JP2880041B2 - Edge detection 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 method for detecting an edge of an object from a one-dimensional image captured by an image input device.

[0002]

2. Description of the Related Art Conventionally, a one-dimensional CCD camera is used as an image input device in order to measure the shape and dimensions of a target object from a photographed image, and the image is continuously processed by swinging the camera to perform image processing. Is disclosed, for example, in JP-A-61-209305.

In such an apparatus, first, an edge of an object is detected from the brightness of a photographed image to recognize a shape, and a distance between edges is calculated to obtain each dimension. For example, when a target is a red-hot steel material going from a heating furnace to a rough rolling step, a difference in output (brightness level) between adjacent pixels of infrared image data taken into the image processing device is calculated from initial coordinates to final coordinates. The point at which this difference is maximum is defined as the start edge of the steel material, and the point at which the difference is minimum (negative value) is defined as the end edge.

However, when the temperature at the end of the steel material is low, the temperature difference between the steel material and its surroundings is small, and the rise of the light emission luminance level at the edge portion becomes slow, so that it is difficult to specify the edge position. Was. In particular, since the resolution of a captured image has been increased due to the recent development of CCD cameras, it has been more difficult to specify the edge position only by looking at the difference in luminance level between adjacent pixels one after the other.

[0005]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems of the prior art, and its main object is to reduce the difference in luminance level between the edge of an object and its surroundings. However, an object of the present invention is to provide an edge detection method capable of easily specifying an edge position of an object.

[0006]

According to the present invention, there is provided a method for detecting an edge of an object from a one-dimensional image including the object, the method comprising the steps of: Calculating a sum or an integral value of the luminance signal levels of the predetermined width and obtaining a difference between the sum or the integral value of the luminance signal levels of the front predetermined width and the sum or the integral value of the luminance signal levels of the rear predetermined width; This is achieved by providing an edge detection method characterized in that a point at which is maximum is defined as an edge position of the object.

[0007]

As described above, instead of a simple difference between adjacent pixels, a sum of luminance signal levels of a predetermined width (n pixels) before and after a certain pixel is obtained, and then the sum of luminance signal levels on the front side is calculated. The difference between the sum and the sum of the luminance signal levels on the rear side is obtained, and the point having the maximum value is determined as the edge position of the object, whereby the edge can be easily specified.

At this time, if the difference between the luminance level of the object and the surrounding luminance level is large, n may be small, and if the difference between the above-mentioned luminance levels is large, n is increased in accordance with the difference. The processing speed is not unnecessarily reduced.

[0009]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a preferred embodiment of the present invention.

FIG. 1 is a perspective view showing the configuration of an image input apparatus and an image processing apparatus to which the present invention is applied. A thick red slab 2 is transported from a heating furnace to a rough rolling step on a conveyor 1 shown in FIG. 1. In order to determine the dimensions of the thick slab 2, 2 is stopped at a predetermined position, photographed by a camera device 3 as an image input device disposed at a predetermined height above the transport roll 1, and an image signal of the thick slab 2 is transmitted to an image processing device 4. Then, the image processing device 4 performs image processing.

The camera device 3 is disposed, for example, at a height of 21 m away from the thick slab 2 in a measurement room provided above the thick slab 2 in order to avoid the influence of heat from the thick slab 2. And a camera body 7 as an optical system member integrally attached to a drive motor 5 supported by a frame (not shown) and a swing plate 6 fixed to the drive shaft.
And a lens unit 8. One-dimensional CC that can measure the range indicated by the solid line in the figure
D is provided, and a zoom lens, for example, is used as the lens unit 8 so that the initial adjustment can be easily performed at a long focus (for example, a focal length of 170 mm). Actually, a one-dimensional CCD with 4096 pixels
Is set so that the resolution is 1.0 mm per pixel.

When the swinging plate 6 is rotated by the drive motor 5, the camera body 7 and the lens unit 8 are rotated.
Swings in the direction indicated by arrow A in the figure, and inputs an image of the thick plate slab 2 in a predetermined direction at a predetermined pitch in the direction of the swing. At this time, the photographic field looks like an imaginary line. Although not shown, a camera device in which the measurement direction and the swing direction of the one-dimensional CCD are orthogonal to those illustrated in order to measure the length in the direction orthogonal to the scanning direction illustrated in FIG. Is provided.

In practice, a plurality of one-dimensional CCDs may be used in combination to secure the resolution, and the above-mentioned area may be divided and scanned by each CCD.

Next, the procedure for detecting the edge position of the thick plate slab 2 will be described.

In this embodiment, the number of pixels in the x direction is set to, for example, four.
096, and the coordinates (x, y) are (0, 0)-(4
095, 499) (FIG. 2). As an outline of the processing, the coordinates (X, Y) from coordinates (0, Y) to coordinates (4095, Y) are fixed and the coordinates (X , Y), the sum PF of the brightness levels of the n (predetermined width) coordinates on the front side in the x direction including
(X, Y) is obtained, and then x including its coordinates (X, Y)
The sum PR (X, Y) of the luminance levels of the n coordinates on the rear side in the direction is obtained, and the difference D between these PF (X, Y) and PR (X, Y) is obtained.
(X, Y) is obtained. A similar process is performed from the coordinates (0, Y) to the coordinates (4095, Y), and the local maximum value among the absolute values of the difference D (X, Y) is set to the edge of the thick plate slab 2 at the y coordinate = Y. Judge. Note that it is not actually necessary to perform the above-described processing from the coordinates (0, Y) to the coordinates (4095, Y).

The processing procedure in this embodiment will be described below with reference to FIG.
Will be described along the flowchart of FIG. First, when an image is taken into the image processing device 4 (step 1), first, the number n of calculation pixels described later is determined according to the average luminance level of the thick slab 2 (step 2).

Here, in this embodiment, when the average luminance level of the steel material is 70 or more (800 ° C. or more) of 256 stories, the number n of operation pixels is 4, and the average luminance level of the steel material is less than 70 (800 ° C.). ), N is set to 10. However, it goes without saying that the value of the number n of operation pixels may actually be changed according to the color and environment of the steel material. In the case of performing scanning by a CCD, a plurality of output amplifiers are often used, but the output value changes because each amplifier has an individual difference. For example, when two amplifiers are used, the amplifiers used for the odd-numbered pixels and the even-numbered pixels are separated, that is, the output values change between the odd-numbered pixels and the even-numbered pixels. Therefore, by setting the number n of operation pixels to the number of amplifiers or a multiple thereof, a detection error due to the amplifier can be eliminated, and more accurate detection can be performed.

Next, assuming that the variable Y = 1 (step 3),
By shifting the value of X from the coordinates (0, Y) in the plus direction, the PF (X, Y), PR (X, Y), D (X,
Y) to obtain one edge (start edge) (step 4), and then shift the value of X from (4095, Y) in the negative direction to obtain the PF (X, Y), PR
The processing for obtaining (X, Y) and D (X, Y) is performed to obtain the other edge (end edge). At this time, the start edge and the end edge are points where D (X, Y) becomes maximum.

As shown in FIG. 4, D (X,
The processing speed is improved by terminating the operation 200 pixels ahead in the shift direction from Y). These 200 pixels are set values that can be changed. This can be expressed by an equation:

[0020]

(Equation 1)

The calculation of (409) is performed from coordinates (0, Y) to coordinates (409).
The value of X is shifted in the plus direction in the (5, Y) direction, and one edge is obtained from the maximum value of D (X, Y) therein.

[0022]

(Equation 2)

Is calculated by shifting the value of X in the minus direction from the coordinates (4095, Y) to the coordinates (0, Y), and calculating the other edge from the maximum value of D (X, Y). Ask for. Next, the y coordinate is shifted by one (step 6), and the coordinate (0, Y + 1) is shifted to the coordinate (4095, Y +
The same processing is performed up to 1). Thus, the y coordinate =
The above processing is repeated from 0 to 499 (steps 4 to 7), and the edge of the thick slab 2 along the y direction is obtained.

When all edges of the thick plate slab 2 along the y direction are obtained, the X coordinate and the y coordinate are exchanged, and all edges of the thick plate slab 2 along the X direction are obtained (steps 8 to 12), and step 13 , 14 calculate the lengths of the thick plate slab 2 in the X and y directions, display the results (step 15), store them, and end the process.

In this embodiment, digital processing is performed by using a CCD camera as an image input device. However, a normal analog camera is used to replace n pixels with a predetermined width and integrate luminance signals. Needless to say, this is also good.

[0026]

As described above, according to the edge detection method according to the present invention, not a simple difference between adjacent pixels but a simple difference between adjacent pixels, but a difference before and after a certain pixel. After calculating the sum of the luminance signal levels of a predetermined width (n pixels), the difference between the sum of the front luminance signal level and the sum of the rear luminance signal levels is calculated, and the point having the maximum value is determined. By setting the edge position of the object, the edge position of the object can be easily and reliably detected even when the difference between the luminance signal level of the edge of the object and its surroundings is small, and the recognition shape of the object And the reliability of each measured dimension is improved.

[Brief description of the drawings]

FIG. 1 is a perspective view illustrating a configuration of an image input device and an image processing device to which the present invention has been applied.

FIG. 2 is a plan view of the thick slab of FIG. 1;

FIG. 3 is a flowchart showing a procedure for detecting an edge of the thick plate slab of FIG. 1;

FIG. 4 is an explanatory diagram showing calculation results.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Conveyance roll 2 Thick plate slab 3 Camera device 4 Image processing device 5 Drive motor 6 Rocking plate 7 Camera body 8 Lens unit

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Fuminori Nakano 46-59, Nakahara, Tobata-ku, Kitakyushu Nippon Steel Corporation Machinery & Plant Division (56) References JP-A-56-4003 (JP, A) JP, A 60-6806 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G01B 11/00

Claims (2)

(57) [Claims] 1. A method for detecting an edge of an object from a one-dimensional image including the object, comprising: obtaining a sum or an integral value of luminance signal levels of a predetermined width before and after each point along a scanning direction; A difference between the sum or integral value of the luminance signal level of the predetermined width and the sum or integral value of the luminance signal level of the rear predetermined width is determined, and a point where the difference is maximum is set as an edge position of the object. Edge detection method to be featured. 2. The edge detection method according to claim 1, wherein the predetermined width is changed according to an average luminance signal level of the object.