JP3258828B2 - Optical disc identification symbol recognition 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 recognizing characters, and more particularly, to a method for recognizing an identification symbol attached to a ring-shaped mirror surface portion of an optical disk by a reflection film.

[0002]

2. Description of the Related Art An optical disc is produced through a process in which aluminum is deposited as a reflective film on a molded signal forming surface and a protective coating material is applied on the upper surface in order to prevent oxidation of the reflective film. General. An identification symbol for identifying what is recorded on each optical disk is a ring-shaped mirror surface portion 1 formed by a reflection film in a portion adjacent to the signal forming surface 11 as shown in FIG.
2 are provided in an arc. Conventionally, in order to perform image recognition of the identification symbol string 10 to determine the quality of each symbol and to identify the type of the optical disc 1, first, an image of a ring-shaped mirror surface provided with identification symbols arranged in an arc is taken. The ring-shaped image is converted from the polar coordinate system to the XY orthogonal coordinate system with the predetermined position of the ring-shaped image as a starting point, and arranged linearly on the XY coordinates. Then, the identification symbol image is cut out from the converted linear mirror image, the head of the image sequence of the identification symbol is detected, and the size of each identification symbol image is geometrically adjusted to the size of the corresponding model symbol. To be corrected. Then, the corrected identification symbol image and the corresponding model symbol are compared in order from the first symbol, and it is recognized whether or not the degree of coincidence falls within a predetermined allowable range.

[0003]

According to the above-described method for recognizing identification symbols, it is possible to accurately judge the quality or disc type of a disk if it is a uniform typeface and well-formed characters. In the case of printed dot characters, a problem may occur. That is, when the optical disk is rotated facing the inkjet nozzle, and printing is performed in an arc shape, printing was performed according to the printing conditions, such as uneven rotation of the optical disk, variation in the interval between the inkjet nozzles due to distortion of the optical disk, and the like. Various deformations occur in the dot character, and even if there is no problem as the character, the dot character may be rejected from the viewpoint of the matching degree with the model character. In the present invention, in consideration of such a point, even if the print character shape is deformed and individual variations occur, the inclination of the character is detected and corrected for each disk, and the normalization is appropriately performed. An object of the present invention is to provide a method for accurately recognizing an identification symbol of an optical disk.

[0004]

The character recognition method according to the present invention is a character recognition method for recognizing characters arranged on a ring-shaped plane in an arc concentric with the center of the ring. And the center of the ring is detected. Then, starting from a predetermined position of the ring-shaped planar image,
X from the polar coordinate system based on the ring center
The image data is converted to a Y orthogonal coordinate system, arranged linearly on the XY coordinates, and a character image sequence is cut out from the converted linear planar image to detect the head of the character image sequence. Then, for each character image, a main axis of inertia passing through the center of gravity of the image is obtained, and the median of the inclination values of each main axis is calculated. Then, a straight line intersecting at a predetermined coordinate is set as an initial projection axis with an angle obtained by subtracting an angle indicated by a median from 90 ° with a linear array axis of the plane image, and the predetermined coordinate is centered at a predetermined angular pitch from that. And calculates the image length of the character string projected on the projection axis for each angle. Then, the smallest image length of the calculated character string image is detected, and the angle between the projection axis and the array axis at that time is subtracted from 90 °, and this is defined as the inclination angle of the character image. Then, geometric correction is performed using the inclination angle and the size data of each character image, and the density of each character image is interpolated to normalize the character image. The model characters are compared with each other, and if the degree of coincidence falls within a predetermined allowable range, it is recognized as a model character.

[0005]

[0006]

The tilt angle of the identification symbol image is detected for each optical disc, and the identification symbol image is geometrically corrected using the tilt angle and the character size data. Normalization is performed by performing interpolation. Then, each of the normalized identification symbol images is compared with the corresponding model character, and if the degree of coincidence between the two falls within a predetermined allowable range, it is recognized as a model character.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment in which the character recognition method of the present invention is applied to a method for recognizing identification symbols of an optical disk will be described with reference to the drawings. FIG. 1 is a plan view of an optical disk 1, and FIG. 2 is a schematic configuration diagram for carrying out a method of recognizing an identification mark of the optical disk, wherein 1 is an optical disk, 2 is a table on which the optical disk 1 is mounted, and 3 is An imaging unit 4 for imaging the optical disk 1 placed on the table 2 from above is an image recognition device for recognizing an image captured by the imaging unit 3. As shown in FIG. 1, the imaging means 3 is placed at a distance enough to allow the ring-shaped mirror surface portion 12 of the optical disk to enter the field of view 13,
Deploy. Further, a ring-shaped surface light source may be arranged between the imaging means 3 and the table 2 in order to improve the quality of the image to be taken. The image recognizing device 4 detects the center position of the optical disk 1 from the image information captured by the image pickup means 3, recognizes the character of the identification symbol attached to the ring-shaped mirror surface, determines the quality of the character, and identifies the type of the optical disk. And so on.

The operation of this embodiment will be described. The optical disk 1 is mounted on the table 2 by a transfer unit (not shown), and the imaging unit 3 captures an image of the optical disk surface existing within the visual field 13. The captured image is input to the image recognition device 4, and the image recognition operation is started. The image recognizing device 4 firstly sets the ring-shaped mirror surface portion 12 to which the identification symbol is attached.
In order to cut out the reflected light image from another image, a density histogram of the input image is created, and a threshold value at which the reflected light image can be separated is determined. Then, the image is binarized using this threshold value, and a ring-shaped reflected light image 120 including the identification symbol image 100 shown in FIG. 3 is recognized. further,
The center of the ring is determined by a known method.

The image recognition device 4 converts the ring-shaped image from a polar coordinate system based on the center of the ring to an XY orthogonal coordinate system starting from a predetermined position of the ring-shaped image 120, and as shown in FIG. Arrange in a straight line. In the present embodiment, in the ring-shaped image 120 in FIG. 3, the phase of a straight line extending downward from the center of the ring in parallel with the Y axis is defined as a starting point (phase angle 0 °), and a certain point is located at any phase from the starting point. This is converted from the polar coordinate system to the XY orthogonal coordinate system by using the X axis as the X axis and the distance from the center of the ring as the Y axis. Then, the converted linear mirror image 121 is binarized with an appropriate threshold value, the identification symbol image 101 is cut out from the strip-shaped mirror image, and the head of the identification symbol image sequence is detected. In FIG. 4, the start of the image sequence is detected.
Within a predetermined distance to the left of an extracted identification symbol image,
Judgment is made based on whether or not another identification symbol image exists, and when the identification symbol image does not exist, the identification symbol image is set as the head of the image sequence.

The image recognizing device 4 performs a characteristic analysis on the cut-out identification symbol images sequentially from the top. The center of gravity of each character image area is calculated by a well-known method, and the center of gravity is set as the origin, and a straight line having a slope value through which the center of inertia of the image area is minimized (hereinafter referred to as a main axis of inertia). Ask for. The principal axis of inertia of each of the identification symbol images thus obtained indicates the inclination direction of the image as shown in FIG.
This indicates the degree of inclination of each identification symbol image with respect to the rectangular coordinates. (See "Digital Image Processing Optics" by Keiichi Tezuka, Tadahiro Kitahashi and Hideo Ogawa, published by Nikkan Kogyo Shimbun, pp. 121-123)

The image recognizing device 4 obtains the inclination values of the principal axes of inertia of the respective images, and calculates the median thereof. Assuming that the main axis of inertia shown in FIG. 5 indicates the median value, the image recognition device 4 has the linear array axis of the mirror image and the angle φ obtained by subtracting the angle θ indicated by the median value from 90 °. , A straight line intersecting at predetermined coordinates is set as an initial projection axis. Then, the projection axis is rotated about the predetermined coordinates at a predetermined angle pitch from there, and the image lengths L1, L2, and L3 of the identification symbol sequence projected on the projection axis at each angle are obtained. Then, the sum of them is calculated. (The hatched figure on the projection axis indicates the density distribution of each projected identification symbol image.)
FIG. 6 shows a case where the identification symbol image is projected on a projection axis at an angle different from that in FIG. 5, and the lengths L1 ′, L2 ′, and L3 ′ of the identification symbol images projected on the projection axis in this case. Are greater than in FIG. This is because the distribution length of the image area becomes the smallest when the projected image of the image inclined at the angle θ is projected on a projection axis orthogonal to the axis representing the inclination. Utilizing such a property, the smallest one of the sums of the image lengths of the calculated identification symbol strings is detected, the angle φa of the projection axis in that case is specified, and the angle θa is calculated from 90 °. The subtracted angle θa is defined as the inclination angle of the identification symbol image sequence.

The image recognition device 4 normalizes each identification symbol image by applying geometric correction using the inclination angle θa and the size data of each identification symbol image. FIG. 7 is an explanatory view showing this geometric correction, in which the coordinates (x ′, y ′) of one point of the identification symbol image shown in the left diagram (a) and the normalization shown in the right diagram (b) are shown. The corresponding coordinates (x, y) of the image are represented by the following equation. Where a, b, c, d, e,
f is a coefficient determined by a correlation between the inclination angle of both images and the size of the images. (System Research Institute No. 127, published by The Systems Research Institute, Minami Satoshi
Image information processing technology, see pages 20-21)

X = ax ′ + by ′ + c, y = dx ′ + ey ′ + f

According to the above-described relational expression, geometric correction is performed and normalized in order from the head identification symbol image. Normalized identification symbol image and model characters (dictionary pattern in which a standard character font corresponding to the identification symbol has been input in advance)
Are compared with each other, and it is determined whether or not the degree of matching falls within a predetermined allowable range.

Further, even when the geometric correction is performed by the above-described relational expression during the above-described geometric correction, the grid points of the image before correction correspond to the grid points of the image after correction. Therefore, density value interpolation is performed by a known method. As shown in FIG. 11, in the present embodiment, f (i, j) is set so that the density value f (i, j) of the corrected grid point (i, j) corresponds to the image before correction (the captured image side). , J), the density values f (m, n), f (m +
1, (n), f (m, n + 1) and f (m + 1, n + 1) are interpolated to calculate f (i, j) by the following equation. Α and β in the expression are internal division ratios indicating the position of the point (i, j) with respect to the grid of the captured image coordinate system surrounding the point, where α is the position in the X direction and β is the position in the Y direction. It shows the direction position.

F (i, j) = f (m, n) · (1−α) · (1
−β) + f (m + 1, n) · α · (1-β) + f (m, n + 1) ·
(1-α) · β + f (m + 1, n + 1) · α · β

As described above, the image recognizing device 4 determines the density value of each grid point by pixel from the density values of four adjacent pixels on the captured image side corresponding to the grid point at the time of geometric correction of the image. Then, after performing geometric correction and density value interpolation, it is determined whether or not the degree of coincidence with the corresponding model character falls within a predetermined allowable range. It may be.

As described above, the embodiment to which the present invention is applied in recognizing the identification symbol of the optical disk has been described. However, for example, it is easy to recognize the type symbol attached to the ring-shaped flat surface of the bearing. Applicable to In short, the present invention can be applied to various cases in which character strings arranged in a ring shape are image-recognized.

[0019]

According to the present invention, since the inclination angle of the identification symbol image is accurately detected for each optical disc, the identification symbol image can be properly normalized by geometric correction and density value interpolation. Because it is possible, the comparison with the model character can be performed reliably. For this reason, even if the shape of the identification symbol is deformed and individual variations occur, characters can be recognized with high accuracy.

[Brief description of the drawings]

FIG. 1 is a plan view of an optical disc according to an embodiment of the present invention.

FIG. 2 is a schematic configuration diagram when the method of recognizing an optical disc identification symbol of the present invention is performed.

FIG. 3 is a diagram illustrating a reflected light image of a ring-shaped mirror surface portion of an optical disc.

FIG. 4 is an explanatory diagram when a reflected light image of a ring-shaped mirror surface portion is subjected to polar coordinate conversion.

FIG. 5 is an explanatory diagram showing a method for obtaining a tilt angle of an identification symbol image.

FIG. 6 is an explanatory diagram showing another mode in FIG. 5;

FIG. 7 is an explanatory diagram showing a geometric correction method.

FIG. 8 is an explanatory diagram showing a density interpolation method.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 optical disc 10 identification symbol 12 ring-shaped mirror surface 100 identification symbol image 120 reflected light image of ring-shaped mirror surface 2 table 3 imaging means 4 image recognition device

Claims (1)

(57) [Claims] 1. A character recognition method for recognizing a character arranged on a ring-shaped plane in a circular arc concentric with the center of a ring, wherein the ring-shaped plane is imaged, the center of the ring is detected, and a ring-shaped plane image is obtained. With the predetermined position as a starting point, the ring-shaped planar image is converted from a polar coordinate system based on the ring center to an XY orthogonal coordinate system, and arranged linearly on XY coordinates.
The character image sequence is cut out from the converted linear plane image, the head of the character image sequence is detected, and for each character image, the principal axis of inertia passing through the center of gravity of the image is obtained, and the median value of the inclination value of each principal axis Is calculated, and the straight-line array axis of the planar image and the angle obtained by subtracting the angle indicated by the median from 90 ° are used as an initial projection axis, and a straight line that intersects at predetermined coordinates is used as the initial projection axis. The projection axis is rotated around the predetermined coordinates, the image length of the character string projected on the projection axis is calculated for each angle, and the minimum image length of the calculated character string is calculated. Detected, the angle between the projection axis and the array axis at that time is subtracted from 90 °, this is used as the inclination angle of the character image, and geometric correction is performed using the inclination angle and the size data of each character image. Interpolation of density values for character images The character image is normalized by the above method, and each character image after the normalization is compared with the corresponding model character. If the degree of coincidence is within a predetermined allowable range, the character image is recognized as a model character. Character recognition method.