KR100716692B1 - Code recognition method and device - Google Patents

Abstract

When determining whether the color of the cell is white or black from the gradation value of each cell of the image, if the gradation value of the cell is sufficiently biased to be white or black, it is determined by comparison with the threshold, but the gradation value of the cell When is in the gray zone, the color of the cell is determined from the state of the cell around the cell. That is, when the gray value of the corresponding cell is gray and the surrounding cells are all black, the corresponding cell is determined as white, and when the surrounding cells are all white, the corresponding cell is determined as black.

2D code, cell, gradation value, long shutter, data codeword

Description

CODE RECOGNIZING METHOD AND DEVICE

The present invention relates to a two-dimensional code recognition method and a recognition device that read-recognize a two-dimensional code image input from an optical reading device such as a scanner or a CCD camera.

In the conventional two-dimensional code reading device, the two-dimensional code image input from the optical reading device is binarized, and the recognition process is performed using the binarized image (Fig. 14 of Patent Document 1 and Fig. Of Patent Document 2). 1).

In order to perform the recognition processing after converting the image data read from the optical device into binary data, there is a problem that the recognition accuracy is lowered due to the influence of the distortion of the data unless the optical resolution is above a certain value. For example, an isolated cell surrounded by cells of four or eight types of different types (in a two-dimensional code, a grating constituting a black or white point is called a cell) has a low optical resolution or focus. When a misalignment or the like occurs, the gradation value is dragged to an adjacent cell, so that it is converted to another value. Such a problem is a remarkable problem in the portable terminal device having a small camera, since the optical device is small when the two-dimensional code is read using the built-in small camera.

Even when the recognition process is performed without performing the binarization process, the above problem cannot be solved by using the simple binarization method when determining the cell value.

In addition, when photographing an image and performing recognition while moving the recognition apparatus with a camera with respect to the two-dimensional code of a recognition object, the time required for recognition processing compared with the imaging time of one image generally As a result of this long, the number of images that can be displayed per unit time is reduced, thereby degrading the function of the user interface. In this case, the image capturing uses a still image capturing control method, and in particular, in a dark place, in order to similarly improve the sensor sensitivity, a plurality of frames exceeding the image data reading time of one frame of the camera. Multiple exposure (long shutter) of the image was performed to capture the image having the gray level secured. On the other hand, by multiple exposure over a plurality of frames, blurry images are picked up for motion, and as a result, there is a problem that the recognition performance is lowered.

[Patent Document 1] Specification of Japanese Patent Application Laid-Open No. 6-042587 (JP-A-7-254037)

[Patent Document 2] Specification of Japanese Patent Application Laid-Open No. Hei 8-341354 (Japanese Patent Application Laid-Open No. Hei 10-187868)

Isolation cells surrounded by cells of different types (cells with different gradation values such as white cells and black cells) in four or eight directions have a gradation value when the optical resolution is low or when a focus shift occurs. Since it is attracted to an adjacent cell, a problem arises in that the value of a cell cannot be accurately converted only by using the simple binarization method.

SUMMARY OF THE INVENTION An object of the present invention is to provide a two-dimensional code reading method capable of accurately reading a two-dimensional code without converting a cell value accurately and degrading recognition accuracy even when the optical resolution is low or a misalignment occurs. There is.

In addition, the problem of the present invention is that by increasing the recognition processing speed per unit time, the number of images that can be displayed is improved, the function of the user interface is improved, and the image data suitable for recognition is photographed even in a dark place. In addition, the present invention provides a recognition device with improved recognition performance.

These problems include a coordinate value detection step of detecting coordinate values of cells on a two-dimensional matrix in recognition of a two-dimensional code in which data is digitized and placed in two dimensions, and a gray level value detection step of obtaining a gray level value on the detected coordinates; A histogram creation process of obtaining a frequency of color components of an image to create a histogram, a threshold detection process of obtaining a threshold from a histogram, a region range determination process of obtaining a certain distance from a threshold, and setting a predetermined distance from the threshold to a designated area; Intra-area cell selection process for selecting cells whose gradation value is within the area range, Out-of-area cell selection process for selecting cells whose gradation value is outside the area range, and selected cells in the out-of-area cell selection process, A cell value determination process for determining the cell type (white or black) from the result of comparing the threshold value with If the cell selected in the area cell selection process detects the gradation value of the cell adjacent to the cell of interest, and the neighbor cell is of the same kind, the cell of interest is determined to be different from the neighboring cell. Achieved by the two-dimensional code recognition method, characterized in that the first cell value determination step and the adjacent cell are not all of the same type, and the second cell value determination step of determining the cell type by comparing the cells of interest with a threshold value is performed. do.

1 is a diagram showing a processing block of a two-dimensional code recognition method according to the present invention.

2 is a diagram for explaining a method for determining a threshold for cell discrimination.

FIG. 3 is a diagram illustrating a method of determining the type of a cell by comparison with four or eight adjacent cells (part 1). FIG.

FIG. 4 is a diagram for explaining a method of determining the type of a cell by comparison with four or eight adjacent cells (part 2). FIG.

FIG. 5 is a diagram for explaining four neighboring neighboring cells and eight neighboring neighboring cells. FIG.

Fig. 6 is a diagram for explaining a method for determining the type of cell according to the embodiment of the present invention.

7 is a flow of recognition processing of a QR code.

8 is a diagram showing a flow of a two-dimensional code recognition method according to an embodiment of the present invention.

9 is a diagram showing a detailed flow of step S203 of FIG. 8;

FIG. 10 is a diagram showing a detailed flow of step S304 of FIG. 9;

FIG. 11 is a diagram showing a detailed flow of step S307 of FIG. 9;

FIG. 12 is a diagram showing a detailed flow of step S308 of FIG. 9;

It is a figure explaining the pretreatment which concerns on embodiment of this invention.

14 is an explanatory diagram of another embodiment of the pretreatment according to the embodiment of the present invention;

15 is a flowchart showing a flow of preprocessing.

FIG. 16 is a view for explaining an exposure time control method at the time of image capturing according to the embodiment of the present invention; FIG.

FIG. 17 is a diagram for explaining a method for separating a two-dimensional code and a background color according to an embodiment of the present invention (No. 1). FIG.

FIG. 18 is a diagram for explaining a method for separating a two-dimensional code and a background color according to an embodiment of the present invention (No. 2). FIG.

1 is a diagram showing a processing block of the two-dimensional code recognition method according to the present invention.

In Fig. 1, an image is read at 101 and photoelectric conversion is performed at 102 as a photoelectric conversion section. In the A / D conversion section 103, an analog signal is converted into a digital signal. In the two-dimensional code recognition processing unit 104, the two-dimensional code recognition process is performed, and the data character output unit 105 outputs the data character obtained from the recognition result of the two-dimensional code.

In the present embodiment, in particular, a mobile terminal such as a mobile phone or a PDA is used as the target device. In this type of device, a built-in camera for taking pictures is prevalent, but nowadays, a technique for reading a two-dimensional bar code using this camera, receiving recognized text data in the device, and simplifying text input is used. It is being proposed.

In the two-dimensional code image read out from the optical device, one cell is composed of a plurality of pixels. Therefore, it is necessary to determine which kind (white or black) each cell belongs to. First, the coordinate value of the cell of a code image is calculated | required, and the gradation component of a cell is calculated | required. The coordinate value is the center of the cell. Next, it determines which kind belongs to the threshold value processing of the gray-scale component of a cell.

Next, a threshold for determining which kind the cell belongs to is determined.

2 is a diagram for explaining a method for determining a threshold for cell discrimination.

When the irradiation amount of light is small or when the range of the sensor sensitivity is low (when the range width of the gradation value is narrow), and the fixed threshold value (for example, when the total gradation is represented by 255 gradation values, (The value is set to 128), the cell value cannot be converted accurately. For example, when the amount of light irradiation is small, the range of gradation values is lowered to a lower level as a whole, and "blur" and "broken" occur. The two-dimensional code is often expressed in black and white binary, so that the histogram of the input image generates two mountains in the light region and the dark region (see Fig. 2). By setting the center position between the mountains as the threshold value, it becomes possible to perform an optimal conversion process without "cloudiness" or "brokenness".

3 and 4 are diagrams for explaining a method of determining the type of a cell by comparison with four or eight adjacent cells.

In an isolated cell surrounded by cells having different kinds of four or eight directions, the gray scale value is attracted by the influence of an adjacent cell when the optical resolution is low or when the focus shift occurs. The method alone cannot accurately convert the value of a cell. For example, a back cell surrounded by black cells may have a low gray level due to the influence of surrounding black cells, and may be misidentified as black cells depending on the threshold value. As a result, as shown in Fig. 3, a comparison is made between the gradation value of the cell of interest and the adjacent cell, and if the difference of the gradation value is more than a certain level, it is converted into a different value from the adjacent cell.

When adjacent cells have values in both black and white (not isolated), the influence of adjacent cells is small. Only when all neighboring cells have the same value, the conversion is regarded as an isolated point.

The influence of the adjacent cells near eight points is large in the horizontal and vertical directions, and small in the inclined direction. However, as shown in Fig. 4, when only the adjacent cells on the left side are white, and all other colors are black, the gradation value of the cell of interest decreases in the black direction. Accordingly, in the case where eight points of proximity are used, even if all of the adjacent cells are not the same value, for example, when seven of the eight neighboring cells are the same type, the same processing as that of the isolated point is performed.

That is, in the left side of Fig. 3, since the gradation values of four adjacent cells around the center cell are 31, 40, 31, and 28, respectively, they are judged to be black when the threshold value is 128. At this time, even if the gradation value of the center cell is equal to or less than the threshold value, when the adjacent cells are all black and the difference between the gradation values of the center cell and the adjacent cell is more than a certain level, the center cell is set to white. In the right side of FIG. 3, when the gradation values of four neighboring adjacent cells are 231, 240, 240, and 228, and the threshold value is 128, all are judged to be white. Here, when the gradation value of the center cell is larger than the threshold value at 143, if the surrounding cells are all white and the difference between the gradation values with the surrounding cells is a certain value or more, the center cell is determined to be black.

In the left side of Fig. 4, when all 8 neighboring cells are black, even if the gray level value of the center cell is below the threshold value, the adjacent cells are all black (less than the threshold value), and the difference between the gray level value between the center cell and the adjacent cell is determined. In the case of a constant value, the center cell is determined to be white. In the right side of Fig. 4, seven of the eight adjacent cells are black and one is white. Also in this case, the gradation values of the center cell and adjacent cells below the threshold are compared, and if the difference is more than a predetermined value, the center cell is made white. Conversely, even when the center cell is surrounded by white, the center cell is made black.

As described above, even when the surroundings are surrounded by different kinds of cells, it is possible to accurately perform the conversion process.

5 is a diagram for explaining four neighboring neighboring cells and eight neighboring neighboring cells.

As shown in FIG. 5, adjacent cells near four represent four cells adjacent to the horizontal and vertical direction of the center cell. In addition, the adjacent cell of 8 vicinity shows the adjacent cell which added four inclination directions to four vicinity.

As described above, the cell in the oblique direction near 8 is far from the center cell compared with the cell near 4, so that the influence on the gradation value of the center cell is small.

6 is a diagram illustrating a method of determining the type of cell according to the embodiment of the present invention.

As shown in Fig. 6, in the histogram of gradation values, a method of determining the cell type by setting a predetermined range centering on the threshold value and determining whether the gradation value of the cell falls within this predetermined range. Change.

As shown in FIG. 6, the cell falling within the region range determines the value of the cell compared with the value of the adjacent cell as described above. Cells outside a certain region are classified into white and black by comparison with a threshold. By setting a certain range of gray zones from the threshold, it is possible to distinguish between cells that are considered to be affected by adjacent cells and cells that are not affected.

In Fig. 6, although a certain range of area is set around the threshold, by this, when most of the isolated cells or adjacent cells have different values from the cells of interest, the gradation values of the cells of interest are in different directions. As it is dragged, it becomes a value around the threshold. By setting a certain range of regions from the threshold, it is possible to distinguish between cells affected by neighboring cells and cells not affected.

Cells outside the area range determine the cell value compared to the threshold. By discriminating between cells affected by neighboring cells and cells not affected, it is possible to obtain an accurate cell value without errors.

Summarize the above.

In order to obtain the value of the cell arranged as a pattern on the two-dimensional matrix, first, the coordinate value of the cell in the image memory space is detected. Next, the gradation value of the cell on the detected coordinate is calculated. Next, a histogram of the gradation components of the cell is created, and the threshold value is determined. An area of a certain width is provided around the threshold, and the cell is divided into cells affected by adjacent cells and cells not affected. The cell in the area then determines the value of the cell compared to the adjacent cell. Cells outside the region determine the value of the cell compared to the threshold.

By performing the above process, even when the optical resolution is low or when the deviation of the focus occurs, the cell value can be converted accurately and the two-dimensional code can be read without degrading the recognition accuracy.

Moreover, in embodiment of this invention, the following means are implemented.

When photographing and recognizing an image while moving a recognition device with a camera with respect to a code to be recognized, in general, the time required for the recognition process is longer as compared with the shooting time of one image, and as a result, The problem that the number of images which can be displayed per unit time is lowered and the function of the user interface is lowered is as follows.

Preprocessing means for evaluating the possibility of recognizing whether or not a code is included in the target image is provided, and the actual recognition process is performed only when it is possible to determine that the object is a code by the determination result of the preprocessing means. As a result, the number of recognition trials per unit time is increased, the recognition processing speed per unit time is improved in appearance, and the number of images that can be displayed is improved, thereby improving the function of the user interface.

In addition, conventionally, in the case of recognition which targets one-dimensional code and two-dimensional code, the user judges the target code and the user switches the recognition process itself. The preprocessing means determines the image characteristics of the one-dimensional and two-dimensional code regions, and performs the recognition processing of the one-dimensional and two-dimensional code based on the determination result, thereby automatically switching the recognition processing.

The two-dimensional code has a ratio of around 50% of the black and white dots, and has a characteristic that it is distributed in two dimensions. As a result, preprocessing for evaluating the small region X, Y of the image, which is a two-dimensional code region, can be performed from one or more features of pixel gradation value dispersion, pixel black pixel rate, and monochrome edge number (vertical and horizontal).

The one-dimensional code has a feature in which a plurality of vertical lines having a certain length are combined. This enables preprocessing to evaluate the small area (X, Y) of the image, which is a one-dimensional code area, from one or more features of pixel gray scale value dispersion, number of black and white edges (horizontal), and line correlation.

When photographing and recognizing the image while moving the recognition device with a camera with respect to the code to be recognized, the image capturing uses a still image capturing control method. In order to improve, the image | capture | photographing of the image which ensured the gradation was performed by carrying out multiple exposure (generally called a long shutter) of the image of several frames beyond the reading time of the image data of one frame of a sensor. On the other hand, by multiple exposure of a plurality of frames, a broken image is photographed with respect to movement, and as a result, the recognition performance of the code is deteriorated. This problem is as follows.

By limiting the shooting time of the camera to within the data reading time of the camera, it is possible to provide a recognition device with improved recognition performance by capturing image data suitable for recognition in a dark place.

The two-dimensional code was composed of white and black cells and back cells for N cells representing the boundary of the two-dimensional code. In order to extract the two-dimensional code region, the N-cell backcell following the black cell was detected, and the rectangular portion connecting the detected boundary was extracted as the two-dimensional code region.

On the other hand, since the two-dimensional code is a set of black and white cells which are meaningless to the user, the two-dimensional code has been used for printing such as surrounding the boundary area with colors from the viewpoint of design. If the peaks of both mountains are obtained by a simple fixed threshold or histogram, and the black and white cells of the cell are determined by the determined threshold, the portion outside the colored boundary area is also a black cell, and the area of the two-dimensional code is determined at the time of determining the boundary. Could not be extracted. This problem is as follows.

The histogram of the input image produces a dark region, a bright region, a middle region of the background, and three mountains. The threshold value is determined by the black cell level determination process of determining the level range region on the dark side of the mountain. The cell value is determined at the determined threshold value, and the two-dimensional code can be correctly extracted by detecting the code boundary using the continuous lengths of the black and non-black cells.

Hereinafter, it demonstrates more concretely.

Here, the QR code will be described as an example of the two-dimensional code. Similar processing can be performed with respect to other two-dimensional codes. For details of QR code recognition, refer to JIS standard "JIS X 0510: Two-dimensional code symbol-QR code basic specification explanation".

7 is a flow of recognition processing of a QR code.

In step S101, black / white cells are recognized. Most of the processing of the present invention relates to the details of this step S101. In step S102, the format information included in the QR code is decoded. The format information is a model number of a QR code and a kind of mask which are encoded and described at a position where a QR code is determined. This is decoded in step S102. In step S103, the model number is determined from the read format information, and in step S104, the mask process is released. In the case of generating a QR code, which is a two-dimensional code, by simply encoding the information, when the two-dimensional code is used, the distribution of the black cells and the back cells arranged on the two-dimensional surface is biased toward the white side or the black side. At this time, one of several predetermined masks is superimposed in a two-dimensional code so that the distribution of a black cell and a back cell becomes uniform.

In step S105, data and error correction codewords are restored from the two-dimensional code obtained by canceling the mask process. In step S106, it is determined whether or not there is an error in the restored data by using an error correction codeword. If it is determined in step S106 that there is an error, error correction is made in step S107. If it is determined that there is no error, the process proceeds to S108. In step S108, the data codeword is decoded. In step S109, the decoded data codeword is output, and the processing ends.

8 is a diagram illustrating a flow of a two-dimensional code recognition method according to an embodiment of the present invention.

In step S201, in order to obtain the value of the cell arranged as a pattern on the two-dimensional matrix, from the image input from the optical device, the coordinate value of the cell in the image memory space is first detected. Regarding the calculation method of the cell coordinate value, the method described in "JIS X 0510: Two-dimensional code symbol-QR code basic specification commentary" is used.

In step S202, the gradation value of the cell on the detected coordinate is obtained.

In step S203, the value of the cell is determined.

9 is a diagram illustrating a detailed flow of step S203 of FIG. 8.

In step S301, the frequency of the gradation components of the cell is obtained, and a histogram is created. A histogram may be created from the gradation components of the pixels included in the code area.

In step S302, a threshold is determined. It scans from the left (arm) of the histogram, detects the gradation component whose frequency becomes more than a certain value, and finds the peak position of the region where the gradation component is low. Similarly, scanning is performed from the right side (light), the gradation component whose frequency is above a certain value is detected, and the peak position of the region where the gradation component is high is obtained. Then, the middle of the two peak positions is found, and it is set as the threshold (see Fig. 2).

In step S303, an area range is set.

For example, in the histogram of FIG. 2, attention is paid to the x-axis, m is divided between the peak and the threshold generated in the region having a low gradation value, and the range of n / m away from the threshold is determined. Similarly, even in a region having a high gradation value, the range of n / m away from the threshold is determined. In addition, n is taken as the value which does not exceed m.

In step S304, the cells are classified into cells within the area range and cells outside the area range.

Next, in steps S305 to S309, conversion processing of the cell stored in the two-dimensional code is performed. The parameter is set below.

a) data [i]: gradation value of the cell

b) i: cell number

c) n: total number of cells

d) flag [i]: discriminate whether it is in or out of region 1: out of region 0: in region

The parameter of d) is obtained in step S304.

In step S305, it is determined whether or not conversion processing has been performed for all the cells.

In step S306, two processes are classified according to the value of flag [i]. If flag [i] is 1, the flow advances to cell determination method 1 in step S307. If flag [i] is 0, the flow advances to cell determination method 2 in step S308.

The cell that proceeds to step S307 determines the value of the cell in comparison to the threshold, in a cell outside the region range on the histogram. The cell that proceeds to step S308 determines the value of the cell in comparison with the adjacent cell in the cell within the region range on the histogram.

In step S309, one cell number is advanced, and the flow proceeds to step S305. If the conversion process is performed for all cells, the process ends.

FIG. 10 is a diagram illustrating a detailed flow of step S304 of FIG. 9.

In the flow of FIG. 10, it is determined whether individual cells are within an area range or outside an area range. All cells are discriminated and the value of flag [i] is set. The parameter is set below. The other parameters are the same as in a) to d).

e) threshl: Threshold for the region with low region range gradation

f) thresh2: Threshold for the region with a high range of gradations

In step S401, the cell number is initialized.

In step S402, it is determined whether flag [i] is obtained for all cells.

In step S403, the gradation values of the cells are compared to determine whether they are within or outside the area range.

In step S404, 0 is substituted into flag [i] if it is within an area range.

In step S405, 1 is substituted into flag [i] if it is outside the region range.

In step S406, one cell number is advanced, and the flow proceeds to the next.

When the value of flag [i] is found for all cells, the processing ends.

FIG. 11 is a diagram illustrating a detailed flow of step S307 of FIG. 9.

The cell which performs the process of FIG. 11 is a cell which is out of an area range, and determines the value of a cell compared with a threshold value.

In step S501, the gradation value and the threshold value are compared.

In step S502, if the gradation value is less than the threshold value, the cell value is black (0).

In step S503, if the gradation value is larger than the threshold value, the cell value is set to the bag 255.

The cell value is 0 for black and 255 for white, but other values may be used.

FIG. 12 is a diagram illustrating a detailed flow of step S308 of FIG. 9.

The cell which performs this process is a cell within an area range, and the value of a cell is determined compared with an adjacent cell.

In addition, each parameter is as follows.

j: adjacent cell number

k: number of adjacent cells

K = 4 around 4

Near 8 k = 8

count_b: the number of black of adjacent cells

count_w: the number of bags of adjacent cells

ndata [i]: gradation value of adjacent cell

rev_thresh: If there are more than this number of adjacent cells, the type of the center cell is determined by a value different from the adjacent cells.

In step S601, parameter initialization is performed.

In step S602, it is determined whether the values of all adjacent cells are obtained.

In step S603, it is determined whether the gray scale value ndata [J] of the adjacent cell exceeds the threshold.

If it is below the threshold, one black count is advanced in S604.

If it is more than the threshold, the count of the bag is advanced by one in S605.

In step S606, one reference position of the neighboring cell is advanced, and when all neighboring cells are determined, the process proceeds to the next.

In step S607, it is determined whether neighboring cells are equal to or greater than the threshold number of blacks.

In step S608, if the threshold number is more than black, the cell of interest is determined to be the value of white, and the processing ends.

In step S609, if the neighboring cell is equal to or greater than the threshold number, the cell of interest is determined to be a black value, and the processing is terminated. In step S611, if it does not belong to either, the cell of interest is compared with the threshold to determine the value of the cell. If it is below the threshold, it is black. If it is above the threshold, it is white.

By performing the above process, even when the optical resolution is low or when the deviation of the focus occurs, the cell value can be converted accurately and the two-dimensional code can be read without degrading the recognition accuracy.

It is a figure explaining the pretreatment which concerns on embodiment of this invention.

When photographing and recognizing an image while moving a recognition device with a camera with respect to a two-dimensional code to be recognized, in general, the time required for the recognition process is long compared with the photographing time of one image. As a result, the number of images that can be displayed per unit time is reduced, so that the function of the user interface is reduced.

By providing preprocessing means for evaluating whether or not the target image includes the two-dimensional code, and performing the actual recognition process only when the object can be determined to be the two-dimensional code as a result of the determination of the preprocessing means, It is possible to increase the number of times of recognition, to improve the recognition processing speed per unit time, to improve the number of images that can be displayed, and to improve the function of the user interface.

In Fig. 13, when code recognition is started, the recognition image is acquired in step S700. In step S701, a preview image is created and the preview is presented to the user. The acquired image is preprocessed in step S702, and it is determined in step S703 whether a code is included in the image. If it is determined that no code is included, the recognition image is acquired again. If it is determined in step S703 that a code is included, a two-dimensional code recognition process is performed in step S704. In step S705, it is determined whether or not the decoding was successful. If it fails, the recognition image is acquired again. If the decoding succeeds, in step S706, the decoding result is output, and code recognition is terminated.

The following processing is performed as a preprocess on the picked-up image, and only when these conditions are satisfied, it is determined that the two-dimensional code exists in the target image, and the recognition processing is performed.

Preprocessing: For a small area (X × Y pixel) of the captured image,

Pixel variance (contrast)> α

Β1 <black pixel rate of pixels <β2

Number of monochrome edges (vertical and horizontal)> γ

If the condition is satisfied, the two-dimensional code area is determined. In addition, (alpha), (beta) 1, (beta) 2, (gamma) are previously experimentally calculated | required.

14 is an explanatory diagram of another embodiment of the pretreatment according to the embodiment of the present invention.

In FIG. 14, the same steps as in FIG. 13 are denoted by the same reference numerals, and description thereof is omitted.

In the related art, in the case where the one-dimensional code and the two-dimensional code are targeted, the user determines the target code, and the user switches the recognition process. Preprocessing means for evaluating whether the one-dimensional and two-dimensional codes are included in the target image is provided, and the actual recognition process is performed only when the object can be determined as the one-dimensional or two-dimensional code as a result of the determination of the preprocessing means. This increases the number of times the recognition is performed per unit time, improves the recognition processing speed per unit time, improves the number of images that can be displayed, and improves the function of the user interface. Automatic recognition processing can be realized for two-dimensional codes.

The following processing is performed as a preprocess on the picked-up image, and only when these conditions are satisfied, it is determined that a code exists in the target image, and a recognition process is performed.

Pretreatment:

For a small area (X × Y pixel) of the captured image,

Two-dimensional code

Pixel variance (contrast)> α

Β1 <black pixel rate of pixels <β2

Number of monochrome edges (both vertical and horizontal)> γ

One-dimensional code

Pixel variance (contrast)> α

Monochrome edge number (horizontal)> θ

Line correlation = correlation peak position is the same between lines

In one case, two-dimensional or one-dimensional code region

As in the case of FIG. 13, α, β1, β2, γ, θ, and the like are experimentally set to optimal values.

15 is a flowchart showing the flow of preprocessing.

First, in step S800, it is judged whether or not the gradation value dispersion of pixel values is out of a range. If it is out of a range, a codeless result is output in step S807 and the preprocessing ends. In step S801, the number of horizontal edges is calculated, and when out of range, the number of vertical edges is detected in step S807, and in range, in step S802. In step S802, if the number of vertical edges is out of range, the flow advances to step S804. When the number of vertical edges is within the range, the black pixel rate is detected in step S803. When the number of vertical edges is out of the range, the flow advances to step S807, and when within the range, the flow proceeds to step S804. In step S804, inter-line correlation is taken, and in step S805, it is determined whether there is a code. If it is determined in step S805 that there is no code, the flow proceeds to step S807. If it is determined in step S805 that there is a code, then in step S806, a determination result of whether it is a one-dimensional code or a two-dimensional code is output, and the process ends.

FIG. 16 is a view for explaining an exposure time control method at the time of image capturing according to the embodiment of the present invention. FIG.

Conventionally, in the case of photographing an image and performing recognition while moving a recognition device with a camera with respect to a two-dimensional code to be recognized, the photographing of the image uses a still image shooting control method, and in particular, In the place, in order to similarly improve the sensor sensitivity, an image having a gradation secured by taking multiple exposures (long shutter) of images of a plurality of frames beyond the reading time of one image data of the sensor. On the other hand, by multiple exposure of a plurality of frames, an image blurred by motion is captured, and as a result, the recognition performance is reduced (see Fig. 16: conventional exposure time control).

In the embodiment of the present invention, by limiting the exposure time of the camera to the driving time of one frame of the camera (see FIG. 16: exposure time control of the present invention), there is no blur that can follow the movement suitable for recognition even in a dark place. By photographing image data, a recognition device having improved recognition performance is provided.

17 and 18 are diagrams illustrating a method of separating the two-dimensional code and the background color according to the embodiment of the present invention.

The two-dimensional code was composed of white and black cells and back cells for N cells representing the boundary of the two-dimensional code. In order to extract the two-dimensional code region, the N-cell backcell following the black cell was detected, and the rectangular portion connecting the detected boundary was extracted as the two-dimensional code region.

On the other hand, since the two-dimensional code is a set of black and white cells which are meaningless to the user, the two-dimensional code has been used for printing such as surrounding the boundary area with colors from the viewpoint of design. When the peak of the bimodal peak of the gradation value is determined by a simple fixed threshold or histogram, and the black and white cell of the cell is determined by the determined threshold, the part outside the colored boundary area also becomes a black cell. In some cases, the area of the dimension code could not be extracted.

As shown in Fig. 17, the histogram of the input image generates a dark region, a bright region, a middle region of the background, and three mountains. Thereby, the threshold value is determined by the above-described determination method of the black cell level (the threshold value is set to the black cell level in FIG. 2) which determines the level range region on the dark side of the mountain. The cell value is determined by the determined threshold value, and the two-dimensional code can be extracted correctly by detecting the code boundary using the continuous lengths of the black and non-black cells.

In Fig. 18, the relationship between the gradation value of the boundary portion of the two-dimensional code and the recognized cell is shown. In the conventional example, even if it is not determined as the boundary, the boundary can be correctly judged by the present invention. By connecting the detected boundaries and cutting out the rectangular region, it is possible to correctly extract the two-dimensional region.

According to the present invention, even when the optical resolution is low or when a misalignment of the focus occurs, the cell value can be converted accurately, and the two-dimensional code can be read without degrading the recognition accuracy.

By using the recognition method of the present invention, it is possible to provide a two-dimensional code reading device with high recognition accuracy.

In addition, by improving the recognition processing speed per unit time by the determination of preprocessing, the number of images that can be displayed is improved, and the function of the user interface is improved. In addition, it is possible to provide a recognition device with improved recognition performance by photographing image data suitable for recognition even in a dark place.

Claims (16)

delete delete delete A gradation value comparison process for comparing gradation values of a cell of interest and an adjacent cell, A neighbor cell threshold comparison process for comparing a neighbor cell with a threshold; An adjacent cell determination step of determining a type of an adjacent cell by a gray level comparison process or a neighbor cell threshold comparison process; Cell value determination process of determining the cell of interest as a different type from the adjacent cell if all adjacent cells are the same kind or a certain number or more of the same kind Two-dimensional code recognition method comprising a. The method of claim 4, wherein The adjacent cell, A two-dimensional code recognition method characterized by selecting four cells in the x direction and the y direction. The method of claim 4, wherein The adjacent cell, A two-dimensional code recognition method characterized by selecting eight cells in the x direction and the y direction. The method of claim 4, wherein Self-determination process, A method for recognizing a two-dimensional code, wherein a gradation value is applied only to a cell within a range for a partial region to be recognized. delete The method of claim 7, wherein Self-determination process, Selecting a cell whose gradation value is out of a range of the partial region to be recognized; And determining a cell value by comparing the selected cell with a threshold value. In the two-dimensional code recognition method in which the data is cellized and arranged in two dimensions, A coordinate value detection process for detecting the coordinate values of the cell, A gradation value detection process of obtaining a gradation value on the detected coordinates, A histogram making process of obtaining a frequency of grayscale values of an image to create a histogram, A threshold detection process of obtaining a threshold value from a histogram, An area range determination process of obtaining a certain distance from the threshold value and setting the distance from the threshold value to a predetermined distance as a designated area; An intra-cell selection process for selecting cells whose gradation values are within the range of the region, An out-of-area cell selection process for selecting cells whose gradation values are outside of the area range, A cell value determination step of determining a cell type from a result of comparing the gray level value with a threshold value selected in the cell selection process outside the area; A neighboring cell gradation value detecting step of detecting a gradation value of a cell adjacent to the cell of interest, the cell selected in the cell selection process in the region; A first cell value determination process of determining a cell of interest as a different type from an adjacent cell if all adjacent cells are the same kind; A second cell value determination process of determining a cell type by comparing a cell of interest with a threshold value if all adjacent cells are not the same type; Two-dimensional code recognition method comprising a. In a recognition device for recognizing a code included in an image acquired by a camera, Provide preprocessing means for determining the characteristics of the acquired image and for determining whether the image contains one-dimensional or two-dimensional code, And a process of updating the input image when none of the one-dimensional code and the two-dimensional code is included as a result of the determination of the preprocessing means. delete The method of claim 11, The pretreatment means, A code recognizing apparatus, characterized in that it is determined whether or not a small area of an image is a two-dimensional code based on one or more determinations of pixel gradation value dispersion, pixel black pixel ratio, and monochrome edge number. The method of claim 11, The pretreatment means, A code recognizing apparatus, characterized in that it is determined whether or not it is a one-dimensional code for a small area of an image based on one or more determinations of pixel gradation value dispersion, black-and-white edge number, and line correlation. The method of claim 11, And the exposure time of the camera does not exceed a period of one frame of the camera. delete

Images