KR101016821B1 - Marking recognition method and apparatus using the same - Google Patents

Abstract

The present invention is directed to an apparatus and method capable of recognizing text or markings on complex backgrounds.

The marking recognition method according to the present invention includes a first image calculating step of calculating pixel values of an input image using a first set, a second set having a size smaller than the first set, and an input. A second image operation step of calculating a pixel value of the second image using the image and a third image operation of calculating a pixel value of the third image by subtracting the pixel value of the second image from the pixel value of the first image Consists of steps.

The marking recognition apparatus according to the present invention includes a marking recognition unit capable of recognizing a marking and a controller capable of controlling whether the marking recognition unit recognizes a marking according to the marking recognition method.

Since the present invention can accurately recognize characters or markings, the present invention can be utilized in various fields requiring recognition of characters or markings.

Marking recognition, threshold, recognition device, extraction

Description

MARKING RECOGNITION METHOD AND APPARATUS USING THE SAME}

The present invention relates to an apparatus and a method for accurately recognizing marking in a complex background.

Recently, techniques for recognizing specific characters or markings in various backgrounds have been developed. The various backgrounds may be still images, moving pictures, chips, or complex circuits. In particular, the recognition device recognizes the characters or markings displayed on the chip or the complex circuit, so as to know the type of the chip or the circuit and the product characteristics or the manufacturer.

In addition, the producer can use the above information to perform quality control, such as a defect of marking printed on a chip or a circuit produced, selection of defective products, selection of other kinds of products, and the like. Accordingly, techniques for more accurately recognizing the character or marking have been developed.

The present invention is to provide a marking recognition method and apparatus using the same that can accurately extract a character or marking even in a complex background.

According to an aspect of the present invention, there is provided a marking recognition method, comprising: a first set of n * n blocks having a width equal to a width of marking present in an input image; A first image operation step of calculating pixel values of the first image using the input image, a second set having a size smaller than the first set, and a pixel of the second image using the input image A second image calculating step of calculating a value and a third image calculating step of calculating a pixel value of a third image by subtracting the pixel value of the second image from the pixel value of the first image.

The method may further include converting the third image value by using a threshold value transformation and dividing the converted value.

The apparatus may further include a filter for removing noise that is not related to the marking to be recognized.

As described above, by using the marking recognition method according to the present invention, markings existing in a complex background can be easily extracted.

According to an aspect of the present invention, a marking recognition apparatus includes a pixel of a first image using a first set, which is an n * n block having a width equal to a width of a marking to recognize an input image. Calculate a value of the second image using a second set having a size smaller than the first set of input images, and calculate a pixel value of the second image from the pixel value of the first image. And a marking recognizing unit for recognizing a marking by calculating a pixel value of the third image by subtracting the pixel value, and a control unit for controlling whether the marking recognizing unit recognizes the marking.

The control unit may convert the calculated pixel value of the third image by using a threshold transform, and expand the pixel value of the converted third image.

As described above, by using the marking recognition apparatus according to the present invention, markings existing in a complicated background can be easily extracted.

Character recognition method and apparatus according to the present invention configured as described above has the advantage that can accurately extract the character or marking (marking) even in a complex background.

In addition, there is an advantage that can accurately extract the character or marking regardless of the form of the character.

In addition, there is an advantage that can accurately extract the character or marking even when there is uneven illumination.

In addition, there is an advantage that can accurately extract the character or marking even in the noisy environment.

1 is a flowchart illustrating a marking recognition method according to an embodiment of the present invention.

Referring to FIG. 1, the method of recognizing a marking according to the present invention includes inputting an image (S100), and a first set of n * n blocks having a size equal to a width of marking to recognize the input image. calculating a first value using a set) (S110), calculating a second value using a second set having a size smaller than the first set of the input image (S120), and Calculating a third value by subtracting the second value from a first value (S130); converting the third value using a threshold value transformation (S140); marking and relation to be recognized Removing noise (S150); and dividing the noise-removed value (S160). Not all steps of the flowchart shown in FIG. 1 are essential components of the present invention. Thus, there may be more or less steps than the above steps. In addition, it is not limited to the order of the said flowchart.

Mathematical Morphology is a useful image processing technique for analyzing and processing shapes in an image. Basic calculation methods of the mathematical morphology include dilation, reduction, opening, closing, skeleton technique, and the like. It is a technique of expanding the size of an object, text, or marking and reducing the background. The reduction technique is a technique of reducing the size of an object, a text, or a marking, and expanding a background.

According to the marking recognition method according to the present invention, an image including a marking to be recognized is received from the outside (S100). The pixel value of the first image is calculated by using a first set, which is an n * n block having a size equal to a width of marking to recognize the image input from the outside (S110). The pixel value of the second image is calculated by calculating the image input from the outside using a second set having a smaller size than the first set (S120). The pixel value of the third image is calculated by subtracting the value of the second image from the value of the first image (S130). The calculating of the pixel value of the third image may be performed by subtracting the value of the first image from the value of the second image. A method of calculating pixel values of the second image and the third image will be described below with reference to FIGS. 2 to 5.

The pixel value of the third image is converted using a threshold value conversion (S140). For example, the threshold transform converts a value less than a specific threshold T to 0 and a value greater than the threshold T to 255. The numerical values indicate values of contrast. In the embodiment of the present invention, 0 means light white and 255 means darkest black. The meaning of the number may be set in reverse. The threshold conversion may include a variable threshold method, an iterative threshold method, a double threshold method, or the like. .

 Among the pixel values of the converted third image, a noise part not related to a marking to be recognized is removed (or filtered) (S150). The noise means small points existing in portions other than the marking. The process S150 will be described in detail later with reference to FIGS. 2A and 2B.

The pixel value of the third image from which the noise is removed is expanded (S160).

As described above, according to the marking recognition method according to the present invention, it is possible to accurately recognize a character or marking even in a complicated background.

2 to 5 are diagrams illustrating a specific method of calculating pixel values of images in a marking recognition method according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating pixel values of an image 200 input in the filtering process S100 of FIG. 1.

1 and 2, each block of the input image 200 represents a pixel, and numbers in each block represent gray scales. In the following, the numbers are referred to as pixel values. In general, gray scale uses 8 bits to express the intensity of an image. In this case, the intensity of the gray scale is expressed in 256 steps from 0 to 255. In general, 0 means light white and 255 means darkest black. It is also possible to set the meaning of the numbers in reverse. The number of the input image 200 and the size of the block (n * m) is made arbitrarily, the actual number of the image and the size of the block (n * m) can be changed in various ways depending on the situation.

3A and 3B illustrate the process of calculating the first image value S110 of FIG. 1.

1, 2 and 3A, the first set 310 has a size corresponding to the width W of a character or marking to be recognized. In this embodiment, there are 5 * 5 block sizes and there are u1 to u8 and origin o present in each corner of the first set and in the middle of the respective corners. The u1 to u8 and the origin o mean numbers of gray scales corresponding to the image 200 when the first set 310 is matched to the image 200. The matching means a process of matching the origin o of the blocks of the first set 310 to all the blocks of the image 200. In the method of obtaining the pixel value of the first image using the first set 310 after matching the image 200, u1, u2, u3, u4 of the pixel values corresponding to the u1 to u8 It may be calculated using only pixel values corresponding to and may be calculated using all pixel values of u1 to u8. In addition, there is a method for calculating the maximum value, the minimum value or the average value among the corresponding pixel value. In the method of calculating the maximum value, after removing one to four of the largest pixel values among the corresponding pixel values, the maximum value of the remaining pixel values may be calculated. In the method of calculating the minimum value, one to four of the smallest pixel values among the corresponding pixel values may be removed, and then the minimum value of the remaining pixel values may be calculated.

Hereinafter, a specific embodiment will be described. In addition, in the following embodiment, the pixel values corresponding to u1 to u8 of the first set 310 are used, and an average value of the pixel values is calculated to match the origin o of the first set 310. Based on the calculation method for changing the pixel value corresponding to the image 200 to the calculated average value.

For example, a matching 320 is performed such that the origin o of the first set 310 matches the colored portion (corresponding to the pixel value 201) of FIG. 2. In the case of matching 320, pixel values of the image 200 are obtained at positions corresponding to u1 to u8 (u1 = 123, u5 = 138, u2 = 165, u8 = 138, u6 = 140, u3 = 168, u7 = 98, u4 = 138). The average value of the obtained pixel values is obtained. That is, the average value is (123 + 138 + 165 + 138 + 140 + 168 + 98 + 138) / 8 = 138.5. Since the numerical value after the decimal point is discarded in the result, the average value is calculated as 138. If the method of calculating the maximum value was used, the maximum value would be 168. If the method of calculating the maximum value except the two pixel values in order from the large pixel value is used, the maximum value will be 140. If the method of calculating the minimum value was used, the minimum value would be 98. If the method of calculating the minimum value except the two pixel values in order from the small pixel value is used, the minimum value will be 138. The pixel value (corresponding pixel value 201) corresponding to the image 200 matched with the origin o of the first set 310 is changed to an average value 330. If the maximum or minimum value is obtained, the pixel value corresponding to the image (the corresponding pixel value is 201) is changed to the maximum or minimum value obtained above.

After moving one pixel to the right from the position of the first set 310, it matches 330 with the image 200. If the average value is calculated by the same process as the above calculation process, the average value is 146. The average value is changed to an average value obtained by calculating a pixel value (the corresponding pixel value is 138) corresponding to the image 200 matched with the origin o of the first set 310 (330). Further, after moving the origin of the first set 310 one pixel below the colored portion, it matches 360 with the image 200. When the average value is calculated by the same process as the above calculation process, the average value is 134. The average value is changed to an average value obtained by calculating a pixel value (the corresponding pixel value is 112) corresponding to the image 200 matched with the origin o of the first set 310 (370).

Hereinafter, a case in which all blocks of the first set 310 are not included in all of the blocks of the image 200 will be described. When the origin (o) of the first set (310) is matched (380) to a pixel (pixel value is 123) existing at the point (0, 0) of the image 200, the first set (310) and Only parts that match the image 200 are matched. If the average value is calculated using the pixel values (u6 = 138, u3 = 201, u7 = 138) corresponding to the matched image 200, the average value will be 159. The average value is changed (385) to an average value obtained by calculating a pixel value corresponding to the image 200 matched with the origin o of the first set 310 (the corresponding pixel value is 123).

As above, while moving the first set 310, the above calculation is repeated at all positions that can be matched. In the above calculation process, an embodiment in which u1 to u8 are used and pixel values corresponding to u1 to u8 are calculated using an average value is described. However, a method of using u1 to u4 and calculating a maximum or minimum value can be used. It is natural.

FIG. 3B is a diagram illustrating after obtaining all pixel values of the first image using the first set.

Referring to FIG. 3B, a pixel value is changed in all portions where the first set may be matched to change from the input image 200 to the first image 390.

4A and 4B illustrate the process of calculating the second image value S120 of FIG. 1.

1, 2 and 4A, the second set 400 has a size as wide as W / 2 of the character or marking to be recognized. In the present embodiment, since the first set 310 is 5 * 5, the second set 400 has a 3 * 3 block size, and each of the corners of the second set 400 includes v1 to v4 and origin ( o) exists. The v1 to v4 and the origin o mean the number of gray scales corresponding to the image 200 when the first set 310 is matched to the image 200. After matching the image 200, using the second set 400 to obtain a pixel value of the second image, v1, v2 of the pixel value corresponding to the v1 to v4 or the origin (o) , v3 and v4 may be calculated using only pixel values, or may be calculated using only pixel values corresponding to v1 to v4 and origin (o), or may be calculated using only pixel values corresponding to origin (o). have. In addition, there is a method for calculating the maximum value, the minimum value or the average value among the corresponding pixel value. In the method of calculating the maximum value, after removing one to four of the largest pixel values among the corresponding pixel values, the maximum value of the remaining pixel values may be calculated. In the method of calculating the minimum value, one to four of the smallest pixel values among the corresponding pixel values may be removed, and then the minimum value of the remaining pixel values may be calculated.

Hereinafter, a specific embodiment will be described. In addition, in the following embodiment, the pixel values corresponding to v1 to v4 and the origin o of the second set 400 are calculated, and the maximum value of the pixel values is calculated to calculate the origin of the second set 400. Based on the calculation method of changing the pixel value corresponding to the image 200 matched with (o) to the calculated maximum value. The origin o of the second set 400 is matched 410 such that it matches the colored portion (corresponding to the pixel value 201) in FIG. 2. When matched 410, the pixel values of the image 200 are obtained at positions corresponding to v1 to v4 and origin o (v1 = 138, v2 = 120, v3 = 138, v4 = 176, o = 201). If the maximum value of the obtained pixel values is obtained, the maximum value will be 201. The pixel value corresponding to the image 200 matched with the origin o of the second set 400 (the corresponding pixel value is 201) is changed to the maximum value calculated (420).

After moving one pixel to the right from the position of the second set 400, a match 430 is made with the image 200. If the maximum value is calculated by the same procedure as the above calculation process, the maximum value is 135. The maximum value is changed to an average value calculated by calculating a pixel value corresponding to the image 200 matched with the origin o of the second set 400 (the corresponding pixel value is 120) (330). In addition, after moving the origin of the second set 400 one pixel below the colored portion, the second set 400 is matched 450 with the image 200. If the maximum value is calculated by the same process as the above calculation process, the maximum value is 120. The maximum value is changed to a maximum value calculated by calculating a pixel value corresponding to the image 200 matched with the origin o of the second set 400 (the corresponding pixel value is 112) (460).

Hereinafter, a case in which all blocks of the second set 400 are not all included in the blocks of the image 200 will be described. When the origin o of the first set 310 is matched 470 to a pixel (pixel value is 123) present at the point (0,0) of the image 200, the second set 400 Only parts that match the image 200 are matched. If the maximum value is calculated using the pixel values (v3 = 135, o = 123) corresponding to the matched image 200, the maximum value will be 138. The maximum value is changed to a maximum value calculated by calculating a pixel value (the corresponding pixel value is 123) corresponding to the image 200 matched with the origin o of the second set 400 (385).

As above, while moving the second set 400, the above calculation is repeated at all positions that can be matched. In the above calculation process, an embodiment in which the maximum value is calculated using pixel values corresponding to v1 to v4 and the origin (o) is described. However, only the u1 to u4 or the nuclear power source (o) are used to calculate the average value or the minimum value. It is natural to be able to use the method.

FIG. 4B is a diagram showing after obtaining all pixel values of the second image using the second set.

Referring to FIG. 4B, a pixel value is changed in all portions where the second set may be matched, thereby changing from the input image 200 to the second image 490.

FIG. 5 is a diagram illustrating a process of calculating a third image value S130 of FIG. 1.

Referring to FIG. 5, the pixel values of the third image 500 are obtained by using the first image 390 and the second image 490 obtained in FIGS. 4B and 5B.

The pixel values of the third image 500 are obtained by subtracting pixel values of the second image 490 from the pixel values of the first image 390. For example, a (3,3) pixel value (pixel value 138) present in the third column of the third row of the first image 390 in the third column of the third image of the second image 490 Subtract the existing (3,3) pixel value (the pixel value is 201). The result value corresponds to -63, but is changed to 0 when the result value is negative. Thus, the result of the calculation is zero, and the pixel values (3,3) present in the third column of the third row of the third image are zero. If the calculation is performed in the same manner as above, the (4,3) pixel value present in the third column of the fourth row of the third image will be 11. By repeatedly executing the calculation method, the pixel values 500 of the third image may be obtained.

6A and 6B illustrate the process of filtering S150 of FIG. 1.

6A and 6B, in the filtering process S150, a filter is used to remove noise. According to one embodiment of the filter, a frame filter which is a type of sieve filter may be used. The sieve filter is a filter that removes small noises and maintains a long pattern of letters or markings. According to the present embodiment, the frame filter inverts the third image converted in step S140 (100), and converts the inverted third image into a 3 * 3 mask 110 and 5 *. Erosion is performed using a 5 mask 120, a 7 * 7 mask 130, or the like. The frame filter may obtain the filtered third image by subtracting the pixel value of the third image eroded from the pixel value of the third image converted in step S140. Since the size of the frame filter is not constant but varied, it is possible to completely remove the noise by repeatedly performing filtering using a mask having an appropriate size. The filter used in the process of removing noise may use various filters such as a ring filter in addition to the frame filter.

FIG. 7 is a diagram illustrating images according to each process of FIG. 1.

1 and 7, an image according to the image input process S100 of FIG. 1 is illustrated in the image 600 of FIG. 7. In the following calculation method used to obtain a third image, the first set uses all pixel values corresponding to u1 to u8, and the second set corresponds to pixel values and origin o corresponding to v1 to v4. A pixel value was used. In addition, the calculation method using the first set of pixel values performed an average value operation, and the calculation method using the second set of pixel values used a maximum value calculation. In this embodiment, the effect of Gaussian noise with an average of 0 and a standard deviation of? Is added. The image 600 is an image in which the background is very complicated and the letter 'A' exists in the complicated background. A first image according to the first image value operation (S110) of FIG. 1 is illustrated at reference numeral 610 of FIG. 7. The first image 610 is an image obtained by calculating the image 600 using the first set. A second image according to the process of calculating the second image value of FIG. 1 (S120) is shown at 620 of FIG. 7. The second image 620 is an image obtained by calculating the image 600 using the second set. A third image according to the process of calculating the third image value (S130) of FIG. 1 is shown at reference numeral 630 of FIG. 7. The third image 630 is obtained by subtracting pixel values of the second image 620 from pixel values of the first image 610. A third image according to the threshold value conversion process (S140) of FIG. 1 is illustrated at reference numeral 640 of FIG. 7. The threshold-converted third image 640 is obtained by threshold-converting the image 630 obtained in the third image value operation S130. A third image according to the filtering ring S150 of FIG. 1 is shown at 650 of FIG. 7. The third image 650 from which the noise is removed is obtained by removing unnecessary noise by using a filter on the third image 640 obtained in the threshold conversion process S140. A third image according to the expansion (S160) process of FIG. 1 is shown at reference numeral 660 of FIG. 7. The expanded third image 660 is obtained by dilation of the third image 650 obtained in the filtering process S150. In the expanded third image 660, only the desired letter 'A' is clearly displayed in the input image 600.

As described above, according to the present invention, even when a character or marking exists on a complicated background, the corresponding character or marking can be accurately extracted.

8A to 8C are diagrams showing the results of experiments on the undetected rate of a character according to the change of the character in using the marking recognition method according to the present invention.

8A is a diagram showing a background used in this embodiment.

FIG. 8B is a diagram illustrating a change in the undetected percentage (%) of a character by changing the letter engraved on the background of FIG. 8A to 'A', 'C', and 'X'.

In this example, the effect of Gaussian noise with an average of 0 and a standard deviation of? Is added. The SNR means 20 * log (S / σ). S denotes the magnitude of contrast between the character and the background, and σ denotes the standard deviation of the Gaussian noise. In the present embodiment, the S is fixed at 60, and the undetected percentage (%) of the characters was calculated while changing the SNR value. In addition, the present embodiment was carried out on the premise that the brightness of the light illuminating the image is uniform (when there is no difference in brightness). The difference in brightness is caused by the angle of illumination or the object being obscured in obtaining the image.

Referring to FIG. 8B, it can be seen that the undetected percentage (%) of the letter as the letter changes to 'A', 'C', and 'X' falls below 5% when the SNR is 20 (dB) or more. Therefore, when the SNR is 20 (dB) or more, it shows that the character can be detected accurately regardless of the type of the character.

As described above, according to the present invention, it is possible to accurately extract the characters or markings existing on a complex background.

In addition, it is possible to accurately extract characters or markings regardless of the form of the characters.

In addition, under an environment with Gaussian noise, a separate filter may be used to accurately extract characters or markings without removing the Gaussian noise.

FIG. 8C is a diagram illustrating a change in the undetected percentage of a character according to the change of the contrast size S between the character and the background.

In this example, the effect of Gaussian noise with an average of 0 and a standard deviation of? Is added.

Referring to FIG. 8C, when the size of S is 20 or more and the SNR is 20 (dB) or more, it can be seen that the undetected percentage (%) of the letter falls below 5%. Thus, the increase in the contrast between the text and the background shows that the text can be detected accurately.

9A to 9C are diagrams showing results of experiments on the non-detection rate of a character according to a difference in brightness of an image in the marking recognition method according to the present invention.

9A is a diagram showing a background used in this embodiment. The background is bright in the middle and dark in the outer. The background refers to a case in which only a center portion of the image is brightly illuminated by illumination or the like. Unlike FIG. 8A to FIG. 8C, the experiment is performed by adding an effect of illumination having non-uniform brightness.

9b shows an image with the effect of illumination of FIG. 9a added. Thus, the center of the image is brighter than the outside.

FIG. 9C is a view showing a result of performing a marking recognition method according to the present invention using the image of FIG. 9B. In the present embodiment,? H is used as a parameter representing the difference between the pixel value present at the origin and the pixel value present outside. The large value of δH means that the difference between the brightness of the center portion and the outer portion is large. For example, δH 150 means that there is a difference of 150 (gray scale) between the pixel value at the origin and the pixel value at the outside. The larger the value, the greater the difference in brightness. Referring to FIG. 9C, it can be seen that the undetected percentage (%) of the character does not change even if the value of δH is changed.

10A to 10C are diagrams showing results of experiments on the non-detection rate of a character according to the difference in brightness of an image in the marking recognition method according to the present invention.

10A is a diagram showing a background used in the present embodiment. The background has a form in which the left part is brighter and becomes darker toward the right. The background refers to a case where only the left part of the image is brightly illuminated by illumination or the like.

FIG. 10B shows an image added with the effect of illumination of FIG. 10A. Thus, the left part of the image is brighter than the right part.

FIG. 10C is a diagram illustrating a result of performing a marking recognition method according to the present invention using the image of FIG. 10B. In the present embodiment, a slope is used as a parameter of the difference in brightness by using an inclination angle from left to right. The larger the inclination value, the greater the difference in brightness between the left part and the right part. Referring to FIG. 10C, it can be seen that the undetected percentage of the character does not change even if the inclination value changes.

Referring to FIGS. 9C and 10C, when the marking recognition method according to the present invention is used, characters or markings may be accurately detected even in an environment in which an unbalance of brightness generated by illumination or the like exists.

Various embodiments described above can be implemented in software. For example, embodiments such as the procedures and functions described herein may be implemented in separate software modules. Each of the software modules may perform one or more functions and operations described herein. Software code may be implemented in software applications written in a suitable programming language.

11 is a diagram illustrating an embodiment of a marking recognition apparatus according to the present invention.

Referring to FIG. 11, the marking recognition apparatus includes an image data input apparatus 1000, a marking recognition unit 1010, a memory unit 1020, a data transmitter 1030, a display unit 1040, and a controller 1050. It may include.

The image data input apparatus 1000 refers to a device capable of data-forming a portion in which a desired character or marking exists. For example, the image data input apparatus 1000 may mean a camera, a scanner, or the like. The image data input apparatus 1000 converts and stores a portion having a desired character or marking into data, and transmits data corresponding to a request for transmission of the stored data.

The marking recognition unit 1010 recognizes a character or a marking from image data input from the image data input apparatus 1000. The method of recognizing the character or marking uses the same method as described in FIGS. 1 to 9C.

The memory unit 1020 may store characters or markings recognized by the marking recognition unit 1010.

The data transceiver 1030 may transmit data to an external device.

The display unit 1040 may display the recognized character or marking.

The controller 1050 may control the marking recognition unit 1010 to recognize a character or marking in an input image.

In addition, the controller 1050 may control to store the characters or markings recognized by the marking recognition unit 1010 in the memory unit 1020.

In addition, the controller 1050 may control to display the character or marking recognized by the marking recognition unit 1010 on the display unit 1040 or to transmit the character or marking to an external device through the data transmitter 1030. have.

As described above, according to the present invention, it is possible to accurately recognize the characters or markings existing on a complex background.

The mobile terminal described above is not limited to the configuration and method of the embodiments described above, but the embodiments may be configured by selectively combining all or part of the embodiments so that various modifications can be made. have.

In addition, although the technical idea of the present invention has been described in detail according to the above preferred embodiment, it should be noted that the above embodiment is for the purpose of description and not for the limitation thereof. In addition, those skilled in the art will understand that various embodiments are possible within the scope of the technical idea of the present invention.

1 is a flow diagram related to a marking recognition method associated with one embodiment of the present invention.

FIG. 2 is a diagram illustrating an image 200 input in step S100 of FIG. 1.

3A and 3B illustrate the S110 process of FIG. 1.

4A and 4B illustrate the S120 process of FIG. 1.

5 is a view embodying the process S130 of FIG.

6A and 6B illustrate the S150 process of FIG. 1.

7 is a diagram illustrating images according to each process of FIG. 1.

8A to 8C are diagrams showing the results of experiments on the undetected rate of a character according to the change of the character in using the marking recognition method according to the present invention.

9a to 9c are diagrams showing the results of experiments on the undetected rate of the character according to the brightness difference of the image in the marking recognition method according to the present invention.

10A to 10C are diagrams showing results of experiments on the undetected rate of a character according to a difference in brightness of an image in a marking recognition method according to the present invention.

11 is a diagram of an embodiment of a marking recognition apparatus according to the present invention;

Claims (16)

Matching each block of the input image with the origin of the first set, which is an n * n block having a width equal to the width of the marking present in the input image, The input image matched with the origin of the first set after calculating any one of a maximum value, a minimum value, and an average value by using values of pixels existing in a portion where the first set of blocks and the image block match. A first image calculating step of converting a value of a pixel of to a value of the computed pixel; A second image calculation step of calculating pixel values of the second image using a second set and an input image having a size smaller than the first set; and And a third image calculation step of calculating a pixel value of a third image by subtracting the pixel value of the second image from the calculated pixel value of the first image calculation step. delete The method of claim 1, The second image operation step, Matching each block of the input image with the second set of origin, and using the values of pixels present in the portion where the second set of blocks and the image block match, using a maximum, minimum or average value After calculating any one of the following, converting a value of a pixel of the input image matched with the second set of origin into a value of the calculated pixel. The method according to claim 1 or 3, The maximum value calculation, And after eliminating one to four maximum values among the values of the pixels existing within the size of the first set, calculating the maximum value among the values of the remaining pixels. The method according to claim 1 or 3, The calculation of the minimum value is, And after removing one to four minimum values among the values of the pixels existing within the size of the first set, calculating the minimum value among the values of the remaining pixels. The method of claim 1, The values of the pixels present in the portion where the first set of blocks and the image block match, And a pixel value matched with four blocks existing at each corner of the first set. The method of claim 1, The values of the pixels present in the portion where the first set of blocks and the image block match, And a pixel value matched with four blocks existing in each corner of the first set and a pixel value matched with four blocks existing in the middle of the respective corners. The method of claim 3, wherein The values of the pixels present in the portion where the second set of blocks and the image block match, ① the value of the four pixels present at each corner of the first set and the value of the pixel present at the origin, The value of the pixel at the origin; or (3) A marking recognition method, characterized in that any one of the values of four pixels present in each corner of the first set. The method of claim 1, Transforming the third image value using a threshold value transformation; and And dividing the converted value. The method of claim 9, And after the converting step, a filtering step for removing noise irrelevant to the marking to be recognized. The method of claim 1, The second set is a marking recognition method, characterized in that the m * m block having a width of half the width of the marking to be recognized. Matching each block of the input image with the origin of the first set, which is an n * n block having a width equal to the width of marking to recognize the input image, the first set The value of the pixel of the input image matched with the origin of the first set is calculated after calculating any one of a maximum value, a minimum value, or an average value using the values of the pixels in the set of blocks and the image block. Convert to the computed pixel value, Compute a pixel value of the second image by using a second set having an input image having a size smaller than the first set, and subtract the pixel value of the second image from the computed pixel value. Marking recognition unit for recognizing the marking (Marking) by calculating the pixel value of the image; And And a controller configured to control whether the marking recognition unit recognizes the marking. 13. The method of claim 12, The control unit, And converting the computed pixel value of the third image by using a threshold transform and dilatating the pixel value of the converted third image. 13. The method of claim 12, Further comprising a memory unit for storing data, The control unit, And a marking recognition device controlled to store the marking recognized by the marking recognition unit in the memory unit. 13. The method of claim 12, And an image data input device for inputting an image on which a marking to be recognized is displayed to the marking recognizing unit. The control unit, And a marking recognition unit controlling the marking recognition unit to recognize the marking from the image when an image is input from the image data input device. 13. The method of claim 12, The apparatus may further include a display unit for displaying the marking recognized by the marking recognition unit or a data transmitter for transmitting to the outside. The control unit, And a marking recognized by the marking recognizer to be displayed on the display or transmitted to the outside through the data transmitter.

Images