JP3296242B2 - Automatic ruled line detection apparatus and method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a ruled line automatic detection apparatus and method for automatically detecting a ruled line drawn to indicate a processing position in an automatic processing facility such as an NC drilling machine.

[0002]

2. Description of the Related Art In a machine factory line, processing equipment to which NC control technology is applied is being automated. In order to achieve labor saving and rationalization of lines, it is necessary to automate operator intervention work and eliminate the need for monitoring. For example,
When it is necessary to determine the drilling position from the ruled line in the NC drilling equipment, or when determining the origin of the processing position in the material applied to the processing machine, the operator draws the previously drawn ruled line visually with the operator. It was necessary to perform the work of accurately determining and specifying the position.

[0003] A scribing line formed as a groove on an iron member is recognized as a bright straight line or curve because a new surface in the groove is strongly reflected when illuminating light is applied. However, the surface of the member on which the scribe line is drawn may be disturbed when detecting the scribe line due to the presence of a large number of bright lines, points, or areas, depending on the state of scratches and external light. In some cases, it is not possible to accurately separate the scribed line portion from the surface of the member by ordinary image processing because the contrast between the image and the reflection portion due to the groove is not clear. Furthermore, depending on the ruled line, even when illuminated, it may appear as a line darker than the member surface. In such a case, overlooking occurs when an attempt is made to detect a ruled line based on brightness.

[0004] Further, due to uneven marking, rust or dirt generated on the surface of the member or the marking line, lighting conditions, etc., a bright marking image is not continuous or the marking line is partially blurred. If it disappears or disappears, the ruled line cannot be easily detected without human judgment. With such a simple device, it was difficult to accurately recognize the ruled line. Therefore, there was no suitable ruled line detecting device for automatically detecting the ruled line, and work by human power was indispensable.

[0005]

SUMMARY OF THE INVENTION An object of the present invention is to provide an automatic ruled line detecting apparatus and method which can more reliably detect a ruled line marked on the surface of a member.

[0006]

Means for Solving the Problems] To solve the above problems, scribed line automatic detection device of the present invention, imaging camera and irradiation
Comprising a light device and an image processing equipment, light of the ruling in writing is scribed line
Applying a bright light, high lightness pixels and shadow representing the received portion of light within at least two scribe lines are classified by the threshold ruling document which receives light according to the brightness of the pixels constituting the imaging screen taken by a camera Select pixels with low brightness representing the part
It is characterized in that a ruled line is extracted by confusing these selected pixels and performing connection processing.

Note that the two thresholds for classifying according to the brightness may be determined to be appropriate values from a histogram relating to the brightness taken for the pixels of the actually captured image screen. Since the image processing extracts ruled lines, it finds a straight line or a curve that best fits a large number of point position information, and a statistical line approximation method is used. In the present specification, this image processing is
Also called. Known methods such as Hough transform and least squares method can be used as the statistical line approximation method. In particular, when the Hough transform is used, the amount of calculation is small and accurate detection is possible. Further, the intersection of the ruled lines can be detected by detecting a straight line component and a straight line orthogonal to the straight line component by a statistical line approximation method, in particular, a Hough transform.

[0008] In addition, scribed line automatic detection method of the present invention, ruffled
The document is illuminated, the image of the ruled document is imaged by an imaging camera, and the pixels constituting the imaged screen are divided according to brightness , and the light within the ruled line
Is extracted by performing concatenation processing by confusing a bright part which is a part hit by a dark area with a dark part which is a shadow part . Note that the image processing is preferably performed by a statistical line approximation method, and more preferably performed by using the Hough transform.

According to the automatic scribe line detecting device of the present invention, when a groove of a scribe line formed by shaving the surface of a member is illuminated, a portion which is brightly reflected along the groove and a portion which is darkly shaded are paired. Note that bright and dark areas of the screen shot by the camera are confused and image processing is performed, so bright and dark areas are paired along the ruled line Of course, in various cases, such as when the scribe line looks just bright or just dark, or when these are mixed, the image of the surface of the member surface with intermediate brightness is eliminated and the scribe line portion is removed. The prominent trajectory information can be appropriately generated.

In the case where a histogram is created from a lightness distribution in a captured image of an object to be actually measured and a threshold value is determined from the analysis result, various environmental conditions around a ruled line change. Also, since the threshold value is determined from the captured image in the changed state, it is possible to accurately extract the ruled line by using an appropriate threshold value according to the situation. In addition, since the original information of the ruled line is generated in various modes as described above, the scratches and stains on the member surface in the screen information become noise, but the ruled line can be accurately recognized by the simple image processing.

In the image processed by the Hough transform, a straight line can be easily detected even from the broken point information, so that the image of the ruled line disappears in the middle and becomes a continuous dot. However, they can be connected to a straight line to restore the correct ruled line. Since the Hough transform can be used for detecting an arc, it is also useful for detecting a ruled line for processing into an arc shape. Also, usually,
In order to indicate a point, the intersection of two orthogonal straight lines is used, but the position of such a point is determined by detecting the most accurate straight line in the Hough transform and detecting the straight line orthogonal thereto. It can be specified by finding the intersection of both.

Further, according to the ruled line automatic detection method of the present invention, a bright portion and a dark portion are removed from the image of the ruled portion taken by the image pickup camera by removing the intermediate brightness portion, and these are confused. The image processing is performed by using the image processing technique, so that even a poor quality ruled line can be accurately extracted by an image processing technique utilizing the fact that bright portions and dark portions are scattered in a straight line or a curved line. When the image processing is performed by the Hough transform, a straight line component can be easily identified and a straight line equation can be calculated. Therefore, not only the straight line machining but also the position determination of the point using the intersection can be performed. Further, since an arc can also be detected, it can be used when machining along the arc.

[0013]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an automatic ruled line detecting apparatus and method according to the present invention. FIG. 1 is a block diagram showing an apparatus configuration of an embodiment of an automatic ruled line detection apparatus according to the present invention. The automatic ruled line detection device of the present invention includes an imaging camera 1, an image processing device 2, and an arithmetic processing device 3.

The image capturing camera 1 captures an image of the object 10 to be converted into a grayscale image, and transmits an image signal to the image processing device 2. The imaging camera 1 may have an illumination device 4 attached around the distal end. The illumination device 4 can control the luminance by the power supply 5. The image processing device 2 stores the image signal received from the imaging camera 1, further separates dark and bright pixels on the screen from other pixels having intermediate brightness, and uses the result as a binary image. Remember. The binarized image is subjected to image processing to extract a ruled line portion. A monitor device 6 is attached to the image processing device 2 to display an imaging screen, a screen in the image processing process, and the like.

The arithmetic processing unit 3 generates a control signal necessary for a machine tool to be applied, for example, an automatic punching machine, based on the ruled line information obtained by the image processing apparatus 2, and generates a control signal (not shown). Transmit to the machine NC controller. The arithmetic processing device 3 can be practically constructed by a personal computer system, and includes input means such as a keyboard and a pointing tool for instructing selection of arithmetic parameters and programs, and a display.

An image of the measurement object 10 on which the ruled lines 11 and 12 are drawn by the imaging camera 1 is converted into a single-color gray-scale image, and a gray-scale image signal is generated for each captured image. The image processing device 2 stores the grayscale image signal received from the imaging camera 1 in an image memory. FIG. 2 is an enlarged perspective view showing a portion of the ruled line portion irradiated with illumination light. Ruled lines 11, 12
The illumination light 13 from the illumination device 4 is applied to the measurement target 10 where
Is irradiated, a part 14 of the ruled lines 11 and 12 drawn as grooves reflects the illumination light 13 and shines brighter than the surrounding surface 10 to be measured. On the other hand, since the ruled lines 11 and 12 are grooves, shadows 15 may be formed along the grooves depending on the condition of the illumination light 13, and lines darker than the surface 10 to be measured may appear. In such a case, the ruled lines 11 and 12 are the clear lines 1
4 and the dark line 15 will be detected.

Further, depending on the state of the ruled lines 11 and 12 and the lighting conditions, when the illuminating light is applied, the entire width or a part of the width of the ruled line shines and no dark line appears. In some cases, the reflection does not occur and the ruled line portion becomes darker than the surroundings. Note that, depending on the types and states of the ruled lines 11 and 12 and the surrounding conditions, the ruled lines can be clearly caught without any additional lighting. As described above, the ruled lines 11 and 12 drawn on the surface of the measurement target 10 are described.
Makes various appearances under the illumination light, but by processing the bright line and the dark line together, the bright line that reflects the illumination light and the dark line that becomes a shadow in the groove of the ruled line In any of the cases where the lines are formed in a pair, the lines are simply bright, and the lines are simply dark, it is possible to catch the lightness change portion caused by the lines.

FIG. 3 is a diagram for explaining image signals stored in the image processing apparatus 2. The image signals are arranged in a memory space 21 corresponding to the pixels of the imaging device 1. In the figure, as can be seen from 22 which is partially enlarged, the density of each pixel is represented by 8 bits and 256 levels. Here, it is assumed that the smaller the numerical value is, the darker the numerical value is, and the brighter the numerical value is. The correspondence between the actual brightness and the numerical value can be largely changed by adjusting the sensitivity of the imaging camera 1 and the corresponding function in the analog-to-digital conversion.

As an image processing procedure, first, a grayscale image signal stored in an image memory is extracted. Next, each pixel in the screen is divided into a bright part, an intermediate part, and a dark part by two thresholds of level 1 and level 2 given in advance based on the brightness.
Divided into two. The bright pixels include many brightly lit portions of the ruled line, and the dark pixels include shaded portions formed along the ruled line. Further, the intermediate portion includes a large number of surface portions to be measured. Therefore, a binary image in which the pixels in the light and dark parts are assigned a value of “1” and the pixels in the middle part are assigned a value of “0” is obtained and stored in the image memory.

FIG. 4 is a drawing showing an example of a binarized image obtained in this way. In the region, the pixel “1” is hatched. Pixels having "1" in the obtained binary image form columns 24 and 25 along the ruled line. If the state of the ruled line is not good, the line connecting the pixels of “1” may be interrupted or a dotted line may be formed. Measurement target 1
In addition to the ruled lines, the bright and dark areas 26
Is present and becomes noise in image processing.

Preferably, the two thresholds are determined from the results of actual measurement. A brightness distribution for each pixel is obtained from a captured image signal to be measured, and a brightness histogram is created. FIG. 5 is a graph showing an example of the brightness histogram. Since the surface of the measurement object that occupies the largest portion in the imaging screen has intermediate brightness, the brightness distribution has a high frequency in the middle. The scribe line constitutes the brightest part and the darkest part depending on the situation, but the area occupying the whole scribe line is at most several percent. Therefore, when the light and dark portions of the brightness histogram are cut out, the two threshold levels 1 and 2 are set so that the bright and dark portions of the ruled line and the shaded portion formed along the ruled line are included in the cut portion. Decide 2.

The image processing device 2 targets a brightly lit part of a ruled line separated from the surface to be measured in the binary image and a shaded part formed along the ruled line by using Hough.
A straight line having the strongest tendency is detected by applying a conversion process or the like. It is very likely that the detected straight line is a ruled line. In the case where a point such as a hole is indicated by an orthogonal point of two ruled lines, the slope of a line orthogonal to the straight line detected by the Hough transform is known. The point of intersection can be determined by conversion.

FIG. 6 is a diagram showing an example of a ruled line detection result obtained in this way. A ruled line 27 is detected, and another ruled line 28 orthogonal to the ruled line 27 can be detected. The intersection of these two ruled lines is the center position of the hole. The Hough transform is a method of performing mapping to a parameter space often used in image processing to determine a line on which a plurality of points on a screen exist, and a type of a line to be detected in an image is predetermined. This is useful when the shape can be expressed in the form of an equation. The Hough transform can be used to detect not only straight lines but also circles and ellipses.

The ruled lines detected by the Hough transform are displayed on the display device 6 and sent to the arithmetic processing device 3.
The arithmetic processing unit 3 creates data necessary for machining, such as NC data, based on the ruled line information, transmits the data to the NC control device of the machine tool, etc., and uses the data for drilling by the machine tool and setting of the initial reference position. Can be

According to the ruled line detecting apparatus of this embodiment, even when the ruled line is not shaded by the irradiation of the illumination light, it is possible to detect a straight line in a brightly lit part due to the reflection at the ruled line. Can be correctly extracted. Also,
Since the pixels in the dark area that do not form the ruled line are distributed over the entire screen, when using a method such as the Hough transform, such pixels have little effect on the straight line detection accuracy.

Even in the case where the ruled line does not reflect the illumination light and becomes darker than the surface to be measured, a straight line in a dark portion can be detected by applying the present invention. Since it is scattered over the entire screen, the ruled line can be detected. Even if zinc flower is applied as is often seen in canning factories without cutting grooves as ruled lines, the ruled line detection method of the present invention is used by utilizing the high brightness of the part of the ruled line. be able to.

It should be noted that the threshold for classifying by lightness may be calculated from the actual measurement result as described above.
More simply, it may be a predetermined value. Further, a brightness histogram may be obtained each time a ruled line is detected, a value suitable for actual measurement may be found from the analysis of the brightness distribution, and used for image processing. In this case, the program may be programmed so that the same threshold value is used if the characteristics of the measurement object or the measurement environment do not change, and that the calculation is automatically performed to deal with a new condition when the characteristic changes.

Further, the lighting device is not limited to the type installed around the imaging camera. When irradiation is performed in a state close to parallel light from a distance, a clear dark line may appear due to the groove of the scribe line. It should be noted that the automatic ruled line detection device of the present invention can also be used to detect a flaw on the surface of a member. This is because a bright part and a dark part often coexist due to illumination, such as a scratch on the iron material surface.

[0029]

As described in detail above, the automatic ruled line detection apparatus and method according to the present invention can automatically detect a ruled line and specify a work position, which are conventionally performed manually. Therefore, by adopting this, it is possible to further promote unmanned operation of the machine tool.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing one embodiment of a ruled line automatic detection device of the present invention.

FIG. 2 is an enlarged perspective view showing a portion where a ruled line portion is irradiated with illumination light.

FIG. 3 is a diagram illustrating an image signal stored in the image processing apparatus according to the embodiment.

FIG. 4 is a diagram illustrating an example of a binarized image obtained by the image processing apparatus according to the present embodiment.

FIG. 5 is a graph illustrating an example of a brightness histogram obtained by the image processing apparatus according to the embodiment.

FIG. 6 is a diagram illustrating an example of a ruled line detection result obtained by the ruled line detection device according to the present embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Imaging camera 2 Image processing device 3 Arithmetic processing device 4 Illumination device 5 Power supply 6 Monitor device 10 Measurement object 11, 12 Ruled line 13 Illumination light 14 Bright line portion 15 Dark line portion 21 Memory space 24, 25 Clear line 27, 28 Ruled line Line 29 intersection

──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Takao Kanamaru 118 Notsuka, Noda-shi, Chiba Kawasaki Heavy Industries, Ltd. Noda Plant (72) Inventor Sadao Kubo 118, Futatsuka, Noda-shi, Chiba Noda Kawasaki Heavy Industries, Ltd. (56) References JP-A-62-239002 (JP, A) JP-A-3-82940 (JP, A) JP-A-6-119447 (JP, A) (58) Fields investigated (Int. Cl. ^7, DB name) G06T 1/00 G06T 7/00 B23Q 17/24

Claims (7)

(57) [Claims] 1. An imaging camera, an illumination device, and an image processing device, wherein the illumination device illuminates a ruled document having a ruled line, and the imaging camera captures an image of the ruled document to acquire an imaged screen. , the shadow of the image processing apparatus in high lightness pixels and the scribed line that represents the surface exposed to light within the scribed line divided by two threshold in accordance with the brightness of the pixels constituting the imaging screen By selecting pixels having low lightness representing the surface and assigning these selected pixels to the same logical value, the pixels are confused and connected, and a straight line or an arc portion is extracted. An automatic ruled line detection device characterized by detecting. 2. The ruled line automatic detection device according to claim 1, wherein the two thresholds are determined by using a brightness histogram of image information of the captured image. 3. The apparatus according to claim 1, wherein the connecting process is performed by a statistical line approximation method. 4. The ruled line automatic detection apparatus according to claim 3, wherein the intersection of the ruled line is detected by detecting a straight line component and a straight line orthogonal to the straight line component by the statistical line approximation method. 5. The automatic ruled line detection apparatus according to claim 3, wherein the statistical line approximation method is a Hough transform. 6. A ruled line having a ruled line is illuminated, an image of the ruled line is captured by an imaging camera, and a pixel forming an image screen is defined by a light portion representing a surface of the ruled line which is exposed to light according to brightness. The dark part representing the shadowed surface in the ruled line is divided, and the bright part and the dark part are set to the same logical value.
A method for automatically detecting a ruled line, wherein the ruled line is detected by extracting a straight line or an arc portion by confusing and connecting processing by assigning a price. 7. The automatic ruled line detection method according to claim 6, wherein the connecting process is performed by a statistical line approximation method.