L3 and L4 are obtained as m2, m3 and m4, respectively.
At step 290, m is compared with mj, e.g., mz. If m is smaller than m2, the reference line segment L is determined at step 300 to be in the left adjoining relationship with the line segments satisfying the adjoining relationship L2. If m is larger than m2, the reference line segment L is determined at step 310 to be in the right adjoining relationship with the line segment L2. If m is equal to mz, the reference line segment Lr is determined at step 305 to have neither right nor left adjoining relationship with the line segment L2. In this exemplary embodiment, the line segment L1 is determined to satisfy the left adjoining relationship with the line segment
L2. Then at step 320, the result of the adjoining relationship obtained at steps 300, 305, 310 are stored in the memory 40.
Steps 280 to 330 are repeated according to the result of checking at step 330 with respect to all the line segments, i.e., L1, L2, L3 and L4. If all the result of the adjoining relationship are stored, the procedure ends.
By using the method presented in the present invention, adjoining relationships of a line segment with other line segments, each line segment being extracted from a scanned image, are effectively obtained and the result thereof is stored. Therefore, time needed in estimating the line segments being extracted from a scanned image can be enormously reduced.
While the present invention has been described with respect to the preferred embodiments, other modifications and variations may be made without departing from the scope of the present invention as set forth in the following claims.
Claims (10)
1. A method for determining an adjoining relationship and a side adjoining relationship between line segments for use in a pattern recognition system, wherein the line segments are extracted from a scanned image and represent a straight-lineapproximated contour of an object, the method comprising the steps of:
a) obtaining sets of line segments satisfying the adjoining relationship with a reference line segment in an Xdirection and a Y-direction, respectively, wherein the reference line segment refers to a line segment selected in turn from the line segments;
b) obtaining a gradient of the reference line segment, and gradients of straight lines connecting the center point of the reference line segment to the center of each of the line segments satisfying the adjoining relationship with the reference line segment; and
c) determining a side adjoining relationship between the line segments by deciding that the reference line segment has a left adjoining relationship with a straight line if the gradient of the reference line segment is smaller than a gradient of the straight line, and by deciding that the reference line segment has a left adjoining relationship with the straight line if the gradient of the reference line segment is larger than the gradient of the straight line, and further by deciding the reference line segment has no side adjoining relationship with a straight line if the gradient of the reference line segment is equal to the gradient of the straight line.
2. The method of claim 1, wherein the method further comprises the step of:
d) storing the results of the step c) in a memory in the pattern recognition system.
3. The method of claim 1 or claim 2 wherein the step (a) includes the steps of: al) obtaining two sets, each set having elements therein representing a value of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
a2) obtaining two sets, each set having elements therein consisting of two neighboring values of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively; a3) obtaining two sets, each set having elements therein consisting of line segments, any part whose coordinate value falls within an interval between the two neighboring values of the step a2), with respect to each set obtained at the step a2);
a4) obtaining two sets, each set having elements consisting of line segments having adjoining relationship with each line segment, with respect to each set obtained at the step a3); and
a5) obtaining a union set with the two sets obtained at the step a4).
4. The method of claim 2, wherein the step a) includes the steps of:
al) obtaining two sets, each set having elements therein representing a value of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
a2) obtaining two sets, each set having elements therein consisting of two neighboring values of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
a3) obtaining two sets, each set having elements therein consisting of line segments, any part whose coordinate value fall within an interval between the two neighboring values of the step a2), with respect to each set obtained at the step a2);
a4) obtaining two sets, each set having elements consisting of line segments having adjoining relationship with each line segment, with respect to each set obtained at the step a3); and a5) obtaining a union set with the two sets obtained at the step a4).
5. An apparatus for determining adjoining relationships between line segments for use in a pattern recognition system, wherein the line segments are extracted from a scanned image and represent a straight-line-approximated contour of an object, the apparatus comprising:
means for obtaining sets of line segments satisfying the adjoining relationship with a reference line segment in an Xdirection and a Y-direction, respectively, wherein the reference line segment refers to a line segment selected in turn from the line segments;
means for obtaining a gradient of the reference line segment, and gradients of straight lines connecting the center point of the reference line segment to the center of each of the line segments satisfying the adjoining relationship with the reference line segment; and
means for determining a side adjoining relationship between the line segments by deciding that the reference line segment has a left adjoining relationship with a straight line if the gradient of the reference line segment is smaller than a gradient of the straight line, and by deciding that the reference line segment has a left adjoining relationship with a straight line if the gradient of the reference line segment is larger than the gradient of the straight line, and further by deciding the reference line segment has no side adjoining relationship with a straight line if the gradient of the reference line segment is equal to the gradient of the straight line.
6. The apparatus of claim 5, wherein the apparatus further comprises means for storing the results from the means for determining the side adjoining relationship.
7. The apparatus of claim 5 or claim 6 wherein the means for obtaining sets of line segments includes:
first means for obtaining two sets, each set having elements therein representing a value of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
second means for obtaining two sets, each set having elements therein consisting of two neighboring values of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
third means for obtaining two sets, each set having elements therein consisting of line segments, any part whose coordinate value falls within an interval between the two neighboring values of the second means, with respect to each set obtained at the second means;
fourth means for obtaining two sets, each set having elements consisting of line segments having adjoining relationship with each line segment, with respect to each set obtained at the third means; and
fifth means for obtaining a union set with the two sets obtained at the fourth means.
8. The apparatus of claim 6, wherein the means for obtaining sets of line segments includes:
first means for obtaining two sets, each set having elements therein representing a value of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
second means for obtaining two sets, each set having elements therein consisting of two neighboring values of projection on a coordinate of each ending point of the line segments, with respect to X and Y coordinates, respectively;
third means for obtaining two sets, each set having elements therein consisting of line segments, any part whose coordinate value falls within an interval between the two neighboring values of the second means, with respect to each set obtained at the second means;
fourth means for obtaining two sets, each set having elements consisting of line segments having adjoining relationship with each line segment, with respect to each set obtained at the third means; and
fifth means for obtaining a union set with the two sets obtained at the fourth means.
9. A method substantially as hereinbefore described with
reference to figures 4A to 4C and SA to SC of accompanying
drawings.
10. Apparatus constructed and arranged substantially as herein described with reference to Figures 4A-4C and 5A-5C of the accompanying drawings.
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1019970049947A KR19990027482A (en) | 1997-09-30 | 1997-09-30 | How to set the relation of object structure straight lines in image in pattern recognition system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| GB9807618D0 GB9807618D0 (en) | 1998-06-10 |
| GB2329738A true GB2329738A (en) | 1999-03-31 |
Family
ID=19521961
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| GB9807618A Withdrawn GB2329738A (en) | 1997-09-30 | 1998-04-08 | Determining relationship between line segments in pattern recognition |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JPH11120366A (en) |
| KR (1) | KR19990027482A (en) |
| GB (1) | GB2329738A (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3861157B2 (en) | 2004-02-27 | 2006-12-20 | 国立大学法人広島大学 | Reference data optimization device and pattern recognition system |
| KR100817298B1 (en) * | 2005-12-08 | 2008-03-27 | 한국전자통신연구원 | Method for detecting and tracking both hands |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1988002158A1 (en) * | 1986-09-18 | 1988-03-24 | Violet Frances Leavers | Shape detection |
-
1997
- 1997-09-30 KR KR1019970049947A patent/KR19990027482A/en not_active Application Discontinuation
-
1998
- 1998-04-08 GB GB9807618A patent/GB2329738A/en not_active Withdrawn
- 1998-04-16 JP JP10106165A patent/JPH11120366A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1988002158A1 (en) * | 1986-09-18 | 1988-03-24 | Violet Frances Leavers | Shape detection |
Also Published As
| Publication number | Publication date |
|---|---|
| KR19990027482A (en) | 1999-04-15 |
| JPH11120366A (en) | 1999-04-30 |
| GB9807618D0 (en) | 1998-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114926839B (en) | Image identification method based on RPA and AI and electronic equipment | |
| CN108615058A (en) | A kind of method, apparatus of character recognition, equipment and readable storage medium storing program for executing | |
| JPH04315272A (en) | Graphic recognizing device | |
| JPH076209A (en) | Method for extraction of feature character in character recognition system | |
| Chhabra et al. | Detection of horizontal lines in noisy run length encoded images: The FAST method | |
| KR19990074775A (en) | Fingerprint Feature Extraction Method | |
| EP0709800A2 (en) | Method for designing classification trees | |
| US4853885A (en) | Method of compressing character or pictorial image data using curve approximation | |
| GB2329738A (en) | Determining relationship between line segments in pattern recognition | |
| Ng et al. | A novel single pass thinning algorithm | |
| Sisodia et al. | A conglomerate technique for finger print recognition using phone camera captured images | |
| JP2000306104A (en) | Method and device for picture area division | |
| JPH11110542A (en) | Method and device for extracting pattern and medium recorded with program therefor | |
| CN112419208A (en) | Construction drawing review-based vector drawing compiling method and system | |
| US5426712A (en) | Method and apparatus for describing a character image based on its contours, and for classifying the character image for character recognition based on a resulting contour-based description of the character image | |
| JP2005149395A (en) | Character recognition device and license plate recognition system | |
| CN109871910A (en) | A kind of hand-written character recognition method and device | |
| JP2653386B2 (en) | Drawing processing equipment | |
| JPH07193716A (en) | Edge emphasizing device | |
| EP1538547A1 (en) | Method and device for extracting skeletal data from image data | |
| JP3344791B2 (en) | Line segment extraction method | |
| CN114694147A (en) | Method and device for segmenting surrounding characters in oval pattern | |
| JPH1027215A (en) | Method and device for recognizing processing of character pattern | |
| JPH1027216A (en) | Method and device for recognizing processing of character pattern | |
| CN118172334A (en) | Electric network wiring diagram electric element cascade detection method based on transducer and residual convolution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| WAP | Application withdrawn, taken to be withdrawn or refused ** after publication under section 16(1) |