JPH1153564A - Image-processing device and machine-readable recording medium recording rotation angle detection program - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To provide an image processing apparatus for detecting a rotation angle of a processing target from a processing target image obtained by capturing a processing target, without performing matching with many model images having different rotation angles. It is an object of the present invention to provide an image processing device that can quickly determine the rotation angle of an image. SOLUTION: A reference point deriving means 7 for obtaining a reference point of an image.
3 and a plurality of feature points having a predetermined feature are detected from points on the circumference of a predetermined radius r centered on the reference point Ko of the reference image K, and the angle of each feature point centered on the reference point And an angle for calculating an angular deviation of a plurality of points corresponding to the characteristic point from among points on a circumference having a predetermined radius centered on the reference point Mo of the processing target image M with respect to the characteristic point. Deviation detecting means 76.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for detecting a rotation angle of a processing target from a processing target image obtained by capturing the processing target, and a machine readable recording medium storing a rotation angle detection program.

[0002]

2. Description of the Related Art Conventionally, there has been known an image processing apparatus which picks up an object to be processed by an imaging device, compares the obtained image to be processed with a reference image registered in advance, and inspects a defect or the like in appearance of the object to be processed. Have been.

[0003] In order to compare the two images, it is necessary to perform rotation correction, position shift correction, and the like on the processing target image so that the processing target image and the reference image coincide with each other within the field of view of the imaging, and then compare the two images. is there.

In a conventional image processing apparatus, in order to detect a rotational deviation of an image to be processed, a reference image is rotated at a predetermined angle, a model image is generated at each rotation angle, and the image to be processed and each rotation angle are generated. With the model image of
When the degree of coincidence with the model image at a certain rotation angle is high, it is recognized that there is a processing object in that angle direction. Then, rotation correction is performed based on the angle so as to match the reference image.

[0005]

In the above-mentioned prior art, since it is necessary to compare the image to be processed with the model image for each rotation angle, it takes a long processing time and a large number of model images can be stored. There are inconveniences such as the need for a memory device with a large storage capacity.

SUMMARY OF THE INVENTION The present invention has been made to solve such a conventional problem, and an image processing apparatus capable of quickly obtaining a rotation angle of an image to be processed without performing matching with a large number of model images. It is another object of the present invention to provide a machine-readable recording medium recording a rotation angle detection program.

[0007]

According to a first aspect of the present invention, there is provided an image processing apparatus comprising: a reference point deriving unit for obtaining a reference point of an image; Feature point detecting means for detecting a plurality of feature points having a predetermined feature from among the points, and obtaining an angle of each feature point with respect to a reference point; And an angle deviation detecting means for calculating an angle deviation of a plurality of points corresponding to the characteristic points from among the points on the circumference with respect to the angles of the characteristic points.

[0008] An image processing apparatus according to a second aspect of the present invention comprises:
In the invention according to claim 1, the feature point is a point whose density value is larger or smaller than a predetermined threshold value, and a plurality of points on the processing target image corresponding to the plurality of feature points have a maximum density value of each point. Or it is the point of the angle which becomes the minimum.

According to a third aspect of the present invention, there is provided an image processing apparatus comprising:
In the first aspect of the present invention, the characteristic point is a difference value between a density value of a point at a certain angle and a density value of a point at an angle near the certain point.

An image processing apparatus according to a fourth aspect of the present invention comprises:
According to the first aspect of the present invention, a plurality of circles having a predetermined radius are set concentrically, and a feature point is detected from points on each circumference.

According to a fifth aspect of the present invention, there is provided an image processing apparatus comprising:
In the present invention, a reference point and a circle having a predetermined radius are moved in an image, and a feature point is detected at each moving position.

An image processing apparatus according to a sixth aspect of the present invention comprises:
According to the first aspect of the present invention, there is provided a shift correcting means for detecting the angular deviation as a shift amount of the image to be processed in the rotation direction and performing rotation correction.

According to a seventh aspect of the present invention, there is provided a machine readable recording medium having recorded thereon a rotation angle detecting program, comprising: a step of deriving a reference point of a reference image; A procedure for detecting a plurality of feature points having a predetermined feature from among points on the circumference and obtaining an angle around the reference point, a procedure for deriving a reference point of the processing target image, and a reference point for the processing target image And detecting a plurality of points corresponding to the plurality of feature points from among points on a circumference having a predetermined radius around the center of the circle and calculating an angle deviation with respect to the angle of the feature point.

According to these inventions, among points on the circumference centered on the reference point of the reference image, a point whose density value is larger or smaller than the predetermined threshold value or a density value of a nearby point is determined. A plurality of feature points satisfying conditions such as a point having a large difference value are detected, and a plurality of points satisfying an angular relationship between the plurality of feature points are detected from the circumference of the processing target image. The rotation angle of the processing target can be detected from the angle deviation with respect to the feature point. Then, the detected angle deviation is detected as the amount of deviation of the processing target image in the rotation direction, and the rotation can be corrected to match the reference image.

[0015]

FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention, which is used for a visual inspection for inspecting a print such as a date attached to a can. In FIG. 1, an image obtained by imaging an object to be processed such as a can by an imaging unit 1 such as a video camera is output as a video signal, converted into digital image data by an A / D conversion unit 2, and stored in an image memory 3. It is memorized.

The image data output from the A / D converter 2 or the image data stored in the image memory 3 is
The data is converted into an analog video signal by the D / A converter 4 and displayed on the screen of the monitor 6 by the display controller 5.

A central processing unit (CPU) 7 controls the entire apparatus, and includes a ROM 71 for storing fixed data such as program data, and a RAM 72 for temporarily storing data such as processing results according to a program. Derivation of the reference point of the reference image and the processing target image, detection of the rotation angle of the processing target image, rotation correction based on the detected angle, detection of the amount of positional deviation from the reference position, position correction based on the detected positional deviation amount, reference Perform comparison processing with the image.

When these processes are represented in a block diagram, a reference point deriving means 73 for obtaining a reference point of a reference image and a processing target image, a density value deriving means 74 for obtaining density values at predetermined positions of the reference image and the processing target image. Feature point detecting means 7 for detecting a feature point of a reference image and obtaining an angle with respect to the reference point 7
5. Angular deviation detecting means 76 for detecting a point corresponding to a feature point from an image to be processed and calculating an angular deviation from a characteristic point of the reference image, and correcting a positional deviation amount and a rotational deviation amount of the processing target image with respect to the reference image. And a recognition processing unit 78 that compares the processing target image after the correction of the shift amount with the reference image to determine the quality of the processing target object.

The CPU 7 has an input unit 8 having a keyboard, a mouse, and the like, a reference image memory 9 for storing a reference image for judging the quality of an object to be processed, and a second image for storing a corrected image to be processed. An image memory 10 and the like are connected.

Although the program data for controlling the CPU 7 is stored in the ROM 71, the program data may be recorded on an external recording medium such as a floppy disk and loaded into the RAM 72 by a reading device (not shown).

Next, registration of the reference image and the reference data performed prior to the inspection will be described with reference to the flowchart shown in FIG. The shift amount of the processing target image is detected based on the registered reference data, and after performing the shift correction, the quality of the processing target image is determined based on the reference image.

First, an object serving as a reference is picked up by the image pickup unit 1 and the obtained picked-up image is stored in the image memory 3 as a reference image K. Further, the reference image K is transmitted via the CPU 7.
Is registered in the reference image memory 9 (step S11).

Next, the coordinates (xo, yo) of the reference point Ko of the reference image K are obtained, and the temporary storage memory 7 in the CPU 7 is obtained.
2 (step S12). The center point or the center of gravity of the reference image K is selected as the reference point Ko. Or if there is a circular pattern that does not change shape even when rotated,
The center point of the circle may be used as the reference point.

Next, while viewing the reference image K displayed on the monitor 6, a circle having a radius r centered on the reference point Ko is set on the monitor 6 from the input unit 8. The setting of the circle is performed so as to pass through a plurality of portions in the grayscale change portion in the reference image K. Further, a gray-scale threshold value t and a step angle α for expressing the gray level in the reference image K at a binary level are set and stored in the RAM 72 (step S13).

After the pre-processing is completed, a reference data registration process is performed. In this process, a density value on the circumference of a radius r centered on the reference point Ko is compared with a threshold value t for each step angle α, and a point having a density value smaller than the threshold value t is detected as a feature point. And the number of feature points.

In the following description, as shown in FIG. 3, an XY coordinate system having an origin at the upper left of the screen, an X axis in the horizontal direction, and a Y axis in the vertical direction is assumed, and the reference point Ko is centered. The description will be made assuming that an axis parallel to the Y axis is rotated clockwise at every step angle α. First, the rotation angle θ is initialized to 0 degrees, and the number n of feature points is initialized to 0 (step S1).
4).

Next, a density value g at a point on the circumference represented by the rotation angle θ is obtained (step S15). The coordinates of a point on the circumference of the radius r are (xo + r · sin θ, yo + r · co
sθ), the coordinates of a point on the circumference are (xo, yo + r) because “θ = 0” in this case.

Next, it is determined whether the density value g is smaller than the threshold value t (step S16). In this example, the black level is low and the white level is high. If the density value g is smaller than the threshold value t, 1 is added to the number n, and the value of the angle θ at that time is set as the storage angle θn ( Step S17). Then, the storage angle θn is stored in the RAM 72 (step S18). If the density value g is equal to or greater than the threshold value t (step S
16), the processing of steps S17 to S18 is jumped.

Next, the increment angle α is added to the rotation angle θ (step S19), and the processing of steps S15 to S19 is repeated until the rotation angle θ reaches 360 degrees (step S20). When the rotation angle θ reaches 360 degrees, the value of the number n of feature points is stored in the RAM 72 (step S2).
1).

In this manner, the density value on the circumference of the radius r is compared with the threshold value t for each step angle α, and the angle θn of the feature point on the circumference having the density value (black level) smaller than the threshold value t, and the feature point Is stored in the RAM 72, and the reference data registration process ends.

FIG. 4 shows a case in which a circle having a radius r is set on the character "ABCD", and the density value on the circumference is set to a threshold value t for each angle α.
FIG. 11 is a diagram in which “•” marks are added to feature points on the circumference having a density value (black level) smaller than a threshold value t as compared with the threshold value t. This example indicates that six feature points have been detected.

Next, the processing for obtaining the rotation angle of the object to be processed will be described with reference to the flowchart shown in FIG. First, the processing target is imaged by the imaging unit 1, and the obtained captured image is stored in the image memory 3 as the processing target image M (step S31).

Next, the coordinates (x1, y1) of the reference point Mo of the processing target image M are obtained and stored in the RAM 72 in the CPU 7 (step S32). The reference point Mo is determined in the same manner as the reference point Ko of the reference image K.
, The reference point Mo is also set as the center point, and the reference point Ko
Is the center of gravity, the reference point Mo is also set to the center of gravity.

Next, the rotation angle θ is set to 0 degree, and the minimum value G of the cumulative density values G of the n feature points detected from now on is set.
min is set to the maximum value, and the minimum value θmin of the rotation angle θ is set to 0 degrees (step S33). Further, the parameter i is set to 1 and the density value G is set to 0 (step S34).

Then, the coordinates (x1 + r ·
sin (θi + θ), y1 + r · cos (θi + θ)
Are accumulated (step S35). These coordinates represent coordinates on the circumference of an angle (θi + θ) obtained by adding the i-th storage angle θi among the n storage angles θ1 to θn to the angle θ. In this case, since “i = 1”, the angle (θ1 +
θ).

The accumulation in step S35 is repeated n times until the parameter i reaches n (step S36).
~ S37). When the accumulation process has been completed n times, it is determined whether the accumulated density value G at that time is smaller than the minimum value Gmin (step S38). If it is smaller, the minimum value Gmin is rewritten to the cumulative density value G, and the angle θ at that time is set as the minimum value θmin (step S39). If it is larger, the process of step S39 jumps.

Next, the step angle α is added to the rotation angle θ (step S40), and the processing of steps S34 to S40 is repeated until the rotation angle θ reaches 360 degrees (step S41). When the rotation angle θ reaches 360 degrees, the rotation angle θ of the minimum value θmin is recognized as the rotation angle of the processing object,
The rotation correction based on the rotation angle θ is performed on the processing target image M. Also, (x1-xo) in the X-axis direction and (y-
1-yo) is performed to make the displacement coincide with the reference image K (step S42). Thereafter, the processing target image M is compared with the reference image K, and an inspection process or the like for inspecting a defect or the like on the appearance of the processing target object is executed (step S43), and the process ends.

The stored angles θ1 to θn of the n feature points detected at the time of registration of the reference image K are added to the rotation angle θ in this way, and the cumulative density value G of the n points θ1 + θ to θn + θ obtained thereby is obtained. It is assumed that the rotation angle θ at which the cumulative density value G is minimum matches the n feature points detected at the time of registration, and the rotation angle θ at that time is detected as the rotation angle of the processing target image M. FIG. 6 shows this state, and shows a state in which the processing target image M is rotated by 90 degrees with respect to the reference image K shown in FIG.

In the above-described embodiment, only one circle having a radius r is drawn and the density value is defined as a function f (θ) of only the angle θ. However, as shown in FIG. Alternatively, the density value may be a function f (r, θ) of the radius r and the angle θ. This can prevent erroneous measurement due to the influence of noise or the like.

If the reference point cannot be determined in the image, the above-described processing may be performed at each moving position while moving the reference point to detect the rotation angle, as shown in FIG.

FIG. 9 is an explanatory diagram showing a specific example for obtaining the reference points Ko and Mo. (A), when the reference image K is circular, the four corners of the circumference are registered as model images K1 to K4, and the processing target image M is registered as the model images K1 to K4.
And a line connecting the center positions of the two images corresponding to the model images K1 and K3, and the model images K2 and K3.
The point at which the line segment connecting the center positions of both images corresponding to No. 4 intersects is set as the reference point Mo of the processing target image M.

(B), when the reference image K is circular, the circumference thereof is registered as a model image K5, and the processing target image M
Is searched in the model image K5, and the detected model image K
The center position of the image corresponding to No. 5 is set as the reference point Mo of the processing target image M.

(C) draws an area K6 indicated by a rectangular dotted line around the reference image K, and sets the position of the center of gravity of this area K6 as the reference points Ko and Mo.

(D), when there is a circular pattern whose shape does not change even when rotated in the reference image K, the circular pattern is registered as a model image K7, and the center point of the circle of the model image K7 is set as the reference point Ko. , Mo.

In the above-described embodiment, the condition (step S16) is that the density value g of the point on the circumference is smaller than the threshold value t in order to obtain a feature point. However, the present invention is not limited to this. Is larger than the threshold value t. In this case, the angle θ at which the cumulative density value G becomes the maximum in the processing shown in FIG. 5 is the rotation angle of the processing target image M.

In the above-described embodiment, the density value g is the density value of a point on the circumference of the rotation angle θ (step S15). The absolute value may be used as the density value g.

The flowchart shown in FIG. 10 shows the processing procedure shown in FIG. 2 in which the density value g is set to a difference value from the density value of the neighboring angle, and the density value of the angle θ is represented by f (θ). “G = | f (θ) −f (θ + α · a) |” (step S15a). Note that a is a natural number (1, 2,...).

It is determined whether or not the difference density value g is larger than the threshold value u (step S16a). If the difference density value g is larger, the angle at that time is stored. In this way, a point where the density value changes significantly can be detected as a feature point.

The flowchart shown in FIG.
Is a method for calculating the rotation angle of the processing target image M based on the feature points detected by the procedure of the flowchart shown in FIG. 5, and is a modified example of the flowchart shown in FIG.

In this processing, since the characteristic point coincides with the characteristic point when the cumulative density value G is maximum, the maximum value Gmax of the cumulative density value G
Is set to the minimum value, and the maximum value θmax of the rotation angle θ is set to 0 degrees (step S33a).

The difference value | f (θi +
θ) −f (θi + θ + α · a) | is accumulated (step S35a). The accumulation in step S35a is repeated n times until the parameter i reaches n (step S3
6 to S37).

Next, when the accumulation process has been completed n times, it is determined whether the accumulated density value G is larger than the maximum value Gmax (step S38a). If it is larger, the maximum value Gmax is rewritten to the cumulative density value G, and the angle θ at that time is changed to the maximum value θmax.
(Step S39a). When the rotation angle θ reaches 360 degrees, the rotation angle θ having the maximum value θmax is recognized as the rotation angle of the processing target, and the rotation correction based on the rotation angle θ is performed on the processing target image M (step S42a).

The flowchart shown in FIG. 12 shows another method for deriving the density value g.
2 is obtained as follows (step S151).

G1 = f (.theta .-. Alpha..a) -f (.theta.) G2 = f (.theta. +. Alpha..a) -f (.theta.) Next, if both the density values g1 and g2 are negative numbers (steps S152 to S153). , And the density value g is set to 0 (step S154). If the density values g1 and g2 are both positive numbers (steps S152 to S153) and the density value g1 is smaller than the density value g2 (step S155), the density value g is set to g1 (step S156). Density value g
If it is 1 or more, the density value g is set to g2 (step S15).
7).

When the density value g is obtained in this manner, only a thin black line (a line thinner than the width α · a) is captured as a feature point, and a large black stain is not captured as noise.
It is strong against such noise, and it is possible to stably detect only the feature points.

The flowchart shown in FIG. 13 shows another method of deriving the density value g.
1, g2 are obtained as follows (step S151)
a). The processing in steps S152 to S157 is the same as the processing in the flowchart shown in FIG.

G1 = f (.theta.)-F (.theta .-. Alpha..a) g2 = f (.theta.)-F (.theta. +. Alpha..a) When the density value g is obtained in this way, the reverse of the case of FIG. Only white thin lines (thinner than the width α · a) are captured as feature points, and white reflected light and the like are not captured as noise. Therefore, it is strong against such noise and it is possible to stably detect only the feature points. it can.

Note that the rotation angle of the processing target image M is obtained based on the feature points detected by the procedure shown in FIG. 12 or FIG. 13, according to the procedure shown in FIG.

[0059]

According to the present invention, a plurality of feature points are detected from points on the circumference centered on the reference point of the reference image, and the image to be processed is determined based on the angular relationship between the plurality of feature points. A plurality of corresponding points are detected from the circumference of the circle, and the angle of the processing target is detected from the angles of these points, so that it is not necessary to compare with the model image for each rotation angle as in the related art. In addition, the processing time can be shortened and the number of memory devices for storing model images can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention.

FIG. 2 is a flowchart illustrating a registration process of a reference image and reference data.

FIG. 3 is an explanatory diagram showing a method for detecting a feature point on a reference image.

FIG. 4 is an explanatory diagram showing a specific example of a method for detecting a feature point on a reference image.

FIG. 5 is a flowchart illustrating a process of detecting a shift amount of an image to be processed.

FIG. 6 is an explanatory diagram showing a specific example of a method for detecting a feature point on a processing target image.

FIG. 7 is an explanatory diagram showing another embodiment for detecting a feature point by drawing a plurality of circles around a reference point.

FIG. 8 is an explanatory diagram showing another embodiment for detecting a feature point at each movement position while moving a reference point.

FIG. 9 is an explanatory diagram showing a specific example for obtaining reference points Ko and Mo.

FIG. 10 is a flowchart illustrating another registration processing of reference data.

FIG. 11 is a flowchart illustrating another displacement amount detection processing of the processing target image.

FIG. 12 is a flowchart illustrating another procedure for deriving a density value g.

FIG. 13 is a flowchart showing another procedure for deriving the density value g.

[Explanation of symbols]

 Reference Signs List 1 imaging unit 2 A / D conversion unit 3 image memory 4 D / A conversion unit 5 display control unit 6 monitor 7 CPU 8 input unit 9 reference image memory 10 second image memory 71 ROM 72 RAM 73 reference point deriving means 74 density value Derivation means 75 Feature point detection means 76 Angle deviation detection means 77 Deviation correction means 78 Recognition processing means

Claims (7)

[Claims] 1. A reference point deriving means for obtaining a reference point of an image, and detecting a plurality of feature points having a predetermined feature from points on a circumference of a predetermined radius centered on the reference point of the reference image. A feature point detecting means for determining an angle of each feature point about the reference point; and a plurality of points corresponding to the feature point from points on the circumference of the predetermined radius centered on the reference point of the image to be processed. And an angle deviation detecting means for calculating an angle deviation of the point with respect to the angle of the characteristic point. 2. The method according to claim 1, wherein the feature point is a point whose density value is larger or smaller than a predetermined threshold, and a plurality of points on the processing target image corresponding to the plurality of feature points are An image processing apparatus characterized in that the cumulative density value is a point at an angle at which a maximum value or a minimum value is obtained. 3. The image processing apparatus according to claim 1, wherein the characteristic point is a difference value between a density value of a point at a certain angle and a density value of a point at an angle near the certain point. 4. The image processing apparatus according to claim 1, wherein a plurality of circles having the predetermined radius are set concentrically, and the feature points are detected from points on each circumference. 5. The image processing apparatus according to claim 1, wherein the reference point and a circle having a predetermined radius are moved in an image, and the feature point is detected at each moving position. 6. The image processing apparatus according to claim 1, further comprising a shift correcting unit that detects the angle deviation as a shift amount of the processing target image in a rotation direction and performs rotation correction. 7. A procedure for deriving a reference point of a reference image, detecting a plurality of feature points having a predetermined feature from points on a circumference of a predetermined radius centered on the reference point of the reference image, A step of obtaining an angle about the reference point; a step of deriving a reference point of the image to be processed; and the plurality of points from among points on the circumference having the predetermined radius centered on the reference point of the image to be processed. Detecting a plurality of points corresponding to the characteristic points and calculating an angle deviation with respect to the angles of the characteristic points.