JP3838079B2 - Image processing apparatus and image processing program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention belongs to the technical field of recognizing patterns such as characters and symbols by image processing. In particular, when recognizing an arrangement of a plurality of dot-printed patterns, each pattern is processed from an image to be processed. Related to the technology to cut out.
[0002]
[Prior art]
In manufacturing plants for beverages such as bottled beverages and beverages in cans, the expiration date, production number, and the like are printed on the surface of the bottles and cans of products before shipment by a dedicated inkjet printer. Further, the products after the printing process are sequentially conveyed to an inspection line including a CCD camera and an image processing apparatus. If the printing state is determined to be defective, the products are excluded from shipment.
[0003]
In a general character string recognition process, an image area including individual characters is projected by projecting an image of a character string to be recognized in each of the horizontal and vertical directions and appearing for each projection axis. After the image is cut out, each character is recognized by using these image areas as recognition target areas and collating the image in the area with a reference image registered in advance. However, when a dot character string is to be recognized as in the above-described inspection, a single character pattern is projected in a discrete state. The projection process and the clipping process are performed on the expanded image (Japanese Patent Laid-Open No. 8-138049).
[0004]
[Problems to be solved by the invention]
Since a large number of products are conveyed one after another in the above-described inspection line, it is desirable that the inspection is completely automated. However, if the number of expansion processing for images is fixed for automation, the same number of expansion processing is used when products with different dot spacing or dot size are mixed due to misalignment during printing or changes in printing conditions. It becomes impossible to correspond. If the character cut-out process fails due to such a reason, a defect determination is made even for a character string that does not need to be determined as defective, resulting in a problem that the accuracy of inspection is greatly reduced.
[0005]
Further, in order to set an appropriate number of expansions, a laborious setting operation such as repeating a test expansion process many times using a model image is necessary. Therefore, if the character size or dot spacing changes due to changes in the product type or printer model, the setting of the number of expansions must be changed each time, resulting in a problem that the setting work is troublesome. .
[0006]
The present invention has been made paying attention to the above-mentioned problems, and an object thereof is to perform highly accurate recognition processing efficiently by automatically performing optimal expansion processing according to the dot pattern to be recognized. .
[0007]
[Means for Solving the Problems]
  The present invention cuts out each dot pattern from an image obtained by imaging a recognition target in which a plurality of dot patterns are arranged.Image processing apparatus having a recognition function, and program for causing a computer to function as the image processing apparatusIt was made about. Here, the “dot pattern” is a pattern of characters, symbols, etc. represented by a plurality of dots, and is mainly printed on a predetermined object by an ink jet printer or the like.PatternPoint to. However, the recognition target of the present invention is not limited to this. For example, a dot pattern displayed on the screen can be set as the recognition target.
[0008]
  An image processing apparatus according to the present invention includes an image input unit that inputs an image obtained by imaging the recognition target, a recognition processing unit that cuts out and recognizes each dot pattern from the input image, and the recognition processing unit. Output means for outputting a recognition processing result.
  The recognition processing means in the first image processing apparatus according to the present invention is a storage means in which the number of dot patterns constituting the recognition target array is registered; a step of expanding the image with respect to the input image; and after the expansion processing The first process of repeating the step of projecting the image toward the dot pattern arrangement direction or the projection obtained by projecting the input image toward the dot pattern arrangement method, Projected image acquisition means for executing a second process for repeating the step of expanding the image; the number of projected images is smaller than the number of dot patterns registered in the storage means during the execution of the first or second process A projection image dividing unit that divides a projection image whose magnitude in the projection axis direction exceeds a predetermined value when the projection image acquiring unit or the projection image dividing unit includes: When the number of the projected image corresponding to the number of dot patterns that are registered in the storage means by the management is obtained, configured to cut the dot pattern on the image based on these projections.
[0009]
  When the projection image acquisition means executes the first process, for example, after binarizing the input image,By the projection process, the distribution state of dots along the arrangement direction of the dot pattern is represented as a two-dimensional histogram.In this case, the histogramIndividually independent mountain portions can be extracted as projected images.AlsoMulti-valued imageIf it is a processing target,Histogram with density added along the text heightcreate. In this case, the histogramA portion having a predetermined threshold value or more can be extracted as a projection image.
[0010]
  "Number corresponding to the number of dot patterns registered in the storage means"IsNumber of dot patterns registeredIs preferably the same number.
  When executing the first process, since each dot is discrete in the first stage, the number of projection images obtained by the projection process is larger than the number of registered dot patterns, but the image expansion process is repeated. Thus, each dot in one dot pattern is connected, and the number of projected images approaches the registered number.
[0013]
  In addition, when the projection image acquisition unit executes the second process, after the projection target image is projected to obtain the projection image, the process of expanding the projection image is repeated, so that it is initially discrete. The projected image is connected for each dot pattern.
  Therefore,The number of dot patterns that make up the recognition target arrayIf registered in the storage means in advance, projection images corresponding to the number of registered dot patterns are obtained by the first or second processing even for recognition objects in which the interval between dots or the size of the dots has changed. Therefore, each dot pattern can be accurately cut out based on these projection images.
[0014]
  In the first image processing apparatus, for example,The spacing between dot patterns is smaller than usual due to printing misalignment.Because it becameA projection image corresponding to two dot patterns is obtained by the expansion process.Connected, the number of projected images constitutes the recognition target arrayDot patternEven if it becomes smaller than the number, it is connected by the processing of the projection image dividing meansCut the projected imageCan be released. Therefore, even when the number of projected images becomes smaller than the registered number of dot patterns during the execution of the first or second process, the number of registered projection images can be obtained by the process of the projected image dividing unit.Each dot pattern can be cut out individually with high accuracy, and a reduction in recognition accuracy can be prevented.
[0015]
  Next, the recognition processing means in the second image processing apparatus according to the present invention comprises the same storage means and projection image acquisition means as in the first image processing apparatus, and performs the first or second processing. When the number of projection images at the time of execution a predetermined number of times or more is larger than the number of dot patterns registered in the storage means, including projection image integration means for executing projection image integration processing based on the distance between the projection images, When the number of projection images corresponding to the number of dot patterns registered in the storage unit is obtained by the processing of the projection image acquisition unit or the projection image integration unit, the dot pattern on the image is cut out based on these projection images. ing.
[0016]
  According to the second image processing apparatus, one dot patternLarge deviation occurs in some dotsEtc., for the first or second processingEven if the expansion process is repeated, the number of projected images exceeds the number of original dot patterns.in case of,It is possible to converge the number of projected images to the same number as the number of registered dot patterns by integration processing.it can. ThereforeThe number of expansion processes can be kept within a predetermined reference number, and a highly accurate cutout process can be performed without reducing the efficiency of the dot pattern cutout process.
[0017]
  Next, according to the present inventionThe recognition processing means in the third image processing apparatus also includes storage means similar to the first and second apparatuses. Further, the recognition processing means projects the step of expanding the projected image obtained by the projection processing after projecting the input image in the direction of arrangement of the dot patterns to the number of dot patterns registered in the storage means. The process is repeated until a corresponding number of projection images are obtained, and a dot pattern on the image is cut out based on the finally obtained projection image.
[0018]
  In the first to third image processing apparatuses,The image input means is for capturing an image to be recognized from an external imaging means or the like, and can include a binarization circuit in addition to an input interface and an A / D conversion circuit. However, the input image is not limited to an analog image, and a digital image can be captured. Further, the binarization circuit is not always necessary, and grayscale image data can be processed.
  The imaging means is not limited to the outside of the image processing apparatus, but can be integrated into the apparatus main body of the image processing apparatus.
[0019]
  An image processing program according to the present invention, when an image obtained by imaging a recognition target formed by arranging a plurality of dot patterns on a computer having a function of inputting an image is input from the image. In order to function as a recognition processing means for cutting out and recognizing a dot pattern, and an output means for outputting a recognition processing result by this recognition processing means, accepting an input of the number of dot patterns constituting the recognition target array, A program for causing the computer to function as registration processing means for registering the number of input dot patterns in the memory of the computer is included. The memory in which the number of dot patterns is registered by the registration processing unit corresponds to the storage unit described above.
  Further, this program configures the recognition processing unit as a unit having the same configuration as the recognition processing unit of the first or second image processing apparatus.
[0020]
Note that the recognition process executed after the cut-out process can be appropriately set according to the purpose of use of the apparatus. For example, when used for the purpose of determining whether a character string is printed correctly, such as the inspection line as described above, each cut dot pattern is checked against each registered model pattern, and each check pattern is checked. The result is integrated to determine the suitability of the character string. In addition, when performing recognition processing for each character string using different character strings as recognition targets, the extracted dot patterns are collated with model models of various characters in order to identify characters corresponding to the respective dot patterns. Processing will be performed.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 shows an application example of an image processing apparatus 1 according to the present invention.
The image processing apparatus 1 according to the present embodiment includes a dot character string printed on the product 5 in a factory (in some cases, a dot pattern of a symbol in addition to a dot pattern of a character. ”Or simply“ characters ”), and the peripheral device such as the CCD camera 2, the monitor 3, and the external device 4 is disposed in the vicinity of the conveying conveyor 6.
[0022]
The CCD camera 2 (hereinafter simply referred to as “camera 2”) is provided with a lens facing the conveyor 6, sequentially images the products 5 conveyed on the conveyor 6, and outputs them to the image processing apparatus 1. . The monitor 3 is used for displaying the image being processed and displaying the inspection result as appropriate. The external device 4 is a device that can be used for data input, such as a controller, a keyboard, and a console, and is used to input setting data such as a character string length of a dot character string to be inspected and a reference value for the quality determination. .
[0023]
FIG. 2 shows an electrical configuration of the image processing apparatus 1.
In the image processing apparatus 1, a memory such as a ROM 12, a RAM 13 and an image memory 14 is connected to a processing unit 11 composed of a microprocessor (MPU), an A / D converter 15, a D / A converter 16, and an input / output. The part 17 and the like are incorporated. The ROM 12 stores not only a program for dot character string recognition processing and pass / fail judgment processing, but also a dictionary file in which model data of various dot patterns are registered. On the other hand, the RAM 13 is used as a work area for the processing unit 11 such as storing setting data input from the external device 4.
[0024]
The A / D converter 15 takes in the grayscale image data from the camera 2 and digitally converts it. The image memory 14 stores the digital grayscale image converted by the A / D converter 15 and an area for storing a binary image obtained by binarizing the grayscale image for recognition processing. It has. The D / A converter 16 converts the grayscale image into an analog display signal and outputs it to the monitor 3. The input / output unit 17 is an interface used when the processing unit 11 inputs / outputs data to / from the external device 4 or the monitor 3.
[0025]
FIG. 3 shows an inspection procedure in the image processing apparatus 1. In the drawings showing the following description and processing procedure, each processing step is indicated as “ST”. In FIG. 3, when a grayscale image from the camera 2 is input to the image memory 14 in the first ST1, the processing unit 11 binarizes the grayscale image with a predetermined threshold value and performs recognition processing. A binary image (hereinafter, this image is referred to as a “processing target image”) is generated (ST2).
[0026]
In the next ST3, a process of cutting out each character constituting the dot character string to be inspected is performed in the processing target image. Although details will be described later in this clipping process, the image expansion process and the projection process are repeated, and finally a rectangular area corresponding to the position and size of each character is recognized for each character. Set as. When this character segmentation process is completed, the processing unit 11 performs a collation process with the model pattern stored in the dictionary file of the ROM 12 for each segmented recognition target area in ST4, and determines whether each character is acceptable. Determine.
[0027]
Thereafter, the processing unit 11 integrates the determination results for each character, performs pass / fail determination for the dot character string, and displays the determination result on a monitor in ST5 and outputs the determination result to a host device (not shown).
[0028]
Next, the principle of character cutout processing will be described.
FIG. 4 shows a specific example of expansion processing performed in character cutout processing. Each rectangle in the figure corresponds to one pixel, and among these, the black pixel is a pixel constituting a dot. In the illustrated example, for each black pixel, pixels in the vicinity of four surroundings (upper and lower, four pixels on the left and right) are black pixels, so that the number of dots constituting the character is increased.
[0029]
Therefore, by repeating this expansion process, each dot of the dot character is connected, so that each character can be cut out by the projection process in the same way as a normal character pattern that is not a dot pattern.
[0030]
FIG. 5 shows an example of cutting out each character of a dot character string by projection processing. The dot character string in the illustrated example has a configuration in which ten characters are arranged in the horizontal direction (x-axis direction). Therefore, when the processing target image is projected toward the x-axis that is the character arrangement direction, Ten projection images A1 to A10 cut out every time can be obtained. Although not shown here, a projection image corresponding to the height of each character can be obtained by individually projecting image portions corresponding to the projection images A1 to A10 in the y-axis direction. Furthermore, when these projection processing results are integrated, rectangular regions R1 to R10 corresponding to the width and height of the character can be set for each character. The rectangular areas R1 to R10 are used for the subsequent recognition process as the above-described recognition target areas.
[0031]
By the way, since the dot character string to be inspected in this embodiment is printed by a high-speed printing process while conveying each product, the dot interval or the dot character string depends on the speed of the conveyor for conveyance and the state of the printer. There is a high possibility that the dot size will vary. For this reason, the number of times of image expansion processing required for appropriately cutting out characters may vary depending on the inspection target.
[0032]
FIG. 6 shows an example in which the number of expansions for obtaining the optimum projection image varies depending on the state of the dot character string to be processed.
6 (1) and 6 (2), the lower row is a binary image of a dot character string to be inspected, and each dot character string includes three characters “A”, “B”, and “C” arranged in order. In both FIGS. 6A and 6B, the image at the left end is the original processing target image, and the image is changed to an image in which the expansion processing is overlapped as it proceeds in the arrow direction. Above these images, the distribution state of projected images obtained when the images are projected in the direction of character arrangement is shown.
[0033]
The dot size of the dot character string to be inspected is different between FIG. 6 (1) and FIG. 6 (2). In FIG. 6 (1) for the character string having a small dot, the expansion process is performed. By performing twice, three projected images corresponding to each character are obtained. On the other hand, in the example of FIG. 6 (2) for a character string having a large dot, a projection image corresponding to each character is obtained by performing the dilation process only once.
[0034]
When the dot size or interval changes in this way, it is necessary to change the number of times of image expansion processing.
On the other hand, the dot character string to be inspected in this embodiment is a character string having a fixed number of digits, such as the date of manufacture, product code, etc. As shown in the above, it is possible to obtain the same number of projection images as the number of digits of the character string to be inspected. Therefore, in the image processing apparatus 1 of this embodiment, the number of characters constituting the dot character string to be inspected is input from the external device 4 in advance and stored in the RAM 13, and the image is obtained until projection images corresponding to the number of registered characters are obtained. By repeating the expansion process and the projection process, the character cutout process is performed appropriately and efficiently.
[0035]
FIG. 7 shows the detailed procedure of the character cutout process in ST3.
First, in ST301, the processing target image is projected in the x-axis direction. Since the processing target image here is a binary image, the projection processing result corresponds to a histogram indicating the number of black pixels along the x-axis direction. In the next ST302, a portion in which a value equal to or greater than a predetermined threshold value continues in this histogram (hereinafter, this portion is referred to as a “mountain”. This mountain corresponds to the above-described projected image). And count the number of peaks.
[0036]
Next, in ST303, the number of extracted peaks is compared with the number of registered characters. If the initial image expansion process is not performed, the number of peaks is larger than the number of registered characters. Therefore, ST303 is “NO” and the process proceeds to ST304, and the first expansion process is performed on the processing target image. Then, the process returns to ST301.
[0037]
Similarly, the image expansion process and the projection process are repeated by executing the loop of ST301 to ST304 a predetermined number of times. During this time, since dots in each character are gradually connected, the number of peaks obtained by the projection processing is gradually reduced. If the number of peaks matches the number of registered characters at a predetermined time, ST303 becomes “YES”, and the process proceeds to ST305.
[0038]
In ST305, the position and width of each mountain obtained by the latest projection processing are specified as the character position and character width, respectively. Next, image portions corresponding to the widths of these peaks are extracted from the latest dilated image and projected in the y-axis direction to specify the height of each character. Further, a recognition target area corresponding to each character is set from the projection processing results in the x-axis direction and the y-axis direction.
[0039]
In the inspection performed thereafter, as described above, the image in each recognition target region is checked against the model pattern in the dictionary file. The image to be checked is the original processing target image or the latest image. Any of the expansion processing images may be used.
[0040]
Next, in the example of FIG. 7, the dots are connected by the image expansion process. However, the present invention is not limited to this, and the histogram obtained by the projection process may be expanded.
FIG. 8 shows a procedure for cutting out characters by histogram expansion processing. In FIG. 8, in order to distinguish from the processing of FIG. 7, each processing is shown by the steps after ST306. However, FIGS. 7 and 8 are independent programs as the processing related to ST3 of FIG. Is executed. (The same applies to the procedures shown in FIGS. 9 and 10 below.)
[0041]
In FIG. 8, first, in ST306, the processing target image is projected in the x-axis direction, and in the next ST307, peaks are extracted from the histogram obtained by the projection processing and the number is counted. If the number of peaks is greater than the number of registered characters, ST308 is “NO” and the process proceeds to ST309, where a process of expanding each peak of the histogram by a predetermined width is performed. Thereafter, the process returns to ST307, and the peaks after the expansion process are counted.
[0042]
As a result of repeating the loops of ST307 to 309, if the number of peaks matches the number of registered characters at a predetermined time, the process proceeds from ST308 to ST310. In ST310, after specifying the position and character width of each character by each mountain obtained by the latest dilation processing, the image portions corresponding to these mountains are individually extracted from the processing target image, respectively, in the y-axis direction. Project the image to identify the height of each character. Further, a recognition target area is set for each character by combining the projection results in the x-axis direction and the y-axis direction.
[0043]
According to the processing of FIG. 7 and FIG. 8 described above, even if the dot interval and size vary, the optimum number of expansion processes can be automatically performed for each image to be processed. Each character constituting the dot character string can be accurately cut out to perform highly accurate recognition processing. Even if the printing status changes due to changes in printing conditions or changes in the printing status of the printer (such as the amount of ink), if the character string to be inspected is the same, the inspection is continued without changing any settings. be able to.
Furthermore, when the dot character string to be inspected is changed, the number of characters in the character string is input and registration processing in the RAM 13 is performed, and settings necessary for cutting out the characters in the dot character string are performed. This can greatly reduce the time and effort required for setting.
[0044]
Next, in FIG. 9 and FIG. 10, the processing target image is expanded as in FIG. 7. However, when the projected images for two characters are connected due to the shift of dots, the projected images for one character are separated. If you do, you can respond.
[0045]
First, in the procedure of FIG. 9, similar to the procedure of FIG. 7, the processing target image is projected, and then the peaks on the histogram obtained by this projection processing are collated with the number of registered characters (ST311 to 315). If the number of peaks is larger than the number of registered characters, the loop of ST311 to ST314 is repeated to expand the processing target image a predetermined number of times. When the number of peaks becomes the same as the number of registered characters by this process, both ST313 and ST315 become “NO” and the process proceeds to ST317, and the recognition target area for each character is set in the same manner as in the procedure of FIG.
[0046]
On the other hand, if two characters are combined in the expansion process and the number of peaks is smaller than the number of registered characters, ST313 is “NO” and ST315 is “YES”, and the process proceeds to ST316. In ST316, the width of each peak obtained in the latest histogram is compared with a predetermined reference value, and a peak having a width exceeding the reference value is divided into two, and then the character cutout process in ST317 is executed.
[0047]
According to the above procedure, even when some character intervals become smaller than usual due to printing misalignment, and mountains corresponding to the characters sandwiching the spaces are connected, these mountains are separated by the process of ST316. Therefore, each character can be cut out with high accuracy.
[0048]
Next, in the procedure of FIG. 10, after performing the projection processing of the processing target image, and further extracting the peaks from the histogram obtained by the projection processing and counting the number, the number of the peaks adds 1 to the number of registered characters. The image expansion process and the projection process are repeated until the value becomes equal to or less than the predetermined value (ST318 to 321).
[0049]
When the number of peaks converges to the number obtained by adding 1 to the number of registered characters by the above processing, ST320 becomes “YES”, ST322 becomes “NO”, and the number of expansion processing executed so far is checked in the next ST323. Here, if the number of expansion processes is within the predetermined specified number, ST323 is “NO” and the process returns to ST321. Subsequently, the expansion process and the projection process are executed by the processes of ST318 and 319, and the number of peaks is re-executed. To check.
The specified number of times is registered in advance in the RAM 13 or the ROM 12.
[0050]
If the number of peaks reaches the number of registered characters before the number of expansion processes reaches the specified number in this way, both ST320 and ST322 become “YES” and the process proceeds to ST325, and a recognition target area for each character is set.
On the other hand, if the number of peaks does not reach the number of registered characters even after performing the specified number of expansions, ST322 becomes “NO” and ST323 becomes “YES”, and the process proceeds to ST324. In ST324, the interval between the peaks in the latest histogram is compared with a predetermined reference value, and the two peaks sandwiching the smallest interval within the reference value are integrated into one. As a result of this processing, the number of peaks becomes the same as the number of registered characters, so the process proceeds to ST325, and a recognition target area for each character is set.
[0051]
According to the procedure of FIG. 10 described above, when a part of dots constituting a predetermined character is shifted and the projection image for the character cannot be converged to one even after performing the specified number of expansion processes, Since the expansion process can be terminated and the number of peaks can be adjusted to the number of registered characters, the time required for the expansion process can be reduced. Therefore, inspection accuracy can be maintained without prolonging the inspection time, and processing with high inspection efficiency can be performed.
[0052]
In the example of FIG. 10 described above, the expansion process is continued regardless of the specified number of times until the number of peaks is one larger than the number of set characters. Regardless of the number, the expansion process may be completed within a specified number of times. If the number of peaks when the number of expansions reaches the specified number greatly exceeds the set number of characters, the character string to be inspected may be determined to be defective without performing the mountain integration process. In this way, it is possible to significantly reduce the inspection time applied to a character string that is clearly regarded as defective due to a large shift in dots.
[0053]
Further, in the procedure of FIGS. 9 and 10, the processing target image is expanded, but the same setting as in FIGS. 9 and 10 can be incorporated in the procedure of expanding the histogram as shown in FIG. Further, if the processes of FIGS. 9 and 10 are executed in series, it is possible to deal with both cases in which a combination of characters occurs and a single character is separated, and a more accurate and efficient recognition process. Can be performed.
[0054]
【The invention's effect】
As described above, according to the present invention, in the process of cutting out each dot pattern from an image obtained by imaging a recognition target in which a plurality of dot patterns are arranged, an image to be processed or a projection image obtained from the image is obtained. Since it automatically expands the appropriate number of times according to the dot size and dot interval in the recognition target array, the character segmentation process can be performed accurately and efficiently, and the accuracy and efficiency required for automatic inspection Can be greatly improved. In addition, even if the type of object or printing conditions changes, it is possible to shift to recognition processing simply by registering the number of dot patterns that make up the new recognition target array, greatly reducing the setting effort. can do.
[Brief description of the drawings]
FIG. 1 is a perspective view showing an installation example of an image processing apparatus according to the present invention.
FIG. 2 is a block diagram illustrating an electrical configuration of the image processing apparatus.
FIG. 3 is a flowchart showing an inspection procedure.
FIG. 4 is an explanatory diagram illustrating a specific example of image expansion processing;
FIG. 5 is an explanatory diagram showing the principle of cutting out characters by projection processing.
FIG. 6 is an explanatory diagram showing an example in which the number of expansions necessary for character cutout varies depending on the size of dots.
FIG. 7 is a flowchart showing a detailed procedure of character cut-out processing.
FIG. 8 is a flowchart showing another procedure for character cutout processing;
FIG. 9 is a flowchart showing another procedure for character cutout processing;
FIG. 10 is a flowchart showing another procedure for character cutout processing;
[Explanation of symbols]
1 Image processing device
2 Camera
5 products
11 Processing unit
12 ROM
13 RAM
14 Image memory
15 A / D converter
17 Input / output section
A1 to A10 projection image

Claims (5)

Image input means for inputting an image obtained by imaging a recognition target in which a plurality of dot patterns are arranged, recognition processing means for recognizing each dot pattern cut out from the input image, and recognition by the recognition processing means Output means for outputting the processing result,
The recognition processing means includes
Storage means for registering the number of dot patterns constituting the array to be recognized ;
A first process for expanding the image and a step of projecting the image after the expansion process toward the arrangement direction of the dot pattern with respect to the input image / or the arrangement of the dot pattern A projection image acquisition means for executing a second process of repeating the step of expanding the projection image obtained by the projection process after projecting toward the method;
When the number of projection images becomes smaller than the number of dot patterns registered in the storage means during the execution of the first or second processing, the projection image whose size in the projection axis direction exceeds a predetermined value is divided. Projection image dividing means;
When the number of projection images corresponding to the number registered in the storage means is obtained by the processing of the projection image acquisition means or the projection image dividing means, the dot pattern on the image based on these projection images An image processing apparatus characterized by cutting out the image. Image input means for inputting an image obtained by imaging a recognition target in which a plurality of dot patterns are arranged, recognition processing means for recognizing each dot pattern cut out from the input image, and recognition by the recognition processing means Output means for outputting the processing result,
The recognition processing means includes
Storage means for registering the number of dot patterns constituting the array to be recognized;
A first process for expanding the image and a step of projecting the image after the expansion process toward the arrangement direction of the dot pattern with respect to the input image / or the arrangement of the dot pattern A projection image acquisition means for executing a second process of repeating the step of expanding the projection image obtained by the projection process after projecting toward the method;
When the number of projection images at the time when the first or second processing is executed a predetermined number of times or more is larger than the number of dot patterns registered in the storage means, the projection image integration processing is performed based on the distance between the projection images. Projection image integration means to perform;
When the number of projection images corresponding to the number of dot patterns registered in the storage means is obtained by the processing of the projection image acquisition means or the projection image integration means, an image based on these projection images is obtained. An image processing apparatus that cuts out the upper dot pattern. Image input means for inputting an image obtained by imaging a recognition target in which a plurality of dot patterns are arranged, recognition processing means for recognizing each dot pattern cut out from the input image, and recognition by the recognition processing means Output means for outputting the processing result,
The recognition processing means includes
A storage unit that registers the number of dot patterns constituting the recognition target array; after projecting the input image in the dot pattern array direction, the projection image obtained by the projection process is expanded; Repeating the step until the number of projected images corresponding to the number of dot patterns registered in the storage means is obtained, and cutting out the dot patterns on the image based on the finally obtained projected images. Image processing device. Recognition processing means for recognizing a computer having a function of inputting an image by cutting out each dot pattern from the image when an image obtained by imaging a recognition target in which a plurality of dot patterns are arranged is input , And a program for functioning as output means for outputting a recognition processing result by the recognition processing means,
A program for receiving the input of the number of dot patterns constituting the recognition target array and registering the number of input dot patterns in the memory of the computer is included. ,
The recognition processing means includes
A first process for expanding the image and a step of projecting the image after the expansion process toward the arrangement direction of the dot pattern with respect to the input image / or the arrangement of the dot pattern A projection image acquisition means for executing a second process of repeating the step of expanding the projection image obtained by the projection process after projecting toward the method;
Projection that divides a projection image whose size in the projection axis direction exceeds a predetermined value when the number of projection images becomes smaller than the number of dot patterns registered in the memory during execution of the first or second processing. And when the number of projection images corresponding to the number of dot patterns registered in the memory is obtained by the processing of the projection image acquisition unit or the projection image division unit. An image processing program characterized by being configured to cut out a dot pattern on an image based on the above. A recognition processing means for recognizing by cutting out each dot pattern from the image when an image obtained by imaging a recognition object in which a plurality of dot patterns are arranged is input to a computer having a function of inputting an image; And a program for functioning as output means for outputting the recognition processing result by the recognition processing means,
A program for receiving the input of the number of dot patterns constituting the recognition target array and registering the number of input dot patterns in the memory of the computer is included. ,
The recognition processing means includes
A first process for expanding the image and an image after the expansion process are repeated on the input image / or obtained by projecting the input image toward the dot pattern arrangement method, and the projection process. Projected image acquisition means for executing a second process of repeating the step of expanding the projected image;
When the number of projection images at the time when the first or second processing is executed a predetermined number of times or more is larger than the number of dot patterns registered in the memory, the projection image integration processing is executed based on the distance between the projection images. When the number of projection images corresponding to the number of dot patterns registered in the memory is obtained by the processing of the projection image acquisition unit or the projection image integration unit. An image processing program characterized by being configured to cut out a dot pattern on an image based on a projected image.

Images