JPH0311483A - Method and device for reading data - Google Patents

Abstract

PURPOSE:To heighten the recovery accuracy of a pattern even when a stain exists by reading image data representing an image from a recording medium, making the data into segments of plural partial data, searching a feature picture element group which features the fraction of a main scan reference pattern hereafter, and finding the position of the fraction by using the group. CONSTITUTION:An encoding image 20 is scanned by moving a manually operated line image sensor 11 consisting of a CCD element, etc., in a lateral direction that is the subscanning direction of the recording medium such as paper, etc., and obtained image output is binarized, and after that, it is transmitted to a control circuit 12 with a timing control signal. Meanwhile, a CPU 13 is operated according to a program stored in a ROM 14, and parallel image data are written on an image RAM 15 at every parallel conversion of serial image data. After the scan of the image is completed, the decoding work of the image data recorded on the RAM 15 is started, and a result is stored in a RAM 16. At such a case, the decoding work is comprised of a processing to recognize a sampling reference pattern, and a processing to identify the light-and-shade of each stitch.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to a data reading method and apparatus for reading encoded data by capturing an encoded image recorded on a recording medium such as paper or a sheet.

[Background] Barcode technology has long been known as an image encoding technology, but due to its structure, barcodes have a limit to the amount of information that can be recorded, making it impossible to increase the recording density. .

Recently, instead of barcodes, the applicant has used a mesh pattern, in which data is encoded by selectively forming brightness and darkness in each mesh, as the data body of an encoded image, by arranging many meshes vertically and horizontally. proposed a technology for capturing and decoding encoded images (patent application No. 63-328028).
In this method, each data bit is expressed by the brightness and darkness of each mesh that is two-dimensionally arranged at minute intervals on the recording medium, so the recording density can be greatly improved.

The purpose of data decoding for this type of coded image is to know the brightness and darkness of each mesh in the mesh pattern, but the method of searching for the mesh directly from the image data captured from the image sensor is generally disadvantageous as an approach. Therefore, when using a line image scanner that manually scans the encoded image on a recording medium as a simple image sensor, erroneous recognition is likely to occur. Due to the changes, considerable image distortion occurs during the scanning stage, and the location of each mesh on the captured image data is significantly offset from the location of each mesh on the recording medium, making interpretation extremely difficult. This problem can be alleviated by providing the image sensor with a special mechanism such as a rotary encoder for measuring the scanning speed, but this increases the cost.

Therefore, in the above-mentioned Japanese Patent Application No. 63-328028, a coded image on a recording medium is provided to indicate the center position of each mesh in the main scanning direction (the center of the mesh in the vertical direction). A main scanning reference pattern and a sub-scanning reference pattern for indicating the center position in the sub-scanning direction (horizontal center of the mesh) are added, and in the decoding process, the image data captured with this encoded image is From this, the position of each scanning reference pattern is recognized, the center position of each mesh on the image data is determined from the result, and image bits (pixels) representing the brightness and darkness of the mesh located there are sampled. In particular, in the embodiment, two main scanning reference patterns are placed on the outside of the mesh pattern in which meshes are arranged at equal intervals vertically and horizontally.
Wooden guidelines are arranged in parallel to sandwich the mesh pattern, and a synchronization mark with a pattern of alternating black mesh and white mesh (clock) is provided along the inside of each guideline as a sub-scanning reference pattern. The vertical mesh rows of the mesh pattern are made to coincide with the lines connecting the corresponding clocks of these two synchronization marks. For such a coded image, a line image sensor is moved roughly along the guideline direction and the image data is captured line by line. During the decoding process, the line image sensor moves from both ends of each main scanning line image of the image data toward the center. The pixel values are checked, and by detecting the first change from white to black, the guideline position in the main scanning line image is obtained. Each main-scanning line image on the image data represents an image portion on a line that diagonally crosses the mesh pattern on the recording medium not completely in the vertical direction but with variations for each line image. However, even in that case, the point that appropriately divides the distance between the two guideline positions (guideline interval) on the diagonal line is the vertical center of each mesh in the mesh pattern (and the vertical center of the mesh of the synchronization mark). The encoded image is configured so that it is located at the center). Therefore, by utilizing this property, the guideline positions of each main scanning line image are equally divided according to the image format, and the image pictograms at each equally divided point are sampled to obtain main scanning decoded image data. The arrays (image dot rows) on both sides of this main-scanning decoded image data represent images centered around the vertical center of the synchronization mark.
As you scan this image dot array, you should be able to see alternating run lengths of white pixels and run lengths of black pixels representing each mesh (clock) of the synchronization mark.

The center of each run length is considered to be the horizontal center of the clock of the synchronization mark, and the line connecting the centers of the corresponding run lengths in the image dot rows of the synchronization marks on both sides is the horizontal center of the vertical mesh row of the mesh pattern. It is considered to pass through the center of the direction. Therefore, according to this principle, the center of each clock of the synchronization mark is detected from the main scanning decoded image data, and each image dot (pixel value) on the line connecting the centers of the corresponding clocks on both sides is calculated. , the brightness of each mesh in the mesh pattern is identified.

As described above, by adding an inexpensive scanning reference pattern to the encoded image and detecting each position of the scanning reference pattern from the captured image data, each mesh of the mesh pattern can be directly detected from the distorted image data. It is now possible to find the position of each mesh without the difficult problem of searching.

However, if the encoded image itself on the recording medium has stains or other scratches, and the scanning reference pattern is partially soaked or chipped or damaged, the above-mentioned decoding method may be used. In this method, there is a high risk of mistaking dirt, etc. as fragments of the scanning reference pattern, and the position of each mesh of the predicament pattern on the image data determined based on the result will be incorrect, resulting in haphazard brightness discrimination results. There was a problem 0 For example, the deviation in the position of the guideline as a main scanning reference that occurs due to dirt is caused by the formation of a mesh that equally divides the lines connecting the corresponding positions of the two guidelines on the main scanning line image. This also affects the position that should be the directional center position, which can result in sampling errors that cannot be corrected later.

[Object of the Invention] Therefore, the object of the present invention is to provide a scanning reference pattern for an encoded image whose data main body is a mesh pattern, especially a main scanning reference pattern for instructing a data sampling standard of the mesh pattern in the main scanning direction on a recording medium. It is an object of the present invention to provide a data reading method and device capable of correctly determining each position of a scanning reference pattern on image data even when the image data is partially destroyed by dirt or the like.

[Structure and operation of the invention] In order to achieve the above object, the present invention reads image data representing the image from a recording medium on which an image including a reticular pattern and a main scanning reference pattern is recorded, and reads the read image data. Segmented into a plurality of partial image data, each segmented partial image data is searched for a characteristic pixel group characterized by a fragment of the main scanning reference pattern from among the partial image data, and in the search, the characteristic pixel group is For undetected partial image data, the position of the main scanning reference pattern fragment in that partial image data is determined from the position of the main scanning reference pattern fragment indicated by the characteristic pixel group detected in the search for other partial image data. By determining this, positional information of each fragment of the main scanning reference pattern in the image data is generated, and based on the generated positional information, the brightness and darkness of each mesh of the mesh pattern in the image data is identified. It is said that

According to this configuration, even if the main scanning reference pattern on the recording medium is partially destroyed due to dirt or the like, the destroyed part is limited to a partial area called partial image data, and furthermore, this partial area Since the position of the main scanning reference pattern in is determined based on the position indicated by the characteristic pixel group obtained from other healthy main scanning reference pattern fragments, the position of the main scanning reference pattern in the case of no destruction can be determined with high precision. Therefore, the brightness and darkness of each mesh of the mesh pattern, which is performed using the position of the main scanning reference pattern as a reference, can be reliably identified.

In one preferred embodiment, the main scanning reference pattern is composed of two continuous lines (guidelines) extending on both sides of the net-like pattern, and the pixel groups characterizing the guideline fragments are defined in the depth direction of the image data (orthogonal to the main scanning line image). A run length (referred to as a guideline set) of three white, black, and white trees with appropriate spacing is used. Such a guideline set is searched for each of the two guidelines on the segmented partial image data, and the position of the guideline part for which the search fails (possibly both) is later replaced by a healthy guideline before or after it. Determined by interpolation from the position. Such a two-tree guideline can accommodate changes in the scanning direction that occur when manually scanning an image sensor, and the pattern itself is a long continuous line, which has characteristics that other image elements do not have. Therefore, it is advantageous for recognition, and furthermore, it can function as a boundary of the encoded image even when an image that is undesirable for decoding, such as a character, is placed around the encoded image.

This invention can be applied to main-scanning reference patterns having any suitable pattern, but (a) those having local characteristics (for example, outside the reticular pattern), and (b) the pattern itself being unique. It is desirable to have a pattern that can be easily distinguished from each mesh of the predicament pattern in order to facilitate recognition of the main scanning reference pattern.

Also, while scanning the encoded image by the line image sensor, a data start mark is formed at a position li (before the mesh pattern) scanned prior to the predicament pattern, and a data start mark is formed at the position scanned after the mesh pattern is scanned. End marks may also be formed. In this case, by detecting these data start and end marks on the image data during the decoding operation, reliable data decoding becomes possible.

For example, if the edge of the main scanning reference pattern is destroyed as in the guideline mentioned above, the data start mark and end mark are located on the main scanning line or line group that crosses the edge of the main scanning reference pattern. By determining the format of the encoded image in such a way, the data start mark or end mark fragment can be estimated based on the detected line of the main scanning line image and the undestroyed main scanning reference pattern part. The destroyed end of the main scanning reference pattern can be determined from the line.

In addition, the size (depth) of partial image data, which is a unit area in which to search for fragments of the main scanning reference pattern, or the size of the fragment to be searched for can be localized depending on the search result, for example, depending on the size of a destroyed fragment. It may be reduced step by step until they become approximately equal.0 For example, if the search for two consecutive partial image data fails consecutively, the size of the fragment to be searched is halved and the search is attempted again. It is. In this way, it becomes possible to determine the position of the main scanning reference pattern with high accuracy.

[Example] Hereinafter, the present invention will be described in detail with reference to the drawings.

Example 51 of "Chopsticks" FIG. 1 shows the overall configuration of a data reading device IO according to a first embodiment. The purpose of the entire apparatus is to read a net-like image 20 (see FIG. 2) recorded on a recording medium such as paper with an image sensor 11 and decipher the data encoded in one image.

The image sensor 11 is, for example, a manual line image sensor including a detection array of COD elements, and is manually moved in the sub-scanning direction (lateral direction) of the recording medium to detect the encoded image 20 on the recording medium. to generate corresponding image data. Specifically, in each detection element of the detection array, the input light, that is, the input light according to the brightness of the pixels on the recording medium facing each detection element, is photoelectrically converted into an analog electrical signal, and the time corresponding to the main scanning life is The analog image output of each detection element is sequentially and periodically taken out via an analog multiplexer etc. so that line image outputs for all detection elements of the detection array can be obtained at each predetermined period (main scanning period). . After that, it is binarized by a binarization circuit (for example, black pixels are
1'', and O'' for white pixels), this binarized serial image data is transmitted to the control circuit 12 together with a timing control signal. The control circuit 12 performs control related to the image sensor 11 and parallel conversion of image data sent in series.

On the other hand, the CPU 13 operates according to the program stored in the ROM 14, and while the image sensor 11 is scanning an image, each time the control circuit 12 converts the serial image data into parallel data, the parallel data is transferred to the image RAM.
After completing the 0 image scan, which is written to 15, the CPU 13
starts decoding the image data recorded in the image RAM 15 and records the result in the RAM 16.

According to this embodiment, the decoding work is roughly divided into a process of recognizing a sampling reference pattern of data from image data, and a process of identifying the brightness and darkness of each mesh of the predicament pattern, which is the main body of data, based on the result. For example, even if the sampling reference pattern is partially destroyed due to dirt, etc., each position of the sampling reference pattern on the image data can be determined with high precision, and this makes it possible to obtain accurate decoding results for the mesh pattern. I have to. The details will be described later.

FIG. 2 is an example of an encoded image on a recording medium to be read by the data reading device 10. The encoded image 20 shown in FIG. 2 has a mesh pattern 22 as a data body inside. The mesh pattern 22 consists of a large number of meshes arranged vertically and horizontally, and each mesh is selectively dark (black) or bright (white) in order to express (encode) bits that are units of data. For example, in the example shown in Figure 0 in which black mesh represents bit "1" and white mesh represents bit "O", mesh pattern 22 has 48x72 bits of information.

Further, the encoded image 20 has a pattern or mark for indicating data sampling standards for main scanning and sub-scanning with respect to the mesh pattern 22 which is the data body.

In the case of FIG. 2, the main scanning sampling reference mark and the sub-scanning sampling reference mark are separate, and the main scanning sampling reference mark extends in the sub-scanning direction along both the upper and lower sides of the mesh pattern 22. The sampling criteria for the sub-scanning is given by the two guidelines 21 that are present, and the sampling criteria for the sub-scanning is provided by two synchronization mark rows 25 having a pattern of alternating black and white meshes along the inside of each guideline 21.
Since the interval between the black and white meshes of the synchronization mark row 25 given by is synchronized in such a manner as to match the mesh interval in the sub-scanning direction of the mesh pattern 22, each mesh of the synchronization mark row 25 can be referred to as a clock. Make it.

The reason why the two guidelines 21 spaced apart as the main scanning reference pattern are provided along the moving direction of the image sensor 11, that is, the sub-scanning direction of the image 20, is because the image sensor 11 is directed during image scanning. This is to be able to respond even if the That is, the image sensor
The image data from 11 is stored in the image RAM 15 in a manner that allows each main scanning line to be identified, but it is not clear in which direction each stored main scanning line image data represents a line image. By detecting the positions of the two guidelines 21 on each main scanning line image data, the detection result corresponds to the vertical center position M (or the vertical existing range) of each mesh of the mesh pattern 22. The position on each main scanning line image data (center position R of main scanning data sampling) is calculated based on the positional relationship of the original image, that is, the positional relationship between the two guidelines 21 and the mesh pattern 22 in the encoded image on the original recording medium. For example, in the case of the image format shown in FIG. 2, the two guidelines 21 are parallel to each other and also parallel to the mesh pattern 22. Since a straight line connecting one point on the upper guideline 21 and one point on the lower guideline 21 passes through a plurality of equally spaced meshes in the mesh pattern 22, this also applies to line image data representing an image of such a straight line. Therefore, by finding the points of the two guide lines 21 on the line image data and dividing them equally, the center position of each mesh in the main scanning direction in the line image can be obtained. It will be done.

On the other hand, the synchronization mark row 25, which is a sub-scanning reference pattern, is used to determine the position of each mesh in the sub-scanning direction in the captured image data. That is, the clocks of the upper and lower synchronization mark rows 25 are detected from the image data, and attention is paid to the corresponding clocks of the upper and lower rows.

In the case of the encoded image format shown in FIG. 2, a straight line connecting the centers of the corresponding clocks passes through the center of each mesh in a certain vertical column of the mesh pattern 22 in the horizontal direction (sub-scanning direction). Therefore, by detecting the position of each clock of a pair of synchronization mark rows 25 from the image data, the center position or the existence range in the sub-scanning direction of each mesh of the mesh pattern 22 on the image data can be determined. You can decide.

Furthermore, in the case of the encoded image 20 in FIG.
There is a data start mark 23 consisting of a checkered mesh group in the area scanned prior to 2, and in the area scanned after the mesh pattern 22 is scanned, there is a horizontally long thin mark 23 in black and white alternately along the main scanning direction. A repeating data end mark 24 is formed.

As in this example, the data start mark 23. End mark 24
These marks can be easily detected on image data by forming periodic or regular patterns. Further, as in this example, by forming the data start mark 23 and the data end mark 24 in a pattern that is easily visually distinguishable, the user can easily determine the scanning direction (movement direction) of the image sensor 11 with respect to the encoded image 20. Easy to judge. Furthermore, Guideline 2 as a main scanning standard
Even if the end of data 1 is destroyed due to dirt, etc., by detecting the data start mark 23 or end mark 24,
It is possible to detect the start and end of data, and even guideline 2
It is also possible to predict or evaluate the collapsed position of 1.

Figure 3 shows the encoded image shown in Figure 2 on the image sensor 1.
This figure shows the image obtained when manually scanning the image in step 1. According to the embodiment, even if the sampling reference pattern is partially missing, that is, the guideline 21 and the synchronization mark row 25 are somewhat destroyed, as long as the degree of distortion shown in the figure is present, it is possible to sufficiently cope with the problem and obtain correct decoding results. can give.

Image data decoding processing according to an embodiment will be described below.

FIG. 4 is a flowchart of the process of storing image data obtained by scanning an encoded image 20, recognizing a guideline 21 which is a main scanning reference pattern from the stored image data, and performing main scanning sampling of the image data. FIG. 5 is a flowchart of a process for performing sub-scanning sampling of image data by recognizing a synchronization mark row 25, which is a sub-scanning reference pattern, from image data subjected to main-scanning sampling. This embodiment is characterized by the recognition of the guideline 21 and the synchronization mark row 25, and as described below, even if some damage occurs to the guideline 21 or the synchronization mark row 25, the accuracy is high. You can determine the location of the destruction.

In FIG. 4, 4-1 to 4-5 are steps for reading the encoded image 20 on the recording medium with the image sensor 11 and writing it into the image RAM 15. Note that the scan end condition (memory cup) shown in 4-3 is merely an example, and the occurrence of any other appropriate event can be used as a signal for the end of the scan.

Further, as shown in 4-4, a byte memory is assumed as the image RAM 15. Figure 6 is image RAM
15, in which one line of image (line image) is written in one line on the side of the diagram (consecutive addresses of example sentences in image RAM 15).

The image decoding process begins at 4-6 in FIG. 4-
6, a guideline set (a group of pixels characterizing a guideline, which is a main scanning reference) is searched from the entire image data stored in the format shown in FIG. As shown in FIG. 2, the guideline 21 is a continuous black line, and has features that other elements of the encoded image 20 do not have. Therefore, for example, as illustrated in FIG. 7, a guideline set can be defined by three parallel lengths 73, 74, and 75 of white, black, and white with appropriate intervals. The initial value of the run length spacing or radius 76 needed to find the guideline set can be a fixed standard limit, or it can be measured by measuring the width of the white or black dot that most frequently appears in the line image. An appropriate three-run length 73.7, which may be determined by
The search for the guideline set defined in 4.75 starts from the positions of self-dots, black dots, and white dots with appropriate intervals on an appropriate line image in the image RAM 15, and searches in the depth direction perpendicular to the line image (see Fig. 6). in the case of,
This can be done by repeating the process of tracking the image in the vertical direction, checking the number of consecutive white, black, and white dots (run length), and comparing the results with the conditions of the guideline set. If the search fails, an error occurs because the information cannot be deciphered (4-7); however, if the search succeeds (4-7), standard values for the width of the guideline on the image data and the inclination in the scanning direction are determined from the information in the guideline set. Furthermore, in 4-8, the margins that take into account the variation of the tilt in the scanning direction from the standard value are set for the left and right (top and bottom in the case of Figure 2, but hereinafter referred to as left and right according to Figure 6). By adding to position a, the left and right search areas (search @ 81 in FIG. 8) of the image data to be handled in subsequent processing can be found. Note that this restriction on the search width 81 is desirable when there are other images such as characters around the encoded image 20 (images that may become noise during decoding), as shown in FIG. This is not particularly necessary if the encoded image 20 has a blank space around it.

4-9 to 4-11 are processes for limiting the search area in the depth direction of image data, and for this purpose, the data start mark 23 and data end mark 24 are detected, and these marks 23 and 24 are detected. The position of the left and right guidelines in the line image is determined. The data start mark 23 and end mark 24 can be detected because the line image includes a periodic black and white pattern, for example,
In the case of the encoded image in Figure 2, start and end marks 23.2
4 is a repetition of 24 black and white pairs, so if about 20 black and white pairs with the same period are found, taking into account the margin, it is sufficient to identify these marks. The data start mark 23 is searched from the top line (first line) of the image data, and the data end mark 24 is searched from the top line (first line) of the image data.
It is convenient to search from the bottom line (last line) of the image data.If the start mark 23 or the end mark 24 cannot be detected, it is possible to detect an error such as only a part of the encoded image 20 being scanned. Since the cause is thought to be an operation, etc., it is treated as an error (4-10.4-12).

From 4-13 to 4-18, the search width 81 and search depth (start line 8 in FIG. 8) are determined by the above-mentioned processing.
2 to the end line 83) is segmented into a plurality of partial images partitioned in the depth direction called search blocks, and the presence or absence of guideline destruction in each segment is being investigated.

In the case of FIG. 8, the image data is divided into eight search blocks 84. The size (depth) of such a search block is determined by 4-13, and the first search plotter starting from the line next to the start line 82 is selected.

The depth of the search block can be determined, for example, by considering the information on the guideline set obtained in 4-6, processing time, accuracy, etc., and its value is determined by the white, black, and white values related to the guideline set searched in 4-14. Determine the required run length. In other words, if a run length in the depth direction of white, black, and white with an appropriate interval and a length (greater than or equal to the depth of the search block) is found, a guideline set is present, and the position (guideline set) is found. (a rectangular area surrounded by ) is stored. If the search fails, it means that there is some kind of deficiency in the guideline of that search block, so set a flag according to the failure status, such as whether the guideline has failed on the left or right, or both. Make it possible to interpolate the guideline position (4-15, 4-1, 6) 0 For example, in the case of Figure 8, there is dirt 85 on the right guideline part of the fourth and fifth search blocks from the top. Therefore, the right guideline is detected as an error in these search blocks.

After inspecting the state of the guideline for one search block, the next search block is selected (4-17) and inspected, and the inspection is repeated until the last search block ending at the end line 83 (4-18).

Although this is not done in the flow of FIG. 4, the size of the search block may be localized variably according to the search results.For example, if a search block with a guideline error is detected in 4-15, Halve the depth of that search block and try again with 2
Guideline searches in each of the three search blocks, or four search blocks with half depth.
The loop from -13 to 4-18 may be restarted.

From 4-19 to the end in Figure 4, for each search block, the positions of the left and right guide lines in each main scanning line image are determined, and based on the position information, the main scanning line image is The scanning sampling position n (whole trajectory is indicated by reference numeral 31 in FIG. 3) is determined, and the image data dot (I) at each sampling position is determined.
i elements, dots) are extracted from the line image to create a main scanning decoding array 100 (see FIG. 10).

The position of the guideline on each main scanning line image (center position of the guideline width) can be determined by actual measurement if no failure flag indicating error or destruction is set for the guideline in the search block to which the main scanning line image belongs. However, if the failure flag is set, interpolation is performed from the position of the guideline actually measured in the part where the guideline is normal.

For example, as shown in FIG. 4, the position of the guideline in the main scanning line image in question is determined by linear interpolation from the position of the guideline in the main scanning line image at the ends of the previous and next search blocks by referring to the list of failure flags. can be obtained.

According to the flow, the check 4-20 will be NO if both the left and right guidelines of the current search block are normal (this can be seen because the corresponding failure flag is lowered), and in that case, the check 4-29 will be NO. The center positions of the left and right guide lines are actually measured for each main scanning line image in the current search block, and the positions of equally divided points according to the format of the encoded image are obtained between the corner positions as main scanning sampling points, and each sampling Extract the image bit at the point, if at least one of the left and right guidelines has a failure flag in 4-20,
21 and below, in order to interpolate the position of the guideline with the failure flag set, the guideline position one line higher than the current search block, that is, the guideline position W in the last main scanning line image of the previous search block (current search block is the first search block, then 4-
Set the guideline position M) of the start line obtained in step 9 as the interpolation starting point 4-21), and search for a search block with a normal guideline from the next search block onwards.
-22.4-26), detect the position of the guideline in the uppermost main scanning line image of the normal search block as an interpolation candidate 4-27), and detect each position of the guideline in the problem search block between the start and end of interpolation. The position of the guideline on the main scanning line image is determined by interpolation, and main scanning sampling is performed (4-28).
Although it is not specified in the flow, if one guideline is normal, its position is directly measured. Additionally, if the guideline near the end line is destroyed, the guide position of the end line evaluated in 4-11 is taken as the interpolation end (4-23.4-24)
.

In this way, the guideline sections with errors are interpolated based on the positions actually measured in other normal guideline sections, and highly accurate main scanning sampling is performed.As a result, main scanning decoding as shown in Figure 1O Array 100 is completed. Here, the columns on both sides of the main scanning decoded image data 100 are one-dimensional arrays of image dots (dot arrays) along the main scanning center locus 31 (FIG. 3) of the synchronization mark row 25, which is the sub-scanning reference pattern. Examine the dot row of this left and right synchronization mark row 25,
The center position in the sub-scanning direction for each clock of the synchronization mark row 25 is determined, and based on the result, sub-scanning data sampling of the mesh pattern 22 is performed to identify the brightness and darkness of each mesh, as shown in FIG. It's a flow.

In Fig. 5, from the main scanning width, that is, the interval between the left and right guide lines 21 (obtained in Fig. 4) in 5-1, the start clock check 5-3 determines the standard value (comparison reference value) of the clock. The length in the depth direction (sub-scanning direction, vertical direction in FIG. 10) is determined. However, in this embodiment, it is assumed that the image sensor 11 is manually scanned with respect to the encoded image 20, and therefore a considerable amount of variation in the clock length is expected, so a fairly large margin is added to the standard value. It is necessary to attach it. Note that instead of calculating the standard value from the main scanning width, for example, the main scanning decoding array 100
Check the dot row in the synchronization mark row 25 above to find the average white,
You may also determine the run length of black and use it as the standard value.

5-2, the first (black) clock of the left and right synchronization mark rows 25 on both sides is detected from the main scanning decoded image data 100, and the center point, length, and distance between the left and right first clocks are determined. The inclination of the image data 100 with respect to the main scanning direction (horizontal direction in FIG. 10) is actually measured. Then, in step 5-3, it is determined whether the actual clock length obtained in step 5-1 is within the standard value range. At this stage, anything that is not within the standard value range will result in a reading error. When it is within the standard value, for sub-scanning sampling, the value of the dot on the straight line connecting the center positions of the left and right start clocks is extracted from the main-scanning decoded image data 100, and the value of the dot in the first row of the mesh pattern 22 is extracted. Data indicating brightness is obtained, and each characteristic parameter (center position, length, inclination, etc.) actually measured in step 5-2 is set as a standard value for the next clock.

In 5-5, the length of the previous clock is added to the center position of the previous clock to predict the center position of the next clock. This prediction is
In this embodiment, the encoded image 20 is
(Figure 2) is due to the fact that the synchronization mark interest is a repeating pattern of black and white clocks of equal length, so that the clock length is equal to the clock interval. The next clock is actually measured by searching above and below the predicted point along the synchronization mark dot row on the image data that has been decoded in the main scan until a dot value different from the dot value of the predicted point is found.6 For example, if the predicted point is If it is "1" indicating a black pixel, consecutive "1"s above and below that point will be used as the next clock. Then, as actual measurement results, the length, center, left and right slope, etc. of the next clock are obtained (5-7). When prediction and actual measurement are performed in this way, if there is no error in the clock, the predicted center point and the actual measurement are obtained. The center point of the clock should be within a certain range, but if the clock is dipped, the clock length and slope will be at standard values (the length and slope of the previous clock), as shown in Figures 9 and 1O. It should change sharply from the previous left and right clock inclinations.

In FIG. 9, the dirt 63 is located at 2 of the synchronization mark row 25 on the right.
Because the second black clock is crushed.

Main scanning decoded image data lOO in FIG.
Above, a black dot row representing a fragment of the stain 63 surrounded by a dotted line in FIG. 10 is formed in the dot l of this synchronization mark row 25 (the row of pixels on the locus 31 in FIG. 9).

Therefore, the length of the next clock is longer than the previous clock (white clock, the center of actual measurement is indicated by Pl) in the dot row of the synchronization mark row 25 on the right by the amount of dirt. be done. Also, the predicted next black 7
The center P2 of the curve will also deviate greatly from the actually measured center P3. On the other hand, since there is no dirt on the corresponding part of the synchronization mark row 25 on the left, the center P2 of the next clock predicted from the previous clock (the center is indicated by PI) and the actual value P
There is only a small difference from 3. Therefore, by comparing the predicted feature parameters and the actually measured feature parameters for the next clock and checking the difference between the two, it is possible to detect the occurrence of a clock error due to dirt or the like.

In the flow shown in Figure 5, the difference between the predicted center point of the next clock and the measured center point of the next clock is determined for each of the left and right synchronization marks (5-8), and it is checked whether the difference is within the allowable range. (5-9.5-10.5-13) It is determined whether the primary clock is appropriate. 5-1
1 is a process that is performed when both the left and right clocks are not appropriate, and in such a situation, only the state before the synchronization mark row 25 can be seen reliably, so the left and right predicted points obtained in 5-5 are used as the center point of the next clock. Confirm as. And 5-
At step 16, a row of dots on a straight line connecting the center points is sampled from the main scanning decoding array 100 and used as data indicating the brightness of each related mesh. 5-12 and 5-14, when one of the left and right next clocks is correct and the other is incorrect, in this case, the center point of the incorrect clock uses the predicted point from the previous clock. However, in order to improve the accuracy as much as possible, the center point of the incorrect clock is calculated based on the slope from the actual measured center point of the correct clock. For example, in Figures 9 and 10, The correct tilt of the visible left and right white clocks is the center point P of the left clock.
It can be evaluated by the vertical difference 2 (dot) between 1 and the center point P1 of the right clock, and among the ring clocks located below each of these two white clocks, the right one is invalid and the left one is correct. Since the center point is actually measured at P2 in the figure, the position is two dots above the intersection of the horizontal line passing through this point and the straight line that is the dot row of the right synchronization mark row 25 (in this case, it happens to be the right The position P2 (which coincides with the predicted position P2 from the previous clock center Pi) is determined as the center point of the right fraudulent clock. Then, in 5-15, update only the clock length that was appropriate as the standard value of the clock interval, and in 5-16, sample the dot row between the centers of the left and right clocks.If both clocks are appropriate,
The center of the actually measured clock is determined and the dot row between them is taken out. In this case, in step 5-15, the standard parameters are all updated by the characteristic parameters of the actually measured clock.

5-17 is a decoding end determination, in which a predetermined number of data according to the format of the encoded image 20 is compared with the number of executions of the decoding process 5-16. If the number of executions of the decoding process 5-16 is equal to the number specified by the format, and the clock does not reach the end line in the subsequent dot row of the synchronization mark row 25 (5-18), that is, main scanning decoding is performed. If the number of clocks detected from the dot row of the synchronization mark row 25 on the image data 100 by actual measurement or interpolation is equal to the number determined by the format, it is assumed that proper processing has been performed and the sampling processing is completed. If incorrect clock interpolation is performed on the way, the end line of array 100 may be reached before the number of data in the format is obtained (
5-19: No), or after the format data count is obtained, another clock is found (5-18: No), so it can be detected as an error.

As described above, according to the first embodiment, reading of the encoded image 20 under fairly severe environmental conditions, for example, due to manual scanning of the encoded image 20 by the image sensor 11, a considerable amount of scanning occurs. Sampling standards for inclusion in the encoded image 20 (Guideline 21, synchronization mark In addition to devising the patterns in column 25), guidelines 21, which is the main scanning reference pattern, have been modified in case a part of the sampling reference pattern is destroyed during recognition of the sampling reference pattern for the read image data. The segment method makes it possible to actually measure the position of the guideline 21 in other parts, and interpolates the guideline position in the problem part based on the position information. The position of the clock is obtained by following the steps of prediction using characteristic parameters of nearby clocks, actual measurement from the predicted point, comparison of the actual measurement result with the predicted result, and re-determination of the position based on the clock error comparison result. Therefore, even in a relatively harsh usage environment with confusing characters, fluctuations in the scanning direction, dirt, etc., each position of the sampling reference pattern can be determined from the image data with the highest possible accuracy, and the results can be used to determine the position of the sampling reference pattern. The brightness and darkness of each mesh of the pattern 22 can be identified with high accuracy.

Figure 11 shows the overall configuration of a data reading device 110 according to a second embodiment of the present invention.
As shown in the figure, the control circuit 112 mechanically scans the image sensor 111.
In this configuration, the stepping motor 117 is driven by the stepping motor 117, and the image sensor 111 is moved along a path such as a predetermined rail. This has the advantage of significantly reducing the distortion of the read image data (
(This holds true even when the motor 117 is not specially controlled by a servo mechanism, etc.) Therefore, by utilizing this property, some of the components and processes described in the first embodiment can be simplified. For example, even if an encoded image 120 without a data end mark is used as shown in FIG.
From an accurate scanning speed of 1, it is possible to accurately predict approximately how many lines below the data start mark 23 the data will end. Furthermore, even if the search block 84 is made large, the position of the destroyed guideline can be interpolated with high precision. Furthermore, in the recognition process for the synchronization mark row 25, which is the sub-scan reference pattern, the standard value for comparison is not a local parameter such as the feature parameter of the previous black 7, but an average from the entire image. It is possible to use global parameters such as, which allows predicting, evaluating, etc. the next clock position with greater accuracy.

[Modification 1 This concludes the description of the embodiment, but various modifications and changes are possible within the scope of the present invention.

For example, the data start mark 23 described in the first embodiment
and data end marks 24 are used when the corresponding parts of these marks on the guideline 21 are destroyed.
Predict the guideline position of that part (4 in Figure 4).
-9.4-11), but the functions of marks 23 and 24 in the normal case, #i.e., the function of indicating the start and end of data, can also be achieved without these marks, e.g. , if there are no data start mark 23 and data end mark 24 in Fig. 2, these parts (white parts)
In the main scanning line image passed through the line, a continuous white pixel appears between the black fragments of the two guidelines 21, so that the start and end of data can be detected from this.

Further, in the above embodiment, the two guidelines 21 are arranged outside the mesh pattern 22 as the main scanning reference pattern in the encoded image 20.120, but any other suitable main scanning reference pattern can be used. For example, the 13th
As shown in the figure, if three (or more) parallel guidelines PR are used as the main scanning reference pattern for the mesh pattern MP, the guideline position can be more strongly restored against partial destruction of the guideline PR. 1.For example, 1 guideline PR out of 3
For the part where there is a fracture, the position of the fracture insulator line can be accurately determined from the corresponding positions of the remaining two guidelines. Furthermore, a main scanning reference pattern may be arranged inside the mesh pattern 22 as exemplified by the guideline PR in the middle of FIG. As illustrated in FIG. 14, the main scanning reference pattern may be composed of one guideline PR along the main scanning direction of the mesh pattern MP. In this case, the scanning direction (movement direction) of the image sensor is as follows. For example, it can be determined by examining a main scanning line image passing through the guideline PR. In addition, by examining such a main scanning line image, the positions of the end points of the guideline PR can be determined, and the positional relationship between the end points of the guideline PR and each mesh of the mesh pattern MP defined as the image format can be determined. From this, main scanning sampling can be performed by determining the center position of each mesh in the main scanning direction on the read image data. Furthermore, although the main scanning reference pattern is advantageously a continuous line of black or another color, it may also consist of discontinuous marks.

Further, in the encoded image 20.120 of the embodiment, two synchronization mark rows 25 that alternately repeat white and black meshes (clocks) on both sides of the mesh pattern 22 are used as the sub-scanning reference pattern. Any suitable pattern that is synchronized with the row of meshes in the sub-scanning direction of the mesh pattern 22 (which can be correlated in position) can be used as the sub-scanning reference. For example, if the scanning direction of the image sensor is stable. In this case, as shown in FIG. 15, it is sufficient to have one row of synchronization marks SR extending along the sub-scanning direction of the mesh pattern 22. In addition, three or more rows of synchronization marks may be arranged at intervals in the main scanning direction to enable more accurate restoration in the event of partial damage. For example, as shown in FIG. 16, a synchronization mark SR in which clocks in the form of elongated black fragments are lined up is placed above and below a mesh pattern MP with lines connecting the corresponding clocks. The sub-scanning reference pattern may be formed by arranging them so as to pass through the mesh boundaries of the mesh pattern MP.Also, as shown in FIG. 17, in addition to the first clock mark row SRI, A second clock mark row SR2 is provided with one clock mark for every predetermined number of clock marks in the first clock mark row SR, and both of them provide a sub-scanning reference for the mesh pattern 17. With this configuration, the number of clocks can be easily managed during image decoding work, and recovery from partial or missing lock marks can be made more reliable.
Initialize the clock mark counter for the first clock mark row SRI at the position found on the image data where the clock marks that match each other are read between the clock mark 15RI and the second clock mark row SR2. After that, the clock mark counter is incremented each time a clock mark on the first clock mark row SR is detected or determined by restoration according to the embodiment. Then, again, when a coincidence between the clock of the first clock mark row SRI and the clock of the second clock mark row SR2 is detected.

Examine the clock mark counter. In the case of the format shown in FIG. 17, if the value is equal to 8n (n is a natural number), it is considered that the first clock mark string has been properly recognized, and if n is 2 or more, the second clock mark string SR
It can be evaluated that 2 is missing, and if it is not equal to 8n, an error can be detected as an error in the recognition of the first clock mark. As a result, the first clock mark row SR is partially destroyed, and through the process of predicting, actually measuring, comparing, and reevaluating the next clock position as described in the embodiment (see FIG. 5), the clock is In the case of restoring
Errors introduced in the restoration process can be narrowed down to a localized area of the clock interval of the second clock mark row SR, and each clock position of that part can be reduced to the two clock intervals of the second clock mark row SR2. It can be obtained by dividing it equally between the clocks. Therefore, it is possible to identify the brightness and darkness of each mesh for the entire mesh pattern MP, even though it is acknowledged that there is such a local error (in the case of the embodiment, 5-17 to 5-19 in FIG. 5).
As shown in the figure, the entire result is an error and is considered unreadable).
In practical terms, since the mesh pattern data includes inspection codes, error correction processing using those inspection codes provides sufficient opportunity for correction of brightness discrimination data where there are local errors as described above. will be given.

In addition, in the embodiment, it was assumed that there is an image near or around the encoded image that is undesirable for decoding characters, etc., but it is of course not necessary to have it, and in that case, recognition of the sampling reference pattern etc. will be considerably easier. , the conditions (arrangement,
conditions such as geometric shape) are reduced.

In addition, in the embodiment, as sampling reference patterns, one for main scanning (guideline 21) and one for sub-scanning (guideline 21) are used.
Although the synchronization mark string 25) is realized by separate components on the image, a sampling reference pattern that serves both may also be used.

For example, returning to FIG. 16, the illustrated synchronization mark row SR can be used as a reference mark for both main scanning and sub-scanning. In the case of FIG. 16, the synchronization mark row SR is located outside the mesh pattern MP which is the data body. Therefore, if there are no unnecessary images other than the mesh pattern MP and the synchronization mark row SR, it is not difficult to detect the synchronization mark row SR from the read image data. Looking at the scanning line images, the black pixels appearing near both ends of the main scanning line image suggest the possibility of fragments of the synchronization mark row SR.In the manual mode, each main scanning line image is completely It is usually read not in the vertical direction, but in a slightly similar direction with fluctuations. Therefore, in each main scanning line image, main scanning line images having black pixel groups near both ends are consecutively distributed for several lines (for example, 2
~3 lines), if it appears, it is the synchronization mark string S
If there is a high probability that it is an R clock mark, and if it is distributed on the image data with a certain degree of periodicity among such black pixel groups, it is safe to assume that it is a clock mark. If we focus on such a group of black pixels for several lines and actually measure the center, we will find that it is the center of the clock mark. For the broken clock mark, for example, the distance between the previous clock mark and the previous clock mark (the distance between the center of the clock mark before the previous one and the center of the previous clock mark) can be determined in the same way as the corresponding process in Figure 5. It is possible to detect a large difference by comparing the predicted value and the actual measurement result. In this case, the predicted point can be determined as the center of the current clock mark, or other methods can be used. The center of the current clock mark can be determined from the center of the current clock mark in the row based on the slope of the previous clock mark. In this way, once the center position of each clock mark is determined, the upper and lower synchronization mark rows S
By selecting corresponding clock marks from R, connecting their centers, dividing the straight line into equal parts according to the image format, and reading the image bits at the equally divided positions, you can know the brightness and darkness at the center of each mesh ( Note that this last explanation assumes that each clock in the synchronization mark row SR in FIG. 16 is shifted by half of the mesh in the horizontal direction so that each clock coincides with the center of each mesh in the mesh pattern MP. ).

As shown in FIG. 16, when the length direction of each clock is arranged to match the boundary between the meshes of the mesh pattern MP, for example, the i-th and (i
+ 1) Center of the 1st clock mark U+, U+・l
, Bi, and B141, and from the position information, it is possible to calculate each center position of the target vertical mesh row. To explain this with reference to FIGS. 18 and 19, in FIG. 18, the centers U+, U+, + (on the read image data ) and calculate the midpoint C1. Similarly, the centers Bi of the i-th and (i+1)-th clock marks of the lower synchronization mark row SR,
Connect B+ and + to find the midpoint C2. FIG. 19 shows a portion of the read image data including midpoints CI and C2, and each row in the horizontal direction in the figure is a main scanning line image. One square in FIG. 19 represents a 1-bit (1 pixel) memory cell. By connecting these midpoints C1 and 02 and dividing the image into equal parts according to the format of the image, the center position, that is, the storage location (indicated by ) is determined by 0 or more, in the case of a midpoint method, the data sampling point is set to 2.
Since the calculation is made from a pair of two adjacent clock centers, the position U which is the fixed point [or interpolation point] of the synchronization mark sequence SRM
i, Ul-Bi, B1. The error included in l can be absorbed and reduced by averaging.

Note that the brightness and darkness of each mesh can be identified by data sampling only the points indicated by the X mark in FIG. The brightness and darkness of the mesh can be determined based on the results, or the pixel values at nearby stops can be determined by the synchronization mark row SR.
If there is a discrepancy with the pixel value of the point marked with an For example, if an error is detected when examining the light/dark discrimination data of a certain block according to the test word, if the previously discovered contradiction is included in that block, it can be detected as an error. So-called erasure correction (
Error correction when the order of error (n) is known is acceptable.

If the relationship between the sampling reference pattern and the mesh pattern that is the data body is a definable positional relationship, the image data is The center position of each mesh in the mesh pattern above can be determined. This will be explained with reference to FIGS. 20 to 22. In this case, it is assumed that there is some variation in the scanning speed and direction of the image sensor with respect to the encoded image, and for convenience of explanation, two rows of spaced synchronization marks with clock marks are used as the sampling reference pattern. It is assumed that a net-like pattern is arranged between the i marks.

Figure 20 shows the positional relationship on the image data read from the image sensor, so U] and U
1*+ is the center position of the i-th and (N+1)-th clocks in the upper synchronization mark row, B, and B1. i is the center position of the 1st and (N+1)th clocks in the lower synchronization mark row 0. For example, Ui and BJ, Ul,
I and B r-+ may be clocks having the smallest distance in the sub-scanning direction on the read image data. In FIG. 20, the coordinates of point U1 are the origin (0, 0
), and the coordinates of the point U i −1 are the point (x+
, O). This is merely a normalized representation of the position on the image memory for convenience of explanation, and the direction of the Y-axis is the direction of the main scanning line image.

FIG. 21 corresponds to FIG. 20, but shows the positional relationship on the recording medium. Ti and Tt・1 are the centers of the i-th and (N+1)-th clocks in the upper synchronization mark row,
A'' and Aj・1 are the i-th and N+ of the lower synchronization mark row
1) It is the center of the th clock mark. Each clock center (T1), (AI
) and the center position (P) of each mesh of the reference pattern are known matters to the data reading device and are held as an encoded image model.For example, all data at these positions may be stored. However, if there is synchronicity between positions, all center positions can be calculated from one or more position data and synchronization data. In Figure 21, T, and T
Although the line of the upper synchronization mark row connecting i and I is drawn parallel to the line of the lower synchronization mark row connecting A3 and A'', 1, this is not necessarily necessary.
The i-th clock center of the upper synchronization mark row is the origin (0,
0), and the (N+1)th claw center is shown as a point (D, O) on the X axis.

Now, suppose that each clock center of the synchronization mark string, which is a reference pattern, is obtained from the read image data in the manner described above, and four clock centers tL+ and Ut as shown in FIG. 20 are obtained. Suppose we are looking at a rectangular image block surrounded by l, Bl, and B1.1, and the problem is to know the center position of each mesh within this rectangular image block. The data reading device then calculates the corresponding clock centers Ti, Ti-+, Aj, Aj, + on the model (and therefore on the recording medium) from the encoded image model.
Select. Since the encoded image model includes the position information (P) of the center of each mesh in all encoded images, the rectangle (parallelogram here) surrounded by T1, TI, +, Aj, Aj, and 1. For example, in the case of FIG.

A set of coordinates (x, y) that satisfies O≦x+Ly≦D (L is the slope) may be selected from the coordinates (P) at the center positions of all the meshes.

The internal points of the rectangle surrounded by TI, Ti・1, Aj・Aj, + on the model correspond to the internal points surrounded by Ui, Ui・+, BJ, Bj−+ on the read image data, and their Therefore, once the center position of each mesh inside the rectangle is known in the x-y coordinate system of the encoded image model, each position is then added to the read image data. It is sufficient to perform coordinate transformation to a position in the coordinate system X-Y, and extract the image bit at the transformed position.

Second example of coordinate transformation? The figure (a) shown in the figure is an image block of the problem on the model, and the center of a certain mesh is indicated by coordinates (x, y).

FIG. 3(f) shows a corresponding image block (target image block) on the read image data, and the coordinates corresponding to the coordinates (x, y) are shown as (x, y). (a) to (b) are horizontal scaling to match the length of the top side of the rectangle, and (b) to (C) are the slopes of the bottom side of the target image block (Ui to BJ).
This is a conversion to match the slope of (C) to (d)
is scaling to match the length of the left side to the length of the left side of the target image block, and up to this point, the four vertices of the target image block Ul, Ui-+, B
j, B). 1, 3 points Ul, Ul-+, BJ
is the matching image block. So (e
), if the corresponding point of this image block (d) is shifted to point B, which is diagonal to point U1, the target image block will be obtained. An image block is a local area, and even in the case of one manual scan, the speed fluctuation and direction change within a local area like q are slight, so the conversion from (d) to (e) is There is no problem in considering this as a −-like distortion (distortion proportional to the area ratio).

As a result, the target coordinates (X, Y) are X-Xd・ΔO
・(raw rim)・(crazy) (1)XI Y2 Y-Yd・-H・(rim) worth (2)Xi
Y2 Here, ΔD= (X3- X2)-Xi
(3)ΔH=Y3-Y2
(4) XI Xd= (−→ −x+ (ML) ・y
(5) Yd=C/y (6) (The multipliers for ΔD and ΔH in equations (1) and (2) are the solid rectangle in (d) of Figure 22 and the coordinate m (
It shows the area ratio with the dotted rectangle whose diagonal points are It is the difference in length in the Y-axis direction, and the other Xi,
X2. For X3, Y2, and Y3, please refer to Figure 20),
In this way, the coordinates (x,
y), the coordinates (x, y) of the center of the corresponding mesh in the target image block are obtained.

Therefore, by sampling the image bits at the position (X, Y), information indicating the brightness of the mesh can be obtained. Note that the image frame formed by points T, , T, 1, Aj, and AJ-1 is actually a very elongated local area (see Figure 2), so it is assumed that one-handed scanning is performed. In the normal case, ΔH is ignored (ΔH=
O), therefore, instead of equation (2), Y=Yd
It can be simplified as

[Effects of the Invention] As described in detail above, in the present invention, image data that captures an encoded image on a recording medium is segmented into a plurality of partial image data blocks, and each segmented partial image data block is is searched for a feature pixel group characterized by a fragment of the main scanning reference pattern as a search unit, and the position of the main scanning reference pattern in the block of partial image data for which the search failed is indicated by the feature pixel group for which the search was successful. Since the position information is determined from the position information of a fragment of a healthy main scanning reference pattern, even if the main scanning reference pattern is partially destroyed due to dirt on the recording medium, the position can be determined with high precision, and the main scanning There is an advantage that brightness and darkness in each mesh of the mesh pattern can be reliably distinguished based on the positional information of the reference pattern.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram of a data reading device according to a first embodiment of the present invention. Fig. 2 is a diagram illustrating an encoded image on a recording medium to be read, and Fig. 3 is a diagram showing an example of an encoded image on a recording medium to be read.
Figure 4 shows an example of an image obtained by scanning an encoded image. Image data obtained by scanning an encoded image is stored, and a guideline, which is a main scanning reference pattern, is recognized from the stored image data to perform main scanning decoding. FIG. 5 is a flowchart for performing sub-scan decoding by recognizing a synchronization mark string, which is a sub-scan reference pattern, from main-scan decoded image data. FIG. 6 is a diagram showing a memory map of the image RAM, and FIG. 7 is an explanatory diagram of a guideline set that characterizes guideline fragments. FIG. 8 is a diagram showing an encoded image with surrounding characters, etc., and a plurality of search blocks in which the encoded image is divided; FIG. 9 is a diagram illustrating an image of a guideline with dirt and a synchronization mark string; Fig. 10 is a diagram showing main scanning decoded image data corresponding to the image in Fig. 9, Fig. 11 is an overall configuration diagram of the data reading device according to the second embodiment, and Fig. 12 is a diagram showing the image data without a data end mark. The figure which shows the example of an encoded image. Figure 13 is a schematic diagram of an example of an encoded image when there are three guideline trees. Figure 14 is a schematic diagram of an example of an encoded image when there is one guideline. Figure 15 is a diagram of an example of an encoded image when there is one synchronization mark row. A schematic diagram of an example of an encoded image. FIG. 16 is a schematic diagram of an example of an encoded image with a synchronization mark row of elongated black marks arranged as a clock, FIG. 17 is a schematic diagram of an example of an encoded image with two types of synchronization mark rows with different periods, Fig. 18 is a diagram used to explain the process of identifying the brightness and darkness of each mesh using the midpoint method from the center of the four clocks of the synchronization mark string, and Fig. 19 is an image including the midpoints C1 and C2 shown in Fig. 18. Figure 5S20 is a diagram showing the arrangement of image data on the RAM; Figure 5S20 is a diagram showing the relationship between the four positions of the sampling reference pattern on the read image data and the positions of internal meshes. FIG. 21 is a diagram corresponding to FIG. 20, but showing the relationship between the four positions of the sampling reference pattern on the recording medium and the positions of internal meshes. FIG. 22 is a diagram used to explain the process of converting the center coordinates of the mesh on the recording medium to the center position of the mesh on the image data. 20...Encoded image, 21...Guideline (main scanning reference pattern), 22...Mesh pattern, 25...Synchronization mark string (sub-scanning reference pattern) pattern), 11.111... Image sensor-113... CPU, 14... ROM
, 15... Image RAM. Fig. 2 Fig. 1 Engraving of drawing Fig. Fig. 9 Fig. 0 Fig. 13 Fig. 4 Fig. 5 Fig. 16 Fig. 5R2-111 5R1,1111111111111111111 Fig. 7 Fig. 8 Fig. 8 (a) i is Figure 1 in [F]) Figure 2 (C) Procedural Amendment (Method) 26th October, 1999

Claims (2)

[Claims] (1) In a data reading method for reading data encoded in an image recorded on a recording medium, (A) a mesh pattern in which data is encoded by brightness and darkness formed in each of a plurality of meshes arranged vertically and horizontally; ,
(B) reading image data representing the image from a recording medium on which an image consisting of a main scanning reference pattern for indicating a data sampling standard in the main scanning direction of the mesh pattern is recorded; (B) reading the read image data; segmenting into a plurality of partial image data; (C) searching for a characteristic pixel group characterizing the fragment of the main scanning reference pattern from among the partial image data for each segmented partial image data; (D) searching; Regarding the partial image data for which the characteristic image group was not detected in the search for other partial image data, the above-mentioned characteristic image group in the partial image data is determined from the position of the fragment of the main scanning reference pattern indicated by the characteristic pixel group detected in the search for other partial image data. (E) generating positional information of each fragment of the main scanning reference pattern in the image data by determining the positions of the fragments of the main scanning reference pattern; A data reading method comprising the step of identifying the brightness and darkness of each mesh of a pattern. (2) In a data reading device that reads data encoded in an image recorded on a recording medium, (A) a mesh pattern in which data is encoded by brightness and darkness formed in each of a plurality of meshes arranged vertically and horizontally; ,
(B) image sensor means for reading image data representing the image from a recording medium on which an image consisting of a main scanning reference pattern for indicating a data sampling reference in the main scanning direction of the mesh pattern is recorded; (C) for each segmented partial image data, a feature pixel that characterizes a fragment of the main scanning reference pattern from among the partial image data; (D) For partial image data in which the above-mentioned characteristic image group was not detected in the search, the above-mentioned characteristic pixel group indicated by the characteristic pixel group detected in the search for other partial image data; A main scanning reference that generates positional information of each fragment of the main scanning reference pattern in the image data by determining the position of the fragment of the main scanning reference pattern in the partial image data from the position of the fragment of the main scanning reference pattern. A data reading device comprising: pattern position generation means; and (E) identification means for identifying the brightness and darkness of each mesh of the mesh pattern in the image data based on the generated position information.