KR101677875B1 - Apparatus and method for line segment detection - Google Patents

Abstract

A segment recognition method and apparatus are disclosed. The segment recognition method includes the steps of: (a) detecting a straight line component by converting a coordinate of an arbitrary pixel among pixels of an edge region detected in an image into a Hough space using PPH (Progressive Probability Hough Transform); (b) Performing line segment tracking on a straight line component detected using a continuous pixel number range; and (c) performing linear verification of a line segment detected through line segment tracking, wherein the continuous pixel number range corresponds to a slope of a straight line (B) is a characteristic value that is preset in accordance with an angle corresponding to a slope of a straight line, and the pixel in the straight line is continuous in a direction of a first axis, which is either a horizontal axis or a vertical axis, And when the value of the second axis, which is one of the horizontal axis and the vertical axis except for the first axis, increases by 1, Head to determines whether or not the continuous appear in a continuous pixel number range.

Description

[0001] Apparatus and method for line segment detection [

The present invention relates to image recognition, and more particularly, to a method and apparatus for recognizing a line segment for an image.

With the increase of image data, pattern recognition in images has begun to be used in various fields. Applications such as augmented reality, security cameras, face recognition, and smart cars are spreading widely. Especially, it is necessary to recognize driving environment of driving assist system of smart car. Recently, relatively low cost camera image is used. Therefore, it is essential to recognize the surrounding environment through image recognition, and related researches have been actively carried out.

Image recognition technology starts with recognizing the edge and color of the object. The driver assistance system needs to detect the straight line after edge detection in order to recognize lane or traffic sign. The Hough transform algorithm is the most representative algorithm for detecting a straight line in a digital image. Canny The Hough transform algorithm that finds a straight line from a binarized image that has undergone edge detection process is widely used for straight line detection because it has the advantage of recognizing a straight line even in images containing data loss and distortion. . However, the Hough transform algorithm requires a large amount of computation because it requires repetitive execution of large overhead operations such as trigonometric function, multiplication, division, etc. in order to detect linear information. In addition, since the Hough transform algorithm finds a straight line passing through specific points, it can know the slope of the straight line and the coordinates that meet the axis, but the actual line segment can not be found in the image.

Progressive Probability Hough Transform (PPHT) is an algorithm that reduces the computational complexity of Hough transform and finds the actual endpoints of straight line components. PPHT is generally performed after edge detection. However, since the edge detection result is complex, the recognition rate for the line segment is low when many lines exist as noise. That is, in the case of the image received from the camera, the edge detection result includes a noise component that is not visible to the human eye depending on the weather, illumination, and performance of the image sensor. Since PPHT finds only an accurate straight line component in consideration of these components, there is a problem that a line segment recognized as a straight line is not found or the line segment recognized as two or more line segments occurs.

The present invention is intended to provide a method and apparatus for recognizing a segment, which is modified to correct a PPHT method and obtain a result similar to that of a linear component detection result.

According to an aspect of the present invention, a segment recognition method using Progressive Probability Hough Transform (PPHT) is disclosed.

A line segment recognition method according to an embodiment of the present invention includes the steps of (a) detecting a straight line component by converting a coordinate of an arbitrary pixel among pixels of an edge region found in an image to a Hough space using PPHT (Progressive Probability Hough Transform) (B) performing line segment tracking on the detected straight line component using a preset continuous number of pixel ranges; and (c) performing straight line verification of a line segment detected through the line segment tracing, Wherein the continuous pixel number range is a characteristic value preset in accordance with an angle corresponding to a slope of a straight line, and the step (b) Axis direction in the direction of the first axis, and determines whether or not the horizontal axis or the vertical axis It is determined whether or not the value of the second axis, which is one of the axes, increases continuously by one in the direction of the first axis within the continuous number of pixels.

The continuous pixel number range is? (Natural number) or? -1, and? Is the characteristic value.

The step (c) includes selecting pixels of the detected line segment at a first predetermined interval, and verifying whether the detected line segment is straight or not by checking the slope variation amount between the first intervals.

Wherein the step (c) comprises: if the end point of the detected line segment is located inside the last validation period, checking the slope change amount by moving the slope change amount by the first interval in reverse direction starting from the end point, And changing the end point of the line segment to a point that finally maintains a straight line if the line is out of a preset allowable range.

(d) tracking the detected line segment and the broken additional line segment.

The step (d) includes the steps of: determining whether a pixel exists at a position having the same slope from a point skipped by a second predetermined interval at the end point of the detected line segment; And performing the line segment tracking to track the additional line segment.

Wherein the step (d) comprises the steps of: confirming whether or not the inclination of the additional line segment is the same as the line segment to which the line segment is plotted; and if it is the same, As a final end point.

According to another aspect of the present invention, there is provided a segment recognition apparatus using Progressive Probability Hough Transform (PPHT).

A segment recognition apparatus according to an embodiment of the present invention includes a memory for storing an instruction and a processor for executing the instruction, the instruction including: (a) detecting an edge region (s) in the image using a progressive probabilistic Hough transform (PPHT) (B) performing line segment tracking on the detected straight line component by using a predetermined continuous number of pixel ranges; and (c) Wherein the continuous pixel count range is a characteristic value set in advance according to an angle corresponding to a slope of a straight line, and the (b ) Is performed in the direction of the first axis, which is either the horizontal axis or the vertical axis, depending on the angle corresponding to the slope of the straight line And the number of consecutive pixels in the direction of the first axis is incremented by one every time the value of the second axis which is one of the horizontal axis or the vertical axis except for the first axis increases by one, It is determined whether or not they appear continuously within the range.

The method and apparatus for recognizing a line segment according to an embodiment of the present invention allow a small error of a pixel level that is not visible to the naked eye, thereby increasing the linear component recognition rate and exhibiting a strong noise characteristic.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a diagram illustrating Huff transformation of digital pixel data.
2 is a flowchart illustrating a line segment recognition method according to an embodiment of the present invention.
3 is a diagram illustrating the distribution of pixels along an angle in a straight line.
4 is a diagram illustrating a result of linear recognition.
FIG. 5 is a diagram showing a straight line verification of a line segment and an additional trace of a broken line segment; FIG.
6 is a diagram illustrating a line segment detection result of a building image using a line segment recognition method according to an embodiment of the present invention.
7 is a diagram illustrating a lane recognition result using a line segment recognition method according to an embodiment of the present invention.
8 is a view schematically illustrating a configuration of a line segment recognition apparatus according to an embodiment of the present invention.

As used herein, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise. In this specification, the terms "comprising ", or" comprising "and the like should not be construed as necessarily including the various elements or steps described in the specification, Or may be further comprised of additional components or steps. Also, the terms "part," " module, "and the like described in the specification mean units for processing at least one function or operation, which may be implemented in hardware or software or a combination of hardware and software .

First, before describing a segment recognition method according to an embodiment of the present invention, a method of Progressive Probability Hough Transform (PPHT) will be described with reference to FIG. 1 in order to facilitate understanding and explanation of the present invention.

Figure 1 is a diagram illustrating Huff transform of digital pixel data.

In order to recognize a line segment in an image, it is necessary to perform an operation of finding all edge regions in the image, and then extract and confirm a line segment having a linear component in the edge region. Here, the work for finding the edge region can be performed using the Kanji edge detection method.

The Hough transform method is one of image processing methods for finding a line segment, and detects linear information using the following equation.

That is, in order to detect linear information, the Hough transform method converts each pixel existing in the (x, y) coordinate space of the image into (?,?) Coordinate space referred to as Hough space using Equation (1).

Referring to FIG. 1, two coordinates P1 (x1, y1) and P2 (x2, y2) existing on one straight line intersect at one point in the (?,?) Coordinate space.

The Hough transform method uses this to recognize the frequency accumulated at one point (?,?) As pixel information distributed on a straight line, and if the accumulated frequency value is larger than the predetermined threshold value, it is determined as a straight line. That is, when a pixel on one straight line in the (x, y) coordinate space is Hough transformed, it appears as a point in Hough space. The larger the accumulated frequency at this point, the longer the length of the straight line.

However, such a Hough transform method can recognize that a straight line exists and a length is short or long, but it is not recognized where a straight line is located in the (x, y) coordinate space. In addition, since the conversion must be performed for all the pixels, there is a disadvantage that the amount of calculation is large.

The PPHT method is the method to complement this. The PPHT method converts the coordinates of the (x, y) plane into the coordinates of the (?,?) Plane like the Hough transform method, but does not perform the transformation on all the edge pixels, And traces the line segment probabilistically by finding the maximum value among the values larger than the threshold value. That is, when the arbitrarily selected pixel is converted into the Hough coordinate system, the PPHT method calculates ρ and θ from Equation (1) using the sine value and the cosine value, and calculates an accumulator corresponding to the calculated (ρ, θ) ), And accumulates (voting) the frequency. And, in the PPHT method, if the frequency value stored in the accumulator is larger than the threshold value, it is determined to be a straight line and starts tracking. At this time, the PPHT method predicts and predicts the point at which the next pixel appears on a straight line by using a sine value and a cosine value as a gap given as a transfer factor. Therefore, the PPHT method reduces the amount of computation compared to the Hough transform method and determines the two end points of the straight line by probing the straight line as one line segment.

However, the PPHT method does not recognize a straight line if there is no pixel at the predicted position or if it is located at the immediate coordinate. This is because the difference of about one pixel can not be distinguished by the naked eye, and therefore, the line segment is not recognized. Further, in the case of the PPHT method, when there are many edge components, if a pixel at another edge exists at a predicted position, it is recognized as a straight line, and there is a problem that straight lines are recognized as straight lines.

The PPHT method has many researches for improving the straight line detection rate of the SHT and PPHT methods or for improving the computation speed, such as a method of giving a weight to the accumulated value by changing the voting method, However, there is no improvement in the method of determining the final segment determination after viewing the image.

FIG. 2 is a flowchart illustrating a method of recognizing a line segment according to an embodiment of the present invention. FIG. 3 is a diagram illustrating a distribution of pixels according to an angle in a straight line, FIG. 5 is a diagram showing an example of linear verification of line segments and additional tracking of broken line segments. Hereinafter, a line segment recognition method according to an embodiment of the present invention will be described with reference to FIG. 2, with reference to FIGS. 3 to 5. FIG.

The segment recognition method according to the embodiment of the present invention uses an improved PPHT method. In the binarized image through the search process of the edge region, the pixels constituting the straight line have a certain rule according to the slope of the straight line. Using this characteristic, the line segment recognition method according to the embodiment of the present invention identifies an accumulator whose frequency value is larger than the threshold value from the edge area as in the existing PPHT method, finds a straight line, and confirms the slope change of the found straight line It is possible to track the line segment and determine whether the line segment tracked through the verification process is straight. 2, the line segment recognition method according to the embodiment of the present invention may include a linear line segment tracking step S210, a straight line verification step S220, and an additional tracking step S230.

Step S210 is a step of tracking the line segment using the characteristic of the straight line.

The start of line segment tracking is the same as the existing PPHT method. That is, the voting process is performed through the Hough transform on the pixel values in the edge region, and the segment tracking starts when the accumulated frequency value is larger than the predetermined threshold value. In this case, when calculating a tangent value, a method of recognizing a line segment according to an embodiment of the present invention does not directly calculate all the angles but adds only the difference with respect to the previous angle, .

In a digital image, a straight line is a set of consecutive pixels. That is, there is at least one pixel in the neighborhood of one of the horizontal, vertical, and diagonal directions with respect to the starting pixel. Once the direction of the straight line is determined, the neighboring pixel is present in one of the three pixels, and the position of the neighboring pixel can be determined by the slope of the straight line.

For example, referring to FIG. 3, when a straight line is drawn according to an angle between a straight line and a horizontal line in units of 1 degree, it can be seen that? Or? -1 pixels are continuous. That is, considering the straight line between 0 and 90 degrees, pixels are continuous in the x-axis direction at 45 degrees or less, and pixels are continued in the y-axis direction at 45 degrees or more. 3 shows the distribution of pixels when the straight line has an angle of 19 to 26 degrees. As shown in FIG. 3, it can be seen that a new pixel is displayed every two or three consecutive pixels in the x-axis direction (i.e., .alpha. = 3) and the value of the y-axis increases by one. Thus, the number of consecutive pixels (alpha or alpha-1) may vary depending on the angle between the straight line and the horizontal line. The number of consecutive pixels (? Or? -1), which is a characteristic value of a straight line according to the angle of the slope of the straight line, can be expressed as shown in Table 1 below.

Angle [˚] One 2 3 4 5 6 7 alpha 63 30 20 15 12 10 9 Angle [˚] 8 9 10 to 11 12-14 15 ~ 18 19-26 27 to 45 alpha 8 7 6 5 4 3 2

Using Table 1, it is not necessary to store or calculate the tangent (= sine / cosine) value having a large computational complexity in advance. That is, in the existing PPHT method, a tangent value is required for Hough transform of a pixel at a predicted point in a tracking process. However, the line segment recognition method according to an embodiment of the present invention tracks a straight line using Table 1, Tangent values are not needed in the tracking process.

That is, in the line segment recognition method according to the embodiment of the present invention, by using the alpha value corresponding to the angle [theta] calculated by Hough transform after confirming Table 1, You can trace. At this time, the number of consecutive pixels can be both α and α-1. A complete straight line is arranged by a certain rule of alpha or alpha-1 pixels. If this rule is followed strictly, the same result as the existing PPHT method can be produced. However, since the detection result of the edge region can not express a line segment as a perfect straight line, it is not easy to satisfy 100% of the above-mentioned rule if the portion actually viewed as a straight line is examined in detail on a pixel basis. In the case of the existing PPHT method, some errors are allowed by skipping some pixels through the gap, but the width is not large. On the other hand, the line segment recognition method according to the embodiment of the present invention accepts both alpha or alpha-1 pixels irrespective of such a rule, thereby widening the allowable range of the error. This allows straight line recognition even when a pixel is missing or added due to the influence of noise, so that line segment recognition can be performed even if the detection result of the edge area is much affected by noise.

For example, an example of a linear recognition result is shown in Fig. 4 (a) shows an original image, (b) shows a binarized image after edge region detection, and (c) and (d) show pixel images of a line segment. As shown in FIG. 4, the difference of one or two pixels is not classified by the naked eye, but it affects the process of recognizing a straight line, which causes a decrease in the line recognition rate.

Step S220 is a step of performing linear verification of the detected line segment.

As described above, a complete straight line appears according to a predetermined rule of alpha or alpha-1 pixels depending on the slope (or the angle between the straight line and the horizontal line). That is, in the line segment recognition method according to the embodiment of the present invention, the number of consecutive pixels is allowed to be either? Or? -1 for error tolerance in line segment tracking. Because of this error tolerance, a gently curved curve can be recognized as a straight line. In order to solve this problem, when the line segment tracking is completed, a pixel of the detected line segment is selected as a preset interval (threshold 2), and a check is made as to whether or not the line segment is based on the slope variation amount between the intervals. At this time, when the gradient change amount is within the error range, it can be recognized as a straight line. If the position of the last pixel to be verified is not the end point of the line segment after confirming the straight line at the constant interval, since this pixel is located outside the end point of the line segment, the slope is again verified at the end point of the line segment, The error can be reduced. And, in the case of an initially-visited line segment, it can be verified by moving one pixel at the boarding point.

For example, a linear verification process of a line segment is shown in Fig. Referring to FIG. 5, a straight line is traced starting from the P0 coordinate (1) according to a maximum boarding angle (?) Characteristic of a straight line with respect to a point set to a predetermined number (threshold1) in the Hough space The slope change amount can be checked at a preset interval (threshold2) to verify whether it is a straight line (2). If the end point P2 is located within the last verification interval, it can not verify the last segment of the line segment. Therefore, the slope change amount is checked by moving the threshold value 2 in the reverse direction starting from the end point (③). In this process, if the gradient change is out of the allowable range (for example, 1 pixel), the end point of the line segment is changed to a point that finally maintains a straight line because it is not a straight line. To do this, move backward from the end point P2 one pixel at a time, and perform the verification again to check the condition of the straight line. The pixel identified as having the same slope becomes the end point of the line segment.

Step S230 is a step of further tracking the broken line segment.

In the process of detecting the edge region, a line segment of the original image may not be detected due to noise or the like, resulting in binarization in a state in which the intermediate portion is broken. In order to connect the broken line segment, after the straight line verification of the first detected line segment is completed, the broken line segment is tracked.

For example, an additional tracking process of broken line segments is shown in FIG. Referring to FIG. 5, it is checked whether or not a pixel exists at a position having the same slope from P3, which is a point skipped by a predetermined threshold (threshold 3) in P2 (step 4). If a pixel is found, (5). Then, it is verified by linear verification whether the detected length and slope of the additional line segment are longer than or equal to the segment line that is shifted to the threshold 1 (6). If it is confirmed, the additional line segment is recovered from the first detected line segment and the broken line segment, and P4 is determined as the final end point. Finally, when a straight line verification is performed on the section from P0 to P1 (7), the line segment detection is completed. Through the above process, the line segment recognition method according to the embodiment of the present invention can detect and recognize line segments that can not be detected by the existing PPHT method.

FIG. 6 is a diagram illustrating a line segment detection result of a building image using a line segment recognition method according to an embodiment of the present invention.

6 (a) is a detection result image of the binarized edge area calculated using the canine edge area detection method. Figs. 6 (b) and 6 (c) show the result of detecting the line segment by the existing PPHT method using the pixel data of the edge image. 6B shows a case where a parameter is set so as to find a line segment as much as possible. However, a wrong line segment is detected in the lower left portion of the image with strong noise.

6 (b) (c) (d) Number of lines 275 159 207 False positive 89 19 8

In FIG. 6 (b), as shown in Table 2, 275 line segments were detected, but 89 line segments were detected. Adjusting the parameters to reduce these errors will greatly reduce the error, but the error remains and the rate of recognizing the actual line segment decreases.

In Fig. 6 (c), as shown in Table 2, although the number of erroneous line segments is reduced to 19, only 159 actual segment lines are detected.

On the other hand, in FIG. 6 (d) calculated using the line segment recognition method according to the embodiment of the present invention, 207 line segments are detected, and only eight false segments are detected, thereby confirming that the recognition rate is excellent.

FIG. 7 is a diagram illustrating a result of lane recognition using a line segment recognition method according to an embodiment of the present invention.

In the case of the lane recognition, the detection result of the edge area is simpler than that of the building image, so that it is often possible to detect the line segment well by the existing PPHT method. That is, when the road surface is clean, there is almost no noise, and the conventional PPHT method has a recognition rate close to 100%. However, the existing PPHT method has a low recognition rate on the road surface when noises are generated due to shadows, road breakage, partial packaging, or rainy days. Recognition rates were compared by applying the existing PPHT method and the line segment recognition method according to the embodiment of the present invention to a moving image having a lot of noises on the road surface on a clear day and a lot of noise due to raindrops on the windshield in rain. Both videos consisted of 100 frames, and the segment recognition rate was measured for all frames and the average value was calculated. At this time, line segment recognition is performed only for the predetermined region of interest (ROI), so that the image outside the road is not affected.

(C) and (d) show the lane recognition result using the existing PPHT method, and (e) and (d) show the lane recognition result using the existing PPHT method. (f) shows a lane recognition result using the line segment recognition method according to the embodiment of the present invention.

When the lane recognition is performed using the existing PPHT method after detecting the canyon edge pattern from the image as shown in FIGS. 7A and 7B, as shown in FIGS. 7C and 7D, You can see that you do not recognize. On the other hand, if the lane recognition is performed using the line segment recognition method according to the embodiment of the present invention, it can be seen that the lane is correctly recognized as shown in FIGS. 7 (e) and 7 (f).

Table 3 shows the result of checking the line recognition rate applied to the entire moving image.

PPHT Proposed algorithm Clear Rainy Clear Rainy Line Detection 83.4% 70.5% 90.5% 85.2% FP 43.8% 40.3% 13.2% 10.3% FN 11.0% 11.8% 4.0% 6.1% Lane Detection 65% 76% 72% 81%

As shown in Table 3, 70.5% and 83.4% of the conventional PPHT method and 85.2% and 90.5% of the line segment recognition method according to the embodiment of the present invention, respectively. At this time, the most recent improved method of the existing PPHT method was applied.

In addition, when the line segment recognition method according to the embodiment of the present invention is used, the line segment detection rate, that is, the false positive (FP), which is a disadvantage of the existing PPHT method, is reduced by about 30%, and the false negative FN) also decreased by 5% and 7%, respectively.

Especially, when raindrops are frequently seen in the images, the recognition rate is greatly reduced and the false detection rate is increased by applying the existing PPHT method. The recognition rate of the raindrop image using the existing PPHT method was 76%, whereas the accuracy of the segment recognition method according to the embodiment of the present invention was improved to 81%. In this case, the reason why the car is recognized better on the rainy day is because the image on the clear day has a very low recognition rate due to a lot of noise components on the road, the rainwater on the front window is the largest noise component, If you overcome the effects of rain because of the video, it is because rainfall is better for lane recognition. That is, if the rainwater does not make a lot of linear components and overcomes the effect of rainwater, the linear recognition rate can be determined according to the road condition.

FIG. 8 is a diagram schematically illustrating a configuration of a line segment recognition apparatus according to an embodiment of the present invention.

Referring to FIG. 8, a line segment recognition apparatus according to an embodiment of the present invention includes a processor 810, a memory 820, an input unit 830, and a display unit 840.

The processor 810 may be a CPU or a semiconductor device that executes processing instructions stored in the memory 820. [

Memory 820 may include various types of volatile or non-volatile storage media. For example, the memory 820 may include ROM, RAM, and the like.

For example, the memory 820 may store instructions for performing a segment recognition method according to an embodiment of the present invention.

The input unit 830 is a means for receiving various commands, information, and the like from a user for controlling the line segment recognition apparatus or executing an application installed in the line segment recognition apparatus. The input unit 830 may include at least one key button, and may include a touch input module formed integrally with the display unit 840.

The display unit 840 is a means for displaying various data input to the line segment recognition device or stored in the line segment recognition device in the form of time information. For example, the display unit 840 may be a liquid crystal display (LCD).

On the other hand, the components of the above-described embodiment can be easily grasped from a process viewpoint. That is, each component can be identified as a respective process. Further, the process of the above-described embodiment can be easily grasped from the viewpoint of the components of the apparatus.

In addition, the above-described technical features may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and constructed for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

It will be apparent to those skilled in the art that various modifications, additions and substitutions are possible, without departing from the spirit and scope of the invention as defined by the appended claims. Should be regarded as belonging to the following claims.

810: Processor
820: Memory
830:
840:

Claims (8)

In the segment recognition method,
(a) converting a coordinate of an arbitrary pixel among pixels of an edge region found in an image to a Hough space using Progressive Probability Hough Transform (PPHT) to detect a linear component;
(b) performing line segment tracking on the detected straight line component using a preset continuous pixel number range; And
(c) performing linear verification of a line segment detected through the line segment tracking,
Wherein the continuous pixel number range is a property value preset in accordance with an angle corresponding to a slope of a straight line,
The step (b)
Determining whether or not the pixels constituting the straight line appear continuously in the direction of the first axis which is one of the horizontal axis and the vertical axis within the continuous pixel number range according to the angle corresponding to the slope of the straight line, When the value of the second axis is incremented by 1, whether or not the second axis continues to be continuously displayed in the direction of the first axis within the continuous number of pixels.
The method according to claim 1,
The number of consecutive pixels may be α (natural number) or α-1,
And the alpha is the characteristic value.
The method according to claim 1,
The step (c)
Selecting a pixel of the detected line segment at a first predetermined interval, and verifying whether the detected line segment is a straight line by checking a slope change amount between the first intervals.
The method of claim 3,
The step (c)
If the end point of the detected line segment is located within the last validation period, checking the slope change amount by moving in the opposite direction from the end point by the first interval; And
Further comprising the step of changing the end point of the line segment to a point at which the line is finally maintained if the inclination change amount is out of a preset allowable range.
The method according to claim 1,
(d) tracking the detected line segment and the broken additional line segment.
6. The method of claim 5,
The step (d)
Confirming whether a pixel exists at a position having the same slope from a point skipped by a second predetermined interval at the end point of the detected line segment; And
And tracking the additional segment by performing the segment tracking if a pixel is found at a position having the same slope.
The method according to claim 6,
The step (d)
Confirming whether the slope of the additional line segment is the same as the line segment being viewed; And
And if it is the same, restoring the broken line segments connected to the detected line segment, and determining the end point of the additional line segment as the final end point.
A segment recognition apparatus comprising:
A memory for storing instructions; And
And a processor for executing the instruction,
Wherein the command comprises:
(a) converting a coordinate of an arbitrary pixel among pixels of an edge region found in an image to a Hough space using Progressive Probability Hough Transform (PPHT) to detect a linear component;
(b) performing line segment tracking on the detected straight line component using a preset continuous pixel number range; And
(c) performing a linear verification of a line segment detected through the line segment tracking,
Wherein the continuous pixel number range is a property value preset in accordance with an angle corresponding to a slope of a straight line,
The step (b)
Determining whether or not the pixels constituting the straight line appear continuously in the direction of the first axis which is one of the horizontal axis and the vertical axis within the continuous pixel number range according to the angle corresponding to the slope of the straight line, When the value of the second axis, which is one of the first axis and the second axis, increases by one, whether or not the second axis continues to appear continuously within the number of consecutive pixels in the direction of the first axis.

Images