JPH08161494A - Method for detecting line segment and its device - Google Patents

Abstract

PURPOSE: To attain accurate detection of a line segment independently of a position of discrimination by calculating an overlap area interposed between an extension line of a 1st line segment object and a 2nd line segment object and discriminating it that the 1st and 2nd line segment objects form a same line segment when a level corresponding to the area is a prescribed threshold level or below. CONSTITUTION: The device is made up of a pre-processing section 1, a line segment detection section 2, a line segment discrimination section 3 and a line segment synthesis section 4. Then an overlap area S interposed between an extension line of a line segment object 9MAX in a direction of an line segment object g1 and the line segment object g1 is calculated and whether or not the line segment object g1 forms a same line segment with the line segment object gMAX, is discriminated based on the magnitude of the overlap area S. Thus, it is avoided that a straight line is adopted for a curve in mistake or a substantially one-line-segment is thinned and a line segment is accurately detected independently of a position of discrimination.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a line segment detecting method and apparatus for detecting a line segment from an input image.

[0002]

2. Description of the Related Art Handling of parts is a very important problem for automating work in an automobile assembly factory, for example. In particular, in order to realize multifunctional and accurate automatic handling by a robot, which is required in the automation of the assembly area, thinning the image of the object is performed to detect the line segment, and from this line segment A technique for accurately extracting the shape of an object is needed.

The following methods are known as conventional line segment detection methods.

(1) The input image is subjected to pre-processing such as edge detection, binarization and thinning to obtain a thin line image.

(2) The following processes (3) to (7) are performed on all the segments included in the thin line image. Here, the "segment" means a continuous pixel row having a line width of 1 and having no branch.

(3) A straight line is applied to the k pixels n _{1 to} n _k (k = 3, for example, as an initial value of k) from the end points of an arbitrary segment by applying the least square method or the like. . Let the obtained straight line be L ₁ .

(4) The (k + 1) th pixel n _{k + 1} is a straight line L
Determine whether it is included in ₁ . That is, the pixels n ₁ to
A straight line is applied to n _{k + 1} by applying the least square method or the like, and the obtained straight line is _defined as L ₂ .

[0008] (5) comparing the inclination alpha ₂ of the straight line L ₁ of inclination alpha ₁ and the line _{L 2, | α 1 -α 2} | If the value is equal to or less than the predetermined threshold value, the pixel n _{k +} It is determined that ₁ is on the straight line L ₁ , and if the value of | α ₁ −α ₂ | is larger than the predetermined threshold value, it is determined that the pixel n _{k + 1} is not on the straight line L ₁ .

(6) If the pixel n _{k + 1} is on the straight line L ₁ , a new straight line that fits the pixels n _{1 to} n _{k + 1} is newly set as L ₁ .
Return to process (4). If the pixel n _{k + 1} is not on the straight line L ₁ ,
The pixels n _{1 to} n _k are regarded as one line segment, the pixel n _{k + 1} is used as an end point, and the process returns to the process (3) to detect a new straight line.

(7) Processes (3) to (6) are repeated until the end point of the segment is reached.

In addition to the above, as a line segment detection method, the contour of the object is extracted from the image information to obtain the contour point coordinates, and the Hough transform is performed from the contour point coordinates to obtain the function ρ = xicosθ + yisinθ. There is a method of detecting the line segment information indicated by the above-mentioned contour point coordinates by converting into a sine wave represented, creating a two-dimensional histogram according to the function value ρ, and detecting a peak point on the histogram (Japanese Patent Laid-Open No. 2000-242242). (See JP-A-64-74680).

[0012]

By the way, the above-mentioned processing
The conventional line segment detection method consisting of (1) to (7) has the following problems, which makes it difficult to accurately extract line segments. That is, (1) pixels that originally form the same line segment are not located on the same straight line as shown in FIG. 5 due to the influence of noise and the quantization error. In particular, the above process
If the threshold value used in (5) is small, it is determined that such pixels that should be on the same line segment are not on the same line segment, and the pixels are subdivided into many line segments.

(2) When the threshold value is increased in order to solve the problem (1), the line segment is fitted to the pixel on the curve, and the erroneous line segment detection result is obtained as shown in FIG. You will get

Further, the method using the Hough transform requires complicated processing and has a problem in terms of the amount of memory used and the processing cost.

Therefore, the present invention has been made to solve the above problems, and an object of the present invention is to provide a line segment detecting method and an apparatus therefor capable of easily performing accurate line segment detection.

[0016]

A line segment detection method according to the present invention is a method for detecting the same line segment from a line drawing image composed of a large number of line segment candidates obtained from an input image, and is a first line segment candidate. Is calculated and the area corresponding to the area of the overlap region interposed between the extension line and the second line segment candidate is calculated. It is characterized in that it is determined that the second line segment candidates form the same line segment.

In the line segment detecting method of the present invention, the first straight line fitted to the _k pixels n _{1 to} n _k from the end point of an arbitrary segment and the k + 1 pixels n ₁ to n k from the above end point. the angle between the second straight line that is fitted against n k _{+ 1} pixel n _{k + 1} if more than a predetermined threshold value is extracted based on the determination that in the first straight line The longest line segment candidate in the line segment candidate group can be selected as the first line segment candidate.

Further, for the pixels included in the first and second line segment candidates which are determined to form the same line segment,
The least-squares method can be applied to synthesize a new line segment.

A line segment detecting apparatus according to the present invention is provided with a pre-processing unit for performing edge detection, binarization, thinning, labeling, etc. on an input image, and a line drawing image obtained by the pre-processing unit. A first straight line fitted to _k pixels n _{1 to} n _k from the end points of any segment included, and a first straight line fitted to k + 1 pixels n _{1 to} n _{k + 1} from the end points. A line segment detection unit that extracts a line segment candidate based on the determination that the pixel n _{k + 1} is on the first straight line when the angle formed by the second straight line is less than or equal to a predetermined threshold value; The longest line segment candidate in the line segment candidate group extracted by the line segment detection unit is selected as the first line segment candidate, and the extended line of the first line segment candidate and the second line segment candidate are selected. Calculate the area of the overlapping area between them and set the value corresponding to the area of the overlapping area to the specified threshold. A line segment determination unit that determines that the first and second line segments form the same line segment when the value is less than or equal to a value, and the first and first line segments that are determined to form the same line segment by the line segment determination unit. It is characterized by further comprising a line segment synthesizing unit for synthesizing a new line segment by applying the least squares method to the pixels included in the second line segment.

[0020]

According to the line segment detecting method and apparatus of the present invention, between the extension line of the first line segment and the second line segment adjacent in the extension direction of the first line segment. Since the area of the overlap area intervening in is calculated and it is determined whether or not the first and second line segments form the same line segment based on the size of the area of the overlap area, The problem that a straight line is erroneously fitted to a curve or one line segment is originally subdivided is solved, and accurate line segment detection is possible regardless of the determination position.

[0021]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a line segment detecting device according to the present invention, which is composed of a preprocessing unit 1, a line segment detecting unit 2, a line segment determining unit 3 and a line segment synthesizing unit 4. .

The preprocessing unit 1 has a function of performing preprocessing such as edge detection, binarization, thinning, and labeling on the input image, and the line segment detection unit 2 performs the preprocessing. The first straight line fitted to the _k pixels n _{1 to} n _k from the end points of any segment included in the line drawing image obtained in the section 1, and the k + 1 pixels n _{1 to} n _{k +} from the end points. ₁
A function of extracting a line segment based on the determination that the pixel n _{k + 1} is on the first straight line when the angle formed by the second straight line fitted to Have.

The line segment determination unit 3 first determines the longest line segment candidate in the line segment candidate group extracted by the line segment detection unit 2.
Value of the overlap region intervening between the extension line of the first line segment candidate and the second line segment candidate, and a value corresponding to the area of the overlap region. Is less than or equal to a predetermined threshold value, the first and second
Has a function of determining that they form the same line segment, and the line segment combining unit 4 determines that the line segment determination unit 3 determines that the line segment combination unit 3 forms the same line segment. It has a function of applying the least squares method to the pixels included in the segment candidate to synthesize a new line segment.

Next, the processing executed by the line segment detecting device will be described with reference to the flowchart of FIG.

(I) Pre-processing The input image is sequentially subjected to processing such as edge detection, binarization, thinning and labeling.

(II) Line Segment Detection Similar to the above-mentioned conventional technique, the first straight line fitted to the _k pixels n _{1 to} n _k from the end point of an arbitrary segment included in the line drawing image, and the above When the angle formed by the second straight line fitted to the _{k + 1} pixels n _{1 to} n _{k + 1} from the end point is equal to or less than a predetermined threshold value, the pixel n _{k + 1} is placed on the first straight line. A line segment candidate is extracted based on the determination that there is. At this time, the threshold value is set small so that a large number of subdivided line segment candidates are detected. The detected line segment candidate is labeled with the same label as the segment (S
1).

[0028] (III) with respect to the segment candidate segment discrimination (a) label 1 is affixed g ₁ to g _n, the following processing is performed.

(B) The longest line segment candidate g _m is selected from the line segments g _{1 to} g _n , and it is set as g _MAX (S2).

(C) Line segment candidates g _MAX and g _i (i = 1,
2, ... M-1, m + 1, ... N) are determined to be the same. As shown in FIG. 3, the same determination is performed by drawing perpendicular lines from both ends of the line segment g _i with respect to the extension line of the line segment candidate g _{MAX in} the direction of the line segment candidate g _i to obtain the line segment g _MAX and the line segment g _i. A region surrounded by _i and both perpendiculars, that is, an extension of line segment g _MAX and line segment g
Overlap area S (shaded area) between _i and
An evaluation value E corresponding to the area of the line segment is calculated by the following evaluation formula, and when the evaluation value E is less than or equal to a predetermined threshold value, the line segment candidate g _i constitutes the same line segment as the line segment candidate g _MAX. If the evaluation value E is larger than the predetermined threshold value, it is determined that the line segment candidate g _i constitutes a line segment different from the line segment candidate g _MAX (S3).

[0031]

[Equation 1]

However, g _MAX : ax + by + c = 0, a vector parallel to the line segment candidate g _i is (p, q), (x _C , y
Let _C ) be the midpoint of the line segment candidate g _i and L be the length of the line segment candidate g _i .

(IV) Line segment composition (a) After the same determination is made for all line segment candidates (S4), the pixels forming the same line segment are combined by applying the least squares method. Find the equation of the line segment (S
5). In practice, the equation of the line segment y = a
Find the parameter of x + b.

[0034]

[Equation 2]

(B) Process of (III) above (b) (S2)
Then, the process is repeated for the remaining line segment candidates g _k .
When the line segment candidates g _k are exhausted, the process proceeds to (c).

(C) The next label is selected, and the process returns to (III) (B) (S2). When the processing is completed for all the labels, the processing ends.

The above is the description of the embodiment of the line segment detecting method and apparatus according to the present invention. According to the present embodiment, the extension line and the line segment of the line segment candidate g _{MAX in} the candidate line segment g _i direction. calculating the area of the overlap region S interposed between the candidate g _i, whether based on the magnitude of the area of the overlap region, the line segment candidate g _i constitutes the same segment and the line segment candidate _MAX Since it is determined whether or not a straight line is accidentally applied to a curve or one line segment is originally subdivided, accurate line segment detection is possible regardless of the determination position. .

In this embodiment, in the pre-processing (I), each processing such as edge detection, binarization, thinning and labeling is sequentially performed on the input image, and the line (II) is processed. In the minute detection, a first straight line fitted to the _k pixels n _{1 to} n _k from the end points of any segment included in the line drawing image and the k + 1 pixels n _{1 to} n _{k + 1} from the end points. the angle between the second straight line fitted against the case is less than a predetermined threshold value, but the pixel n _{k + 1} is determined to be in the first straight line, instead, The vector (direction and size) of the density gradient of each pixel at the edge detected by differentiating the input image is obtained, and when the difference between the vectors of the density gradient of pixels adjacent to each other is smaller than a predetermined threshold value, It is determined that two adjacent pixels are on the same straight line. It may be.

That is, as shown in FIG. 4, assuming that the densities of pixels in the vicinity of one pixel p ₀ having coordinates (x ₀ , y ₀ ) 8 are f _{1 to} f ₈ , the horizontal direction of the pixel p ₀ is shown. The concentration gradient u and the vertical concentration gradient v are respectively expressed by the following equations.

U = (f ₃ + f ₄ + f ₅ ) − (f ₁ + f ₇ + f ₈ ) v = (f ₅ + f ₆ + f ₇ ) − (f ₁ + f ₂ + f ₃ ). Further, the pixel p ₀ (x ₀ , x ₀ , If the straight line indicating the direction of the vector of the concentration gradient of y ₀ ) is y = k × (x−x ₀ ) + y ₀ , the coefficient k indicating the direction of the vector and the size P _{0 of the} vector are respectively expressed by the following equations. It is represented by.

K = tan ⁻¹ (v / u) P ₀ = √ (u ² + v ² ) Such a calculation is performed for each pixel to obtain the direction and magnitude of the vector of the density gradient of the pixel, and they are adjacent to each other. It may be determined that the pixels whose difference in the direction and magnitude of the vector of the density gradient of the pixels are within the predetermined thresholds are on the same straight line.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of the configuration of a line segment detecting device according to the present invention.

FIG. 2 is a flowchart showing the flow of processing in the line segment detecting method according to the present invention.

FIG. 3 is an explanatory diagram of the same action.

FIG. 4 is an explanatory diagram of a method for obtaining a density gradient of a pixel from the densities of pixels in the vicinity of the pixel.

FIG. 5 is an explanatory diagram of problems of the conventional line segment detection method.

FIG. 6 is an explanatory view of problems of the conventional line segment detection method.

[Explanation of reference numerals] 1 preprocessing unit 2 line segment detection unit 3 line segment determination unit 4 line segment synthesis unit

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hiroyuki Yoshida 3-1, Shinchi Fuchu-cho, Aki-gun, Hiroshima Prefecture Mazda Co., Ltd. (72) Inventor Hiroyuki Takahashi 3-1-1 Shinchu, Fuchu-cho, Hiroshima Prefecture Mazda Motor Corporation Within

Claims (4)

[Claims] 1. A method for detecting the same line segment from a line drawing image composed of a large number of line segment candidates obtained from an input image, wherein the line segment is between the extension line of the first line segment candidate and the second line segment candidate. If the area corresponding to the area of the overlap area is calculated and the value corresponding to the area of the overlap area is less than or equal to a predetermined threshold value, the first and second line segment candidates form the same line segment. A line segment detection method characterized by making a determination. 2. A first straight line fitted to _k pixels n _{1 to} n _k from the end point of an arbitrary segment and a k + 1 pixel n _{1 to} n _{k + 1} fitted from the end point. Of the line segment candidate group extracted based on the determination that the pixel n _{k + 1} is on the first straight line when the angle formed by the formed second straight line is equal to or smaller than a predetermined threshold value. The line segment detection method according to claim 1, wherein a long line segment candidate is selected as the first line segment candidate. 3. A new line segment is synthesized by applying the least-squares method to pixels included in the first and second line segment candidates that are determined to form the same line segment. The line segment detection method according to claim 1 or 2. 4. Edge detection for an input image, binarization,
A pre-processing unit for performing each processing such as thinning and labeling, and a first processing applied to _k pixels n _{1 to} n _k from the end point of an arbitrary segment included in the line drawing image obtained by the pre-processing unit. 1 and k + 1 pixels n ₁ from the above end point
If the angle formed by the second straight line fitted to ˜n _{k + 1} is less than or equal to a predetermined threshold value, the pixel n _{k + 1} is
Line segment detection unit that extracts line segment candidates based on the determination that they are on the straight line, and the longest line segment candidate of the line segment candidate group extracted by the line segment detection unit is the first line segment. It is selected as a candidate, the area of the overlap region interposed between the extension line of the first line segment candidate and the second line segment candidate is calculated, and the value corresponding to the area of the overlap region is set to a predetermined value. A line segment determination unit that determines that the first and second line segment candidates form the same line segment when the threshold value is less than or equal to a threshold value, and the first line segment determination unit that determines that the line segment determination unit configures the same line segment. A line segment synthesizing unit for synthesizing a new line segment by applying the least squares method to the pixels included in the first and second line segment candidates.