JP2000285190A - Method and device for identifying slip and storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform slip identification with high performance by previously classifying registration slips having a plurality of characteristics, such as color information, ruled line information or the recognized results of preprinted characters with a common characteristic. SOLUTION: Registration slips having plural characteristics, such as color information, ruled line information or recognized results of preprinted characters are previously classified with a common characteristic. In this device, an image inputting part 1 scans respective slips to be filed and gives an obtained image to a slip identifying part 5. The part 5 performs identification on the basis of plural pieces of format information registered with a format information storing part 4 and according to which slip among registered slips a slip to be filed coincides with. Then, the results are shown on a display device 6, and the obtained identification results are presented to a user to perform confirmation by interactive processing with the user. Subsequently, the user instructs processing by using an instructing device 7.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a form identification method for efficiently identifying and classifying forms.

[0002]

2. Description of the Related Art Conventionally, in order to identify an image, for example,
Japanese Patent Application Laid-Open No. 61-75477 discloses an image of a sample document (plural) of a target document to be stored or a feature pattern thereof is stored in advance, and a feature pattern is extracted from the image of the target document to be stored. There has been proposed a method of automatically determining a classification code by assigning, to an input document, a classification code of a sample document which is determined to be the same as the characteristic pattern of the sample document, and which is determined to be identical.

However, for example, in Japanese Patent Application Laid-Open No. 61-75477, an integrated value of horizontal and vertical line segments of an image (addition of the number of pixels of long line segments in horizontal and vertical directions) is used.
If different documents have the same linear portion, it cannot be distinguished, and if the image is skewed or misaligned, there is a problem that it is difficult to perform alignment and skew correction at the image level. .

[0004] In order to solve this problem, for example, Japanese Patent Application Laid-Open No. 08-255236 proposes a process for performing identification by collating information of ruled lines extracted from a form image between forms using a hash table. are doing. This is to check the connectivity of the ruled line frame extracted from the form image against the registered form, normalize it by the circumscribed rectangle of the connected ruled line frame,
The center coordinates of each ruled line frame are registered in the hash table. On the other hand, a ruled line frame is similarly extracted from a form image for a form to be identified, and normalized based on a linked ruled line frame. Finally, there is a method in which the registered hash table is searched using the center coordinates of each ruled line frame, and the type of the registered form that hits the most is set as the corresponding form.

[0005]

[0006] As a large number of types of forms are handled, a method based on only the physical structure of a ruled line frame may be used to classify as different types of forms even if they have the same structure. However, even if the frame information matches, only the pre-printed characters differ, and only the color information differs, the recognition performance is limited.

Therefore, the present invention is not limited to a physical structure such as a ruled frame, but a form identification method capable of performing a form identification with good performance using a plurality of features such as color information and a keyword recognition result. And a form identification device using the same.

[0007]

According to a form identification method of the present invention, a plurality of features are extracted from an input image of a form to be identified, and one of the features is compared with the features of a registered form. When the type of the form to be identified cannot be identified,
By comparing the other feature amount of the plurality of feature amounts with the feature amount of the registration form to identify the type of the form to be identified, limiting to only a physical structure such as a ruled line frame. Instead, it enables high-performance form identification using a plurality of features such as color information and keyword recognition results.

According to the form identification method of the present invention, at least one of a color, a ruled line, and a character string for specifying a form type from an input form image is extracted as a feature quantity, and a plurality of registered forms are extracted. These are classified in advance by the feature amounts common to them, and among the feature amounts extracted from the input image of the form to be identified, a feature amount used when identifying the type of the form to be identified is selected. Then, by identifying the type of the form to be identified based on the characteristic amount of the registered form classified by the selected characteristic amount, while narrowing down the search range, it is possible to identify a large number and variety of forms. A good form identification result can be obtained.

Preferably, a feature quantity to be extracted from the image of the form to be identified is selected according to the type of the image of the form to be identified.

Preferably, when a form matching the characteristic amount extracted from the image of the form to be identified is not registered, the form is registered using the characteristic amount of the registered form most similar to the characteristic amount. .

Preferably, when extracting a color as a feature of a form, the color of each pixel is determined based on the colors of peripheral pixels.

A form identification apparatus according to the present invention includes an extracting unit for extracting a plurality of feature amounts from an image of an input form to be identified, and one of the plurality of feature amounts extracted by the extracting unit.
A first identifying means for comparing the characteristic amount of the registered form with the registered form to identify the type of the form to be identified; and when the type of the form to be identified cannot be identified by the first identifying means,
A second identification unit that identifies the type of the form to be identified by comparing another characteristic amount of the plurality of characteristic amounts with the characteristic amount of the registration form, thereby providing a ruled line frame or the like. It is possible to perform high-performance form identification using a plurality of features such as color information and a keyword recognition result without being limited to only the physical structure.

[0013] The form identification apparatus of the present invention comprises: an extracting means for extracting at least one of a color, a ruled line, and a character string for specifying a form type from an input form image as a feature amount; A classifying means for classifying the registered form by the characteristic amount common to them, and a characteristic used when identifying the type of the form to be identified among the characteristic amounts extracted from the input image of the form to be identified. Select the amount,
Identification means for identifying the type of the form to be identified based on the characteristic amount of the registered form classified by the selected characteristic amount, thereby enabling identification of a large number and variety of forms while narrowing the search range. And a high-performance form identification result can be obtained.

Preferably, a feature quantity to be extracted from the image of the form to be identified is selected according to the type of the image of the form to be identified.

Preferably, when a form matching the characteristic amount extracted from the image of the form to be identified is not registered, the form is registered using the characteristic amount of the registered form most similar to the characteristic amount. .

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows an example of the configuration of a form identification apparatus for identifying a form image according to the present embodiment. The form input unit 1, display device 6, input device 7, and various storage means (input data) The memory 3, the storage unit 8, and the format information storage unit 4) are connected to a bus.

FIG. 1 shows the form registration unit 2 and the form identification unit 4 as hardware resources. The image input unit 1, display device 6, input device 7, and various storage means (input data memory 3, storage Section 8, a format information storage section 4) and a program which can be executed by a computer having a CPU.
It can also be recorded on a storage medium such as a D-ROM and distributed.

The form identification apparatus shown in FIG. 1 has two modes, a form registration mode and a form identification mode, and operates in the form registration mode as follows. That is,
The image input unit 1 scans a form (registration form) to be registered and provides the obtained image to the form registration unit 2. The input data memory 3 stores an input image.
The form registration unit 2 performs recognition processing such as color separation, line segment extraction, character frame extraction, character recognition, and frame extraction on the input image, and displays the recognition result on the display device 6 to present to the user. The various components of the registration form are extracted, the format information is generated, and registered in the format information storage unit 4 by the interactive processing of.

On the other hand, the following operation is performed in the form identification mode. That is, the image input unit 1 scans each form to be filed and provides the obtained image to the form identification unit 5. The form identification unit 5 is a format information storage unit 4
Based on a plurality of pieces of format information registered in, a form to be filed is identified based on which of the registered forms matches. Then, by displaying the result on the display device 6, the obtained identification result is presented to the user, and after confirming the result by interactive processing with the user, for example, from an input device such as a mouse, a keyboard, and an electronic pen. Processing is instructed by the user using the instruction device 7 configured. Finally, based on the format information of the registered form registered by the form registration unit 2 and the format information of the form obtained by the form identification unit 5,
The input image is filed in the storage unit 8. However, it is also possible to omit this interactive processing, perform automatic saving, and perform a confirmation operation after saving a plurality of sheets.

Next, the processing performed in each of the above operation modes will be described in more detail. First, the form registration mode will be described.

FIG. 2 is a flowchart for explaining the processing operation in the form registration mode. First, the image of the registration form input by the scanner is displayed on the display device 6 (step S1). Then, the form registration unit 2 performs a color separation process on the input image to extract a background image, ruled lines, and a pre-printing unit (step S2). Next, a color analysis process (step S3) and a binarization process (step S4) are performed on the data obtained as the ruled line / preprint portion. Further, an element extraction process (step S5) is performed to extract a characteristic element (feature amount) used for identifying a form type such as a ruled line, a character frame, and a character code. Finally, the data is edited by the addition / correction processing (step S6) according to the instruction from the user, and finally the format information is created and registered in the format information recording unit 4 (step S7).

Next, the background image detection and its removal processing in the color separation processing in step S2 will be described.

For the input image, the RGB values are converted into HSV (hue, saturation, lightness) for each pixel. From this value, a histogram is taken, and the value with the largest value is set as the background color value. After the background color is determined, the background portion is removed from the entire image. Of course, other methods may be used to detect the background color.

Since the input image can be considered to have a background portion, a ruled line portion, and a character portion, the image left when the background image is removed is considered to be a ruled line / pre-printed character portion. Therefore, after binarizing these images (step S4), performing color analysis and extracting color features (step S3), various elements such as graphic features and character codes are extracted from the images. (Step S
4).

In the binarization process in step S4, for example, since the background portion has been deleted by the color separation process in step S2, the remaining image is considered to be elements such as ruled lines, preprinted characters, and handwritten characters. Therefore, it is only necessary that binarization can be performed so that data other than the background becomes black.

The color analysis in step S3 is performed by, for example, the following method. Since the background portion has already been deleted from the input image, only the color portion is processed. RGB space or H (hue) -S only for color part
(Saturation) Voting in the space, and determining the colors present in the form by clustering. For example, as shown in FIG.
A ballot box corresponding to a color component of a ruled line used in a form is previously arranged in an HS color space, a color component is calculated for the extracted pixels, and a vote is cast in the ballot box. Take the way to go. For example, a ballot box (region) R1 corresponding to red and a ballot box (region) R5 corresponding to blue are provided.
Voting is performed using the color component of each pixel, and finally, the number of votes and the ratio of the voting pixels to the total number of pixels are calculated to determine whether a winning or a failure has occurred. With this method, the representative color of the target form is recorded. Not only the color type but also the ratio of the color used in the form may be stored. The method of analyzing the color feature may use other methods that have been proposed.

In the element extraction in step S5, a ruled line frame (ruled line, character frame), a character code, and the like are extracted. The ruled line is extracted by, for example, the following method. That is, here, the ruled line is considered as frame information composed of four points, and extraction is performed. First, after input data is raster-scanned and its tracking start point is detected, a contour is extracted by tracing a closed curve having a figure boundary clockwise or counterclockwise from the tracking start point. The information on the contour extracted by such closed curve tracking is stored as a coordinate point sequence or a chain code which is a sequence of direction codes.
It goes without saying that other outline extraction methods conventionally proposed variously may be used for the outline extraction processing.

Next, feature points such as intersections and corner points of the image are extracted based on the contour data. In this method, for example, a sequence of convex points is detected from a sequence of contour points, and a sequence of concave points corresponding to the sequence is extracted. In this feature point extraction, for example, the curvature k obtained by the curvature calculation processing is compared with predetermined thresholds K0 and K1, and k ≦
This is realized by setting the point sequence K0 as a concave point sequence and the point satisfying k ≧ K1 as a convex point sequence.

The feature point extraction processing may be realized by a graphic thinning processing or a core line processing. For example, in the thinned data, since a convex portion exists at an end point of a line segment and a concave point exists at an intersection of the thin line, feature extraction becomes possible.

Next, a frame is extracted from the combination of these point sequences. The frame can be extracted easily by judging whether or not the figure forms a rectangle or a parallelogram from a combination of coordinates of four points near the point of interest, for example. At this time, it goes without saying that other methods may be used for frame extraction.

The character frame extraction is performed, for example, by the following method. The image is labeled for each connected region, and a circumscribed figure is obtained. Next, when adjacent circumscribed figures are within a certain threshold distance, they are merged. When the merging is performed, the circumscribed figure is calculated again. If the change in the direction of the underline (inclination) is within a certain threshold, it is determined that the character strings are the same. Note that the character frame extraction method may use another method such as a conventionally proposed method.

The character extraction is performed, for example, by the following method.
The image is labeled for each connected region, and a circumscribed figure is obtained. If the size of the circumscribed figure is within a certain threshold, character recognition is performed as a character candidate. It should be noted that the character extraction method may use another method such as a conventionally proposed method.

For the elements extracted as described above,
In step S6, addition and correction are performed.

Various elements extracted from the input image displayed on the display device 6 are displayed, for example, superimposed on the image of the registration form input in step S1, so that the user can correct the data. ing. The correction method can be performed in the same manner as a conventionally used graphic creation tool or the like.

Finally, the format information finally created by the correction by the user is stored in the format information storage unit 4.
(Step S7). At this time, a keyword or a file name is given to the format information and stored.

Here, the features of the color form will be described with reference to FIG. The form includes ruled line frames, ruled line colors, and preprinted character string (pre-printed character string) regions R1 and R2 as its constituent elements.

FIG. 5 shows four examples of the format information (here, the image is shown) of the registered registration form, and each registration form has a feature amount including a difference in color information. Elements (ruled line frames, character strings (pre-printed characters)) have been extracted.

These forms are common only to color information.
Some of the preprinting are common, some of the ruled line frames are common, and the like.

For example, in a form identification mode to be described later, a registration form and a form to be identified first (a first feature amount) are color information. That is, as shown in FIG. 6, when the ruled line color is used as the first feature amount, the registered forms A and B, and C and D are classified as the same format group. When the preprint # 1 at the head of the form is classified as a second feature amount based on this, form samples C and D
And are classified as the same format county. Further, when the preprint # 2 at the lower right of the form is used as the third feature amount and classified based on this, the form samples A to C are classified as the same format.

Not only the classification method shown in FIG. 6, but also the format information of a plurality of registration forms, only the structure of the ruled line frame matches, the place where the pre-print is made, its character string (character code), the ruled line Since there are many types such as ones that differ only in color, it is desirable to classify registration forms from the viewpoint of being as easy to separate as possible.

Next, the form identification mode will be described in detail.

FIG. 7 is a flowchart for explaining the processing operation in the form identification mode. First, the image of the form to be identified input by the scanner is displayed on the display device 6 (step S11). And
The form identification unit 5 performs an image separation process on this input image (step S12), and extracts a background image, a black area (character string portion), a ruled line, and a preprinting unit.

The data obtained as the ruled line / preprint portion is subjected to color analysis processing (step S13) and binarization processing (step S14). Further, an element extraction process (step S15) is performed to extract elements such as ruled lines, character frames, and character codes.

The background image detection processing, the color analysis processing,
The process of extracting elements such as ruled lines, character frames, and character codes, and the binarization process are the same as those at the time of registering a form, and therefore will not be described. Here, it is not always necessary to perform the character recognition processing. The character recognition processing may be performed in step S19 described below, if necessary.

The black area extraction processing in step S12 is performed, for example, by the following method. In the black region, when RGB values are converted into HSV (hue, saturation, lightness) for each pixel, lightness V and saturation S are reduced. Therefore, the black pixel portion can be extracted by the threshold processing. Of course, other methods may be used to extract the black pixel region. Of course, when black is used for the ruled lines and pre-printed characters of the form, the black area may be handled in the same manner as the ruled line / pre-printed portion.

Next, from the data obtained by the element extraction processing in step S15, a registration form and a form to be identified which are to be first identified (first feature amount) are selected (step S16). Then, form identification is performed (step S1).
7).

Here, as an example of the form identification processing, a case where form identification is performed using a hash table will be described. This method is disclosed in Japanese Patent Application Laid-Open No. 08-25523.
No. 6, which can be divided into pre-processing for creating a hash table from feature amounts extracted from a registration form, and recognition processing for extracting features from an input image and voting and recognizing them.

When a hash table is created here, a result obtained by classifying a plurality of registration forms may be treated as one registration form. In this case as well, the following description is directly applicable.

First, a process of creating a hash table will be described with reference to FIG. As described with reference to FIG. 2, when registering a form, an image of the form is input (step S31), and each component of the form is extracted (step S32). Therefore, in order to create a hash table, based on the format information already stored in the format information storage unit 4, feature amounts such as the center coordinates of the ruled line frame, the center coordinates of the character string frame, and the connection frame information are extracted. , These values are normalized. For example, a relative value when the maximum value is “1” is calculated (step S33).

Using the standardized values, a hash table is created for each feature amount such as the center coordinates of the ruled line frame and the center coordinates of the character string frame.

FIG. 9 shows an example of a hash table when the center coordinates of the ruled line frame are used as the feature values. This shows a case where the center coordinates of the ruled line frames of all the registered forms (A to F) are embedded in the hash table. Since it is assumed that the inclination has been corrected based on the straight line data, the ruled line frame may be considered to be composed of only horizontal and vertical line segments. Here, how to take a base (a parameter of a feature amount to be normalized) when creating a hash table is represented by a point located on a horizontal / vertical line segment. In particular, in the case of a form, since there are many connected rectangular frames, the width and height information of the circumscribed figure of the connected rectangular frame is used as the base. For the rule, the size of the form itself may be used, or the standard may be unified.

Next, a form identification processing operation using a hash table will be described with reference to a flowchart shown in FIG. As described with reference to FIG. 7, when identifying a form, an image of the form to be identified is input (step S41), and features such as ruled lines and character frames are extracted (step S42).

Further, from the extracted feature values, feature values such as the center coordinates of the ruled line frame, the center coordinates of the character string frame, and the connection frame information are extracted, and these values are normalized (step S4).
3).

Then, the area of the feature point corresponding to the feature amount on the hash table is searched using the standardized feature amount (step S44). This area is a search area of a predetermined range, and if there is a characteristic point of the registered form in that area, the registered form is voted for (step S45). Specifically, a histogram is created for each registered form name (format number), and voting is performed for each registered form name (format number) in the searched area. These processes are repeated by the number of feature points of the form to be identified (step S46).
The similarity with the registered form having the largest histogram is calculated and output as an identification result (step S47, step S48).

For example, in a hash table having the center coordinates of a ruled line frame as shown in FIG. 11, it is assumed that there are six registered forms, and a form (sample form) having a format number "A" is identified as a form to be identified. The case will be described. The sample form has nine feature points, and the hatched portion on the hash table in FIG. 11 is a search area of a predetermined range corresponding to each feature point. In this case, as shown in FIG. 12, as a result of voting for each registered form based on the characteristic points of the registered form existing in each search area corresponding to the characteristic point of the sample form, the voting of the registered form of format number "A" is performed. The number is the largest, "8", followed by the number of votes of the registration form of format number "D""2", the number of votes of registration form of format number "E""1", and the registration form of format number "F""1".

Therefore, the similarity with the above four registration forms is determined. The similarity is assumed to be defined by the following equation (1), for example. That is, assuming that the number of votes of each registered form t is Nt, the number of rectangular frames of the form to be identified is k, and the number of rectangular frames of the registered form t is Rt, the similarity S is S = (Nt × 2 × 100) / (K + Rt) (1)

FIG. 12 shows the result of calculating the similarity using the equation (1).

Therefore, it can be seen that the most similar to the sample form is the registered form with the format number “A” having the highest similarity.

Although the method of identifying a form using a hash table has been described above, the present invention is not limited to this case, and another identification method may be used.

Next, it is considered that a detailed check is performed on the form to be identified (sample form) using the second feature value. At this point, if there is only one registered form with the format number "A", it is sufficient to output the registered form as an identification result (step S26), and the detailed investigation using the second feature amount can be omitted.

For example, as shown in FIG. 13, it is assumed that a registration form having a format number "A" includes two types of registration forms as shown in FIGS.
These two registration forms have the same ruled line frame and the same color of the ruled line frame, and the different portions are character strings preprinted. Therefore, the format information of these two registration forms includes a character frame (character area) and a character string necessary for identification.

For example, as shown in FIG.
In the format information of each registration form shown in (a) and (b), two pre-print areas R1 and R2 for character identification are registered. If a detailed investigation is required, the character strings described in these two places are sequentially extracted from the form to be identified (sample form) and character recognition is performed (step S).
18 to Step S19). If it is not possible to determine one registered form by character recognition in one place, character recognition is repeated until it can be determined (steps S18 to S19).

An existing method may be used for character string extraction and character recognition. The character string recognized here is shown in FIG.
The detailed identification result is determined based on whether the form matches one of the forms (a) and (b) or the degree of match is high.

When the character recognition is repeated and the detailed check of the form to be identified becomes unnecessary, the form to be identified matches one registered form (step S18), and the identification result is output or stored (step S18). Step S26).

In the example shown in FIG. 13, if the character is recognized only in the pre-printing area R1, it is possible to determine which form it is. However, if it cannot be determined here, the character is further recognized in the pre-printing area R2. The detailed investigation is repeated using the amount. Here, the identification may be performed by changing the feature amount until the form is determined, or the most similar one may be presented. If there is no corresponding form, it is possible to newly register and add.

(Second Embodiment) Next, the classification of registered forms in form registration will be described in more detail.

FIG. 15 shows the registered registration forms (A to E).
The format information is shown as an image.
In FIG. 15, registration forms A, B and C, and D and E have a common ruled line structure, and have registration forms A and B respectively.
And E, C and D are common as colors used. In this way, by performing classification based on each feature (color, ruled line, preprint # 1, preprint # 2), each feature can be classified into two groups as shown in FIG.

Here, in the form identification mode, assuming that the first feature amount selected in step S16 of FIG. 7 is color information, the registered form narrowed down at this stage is
Further, using the ruled line as the second feature amount, the form is identified using the hash table as described above. In this way, assuming the feature amount used for identification at the time of form registration,
By classifying the registration forms, the identification work can be made more efficient and identification errors can be reduced.

(Third Embodiment) Next, a case where a feature amount used for identification is selected according to the type of an image of a form to be identified will be described with reference to a flowchart shown in FIG.

First, the image of the form to be identified input by the scanner is displayed on the display device 6 (step S51). Then, it is determined which image type (binary image, grayscale image, color image) this input image belongs to (step S52). The image type may be determined by instructing which image type the user has.

The feature amount used for identifying the form differs depending on the image type. That is, for example, in the case of a binary image, a ruled line frame and a pre-printed character (step S54), in the case of a grayscale image, the density distribution of shades, the ruled line frame, a preprinted character (step S55), and in the case of a color image, A ruled line color, a ruled line frame, and a pre-printed character (step S56) can be given as the feature amount. Finally, a first feature value is selected from the feature values (step S57).

For example, in the case of a binary image, a ruled line frame is selected as the first feature amount. Similarly, a density distribution is selected as a first feature amount for a grayscale image, and a ruled line color is selected for a color image. After that, only the feature amount extraction processing corresponding to each image type is performed, and the form identification using the hash table, the collation of the pre-printed characters, and the like may be performed.

In this case, since the amount of features to be extracted in advance is limited according to the image type, the time required to obtain a form identification result can be reduced.

The feature values used for identification are not limited to the above feature values, and other feature values may be used. Similarly, the feature amount used for identification is not limited to the first feature amount, and the second and third feature amounts may be used together depending on the image type.

(Fourth Embodiment) Next, FIG. 18 shows a processing operation when a corresponding form is not obtained as a result of form identification, that is, when an unregistered form is identified. This will be described with reference to the flowchart shown.

The result of performing the processing shown in FIG. 7 (step S
61, step S62), if there is a registered form identical to the form to be identified, it is output as a form identification result (step S63, step S64). The above detailed processing is the same as the flowchart shown in FIG.
On the other hand, when there is no registered form identical to the form to be identified (step S63), it is determined whether there is a similar form (step S65). Whether or not the form is similar is evaluated using a form that can evaluate the similarity of the form, such as the similarity used at the time of the search. Then, those having a similarity higher than a predetermined threshold may be presented, or those having the highest similarity may be presented. If there is no similar form, the entered form to be identified is newly registered as shown in FIG. 2 (step S6).
6).

If there is a similar form, the form information to be identified is additionally registered using the format information of the similar form (step S67).

For example, it is assumed that a registration form of format information as shown in FIG. 19A has already been registered, and the input form to be identified is shown in FIG. 19B. Then, it is assumed that the one most similar to the input form is the registration form shown in FIG.

The form shown in FIG. 19A and the form shown in FIG. 19B differ only in the pre-printed characters. In this case, as a modification of the registration form shown in FIG. 19A, a different part in the form shown in FIG. Additional registration is performed to register a new registration form.

As a registration form registration method using the format information of another registration form as described above, for example, as shown in FIG. 20, the ruled line type, the color, and the pre-printed character are separately registered as parts, and The registration form is associated with the format number, for example, as shown in FIG.
It is expressed by the combination of parts shown in.

As described above, the elements of the registration form are made into components, and by combining them, the format information of one registration form is created. And the amount of registered data can be reduced.

In this case, form identification can be performed in the same manner as described above.

(Fifth Embodiment) The same can be said for the form registration mode and the form identification mode. For example, the color analysis processing in step S3 in FIG. 2 and the step S1 in FIG.
In the color analysis process in No. 3, a process in a process of determining, for each pixel of the input form image, whether or not the color information of the pixel is a color to be extracted will be described. That is, when the color to be extracted is extracted from the state where the color to be extracted and the color to be removed or the background color exists, or when the operation to remove unnecessary ones is performed, Then, the state of the color of the pixels surrounding the pixel determined to be an intermediate color, which is an intermediate state between the two, is checked to determine whether or not the pixel has a color to be extracted.

Referring to the flowchart shown in FIG.
The above-described color determination processing operation will be described.

First, the process of extracting a black area will be described. FIG. 23A shows an original image. Here, the area x1
Is a black line, and the regions x2 and x3 are colored lines. FIG.
FIG. 23B shows image data immediately after scanning the original drawing shown in FIG. In FIG. 23B, the region x
5 is a portion determined to be black, and regions x4 and x6 are portions determined to be colored. In the scanner, like this,
In some cases, the black region may be colored on the boundary with the colored region. Therefore, first, only the black region x5 is extracted from the data as shown in FIG. 23B (step S71). Next, as shown in FIG. 23C, the peripheral areas of the black area x5 are x4 and x6, and the colors of the areas x4 and x6 are checked (step S72). Region x
If the color of 4, x6 is not white, that is, if it is an intermediate color, this area is determined to be black (steps S73 and S74). Then, the area x determined to be black at this time
4, x6 and an area x5 originally determined to be black are defined as a black area as shown in FIG.

On the other hand, in the case of the colored area extraction processing, the reverse is performed. First, regions x4 and x6 determined to be colored
Is extracted, and a region that is in contact with the region and is not in contact with the region x5 determined to be a black region, that is, the region x2,
A portion corresponding to x3 is extracted, and this is determined as a colored region.

However, as shown in FIG. 24, when the left boundary portion x12 of the black region x13 is not colored, the above-described method causes the region x12 to be omitted from extraction without being regarded as colored. . Therefore, the following processing is performed.

First, an area that is determined to be colored and is not in contact with the area x13 determined to be a black area, that is,
An area x11 is extracted, and colored areas x12 and x14 that are in contact with another black area x13 are extracted. Region x
12, x14, the area determined to be colored, that is, the portion in contact with the area x11 is determined to be colored.
By doing so, the regions x11 and x12 are extracted as colored.

Although the above description has been made on the basis of the classification of white, black, and colored, any color may be used. In general, for the color to be extracted (a plurality of colors may be used, similarly in the following, "color" does not necessarily mean a single color. Also, the color includes white, black, gray, etc.) , When extracting the color to be extracted or removing unnecessary ones from the state where other colors or background colors exist, introduce the concept of the intermediate color of the two and judge that it is this intermediate color It is determined whether or not the pixel should be finally extracted based on the state of the pixels around the pixel thus selected, that is, whether or not the pixel should be extracted or not, and whether or not the pixel is an intermediate color.

The intermediate color may be defined as, for example, a color defined as an average of a color to be extracted and a background color.

(Effects) As described above, according to the above embodiment, registration forms having a plurality of characteristics such as color information, ruled line information, and recognition results of preprinted characters are classified in advance by common characteristics. As a result, it is possible to identify a large number and variety of forms while narrowing the search range, and to obtain a form identification result with good performance.

By managing a form having a common element as a combination of the common elements, a form can be registered efficiently.

[0094]

As described above, according to the present invention,
It is possible to identify a large number and variety of forms having a plurality of characteristics, and obtain a form identification result with good performance.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration example of a form identification device according to a first embodiment of the present invention.

FIG. 2 is a flowchart for explaining a vote registration processing operation.

FIG. 3 is a diagram showing an example of a voting space for color determination.

FIG. 4 is a view for explaining features of a form.

FIG. 5 is a diagram showing a specific example of format information of a registration form.

FIG. 6 is a diagram showing a classification example of a registration form.

FIG. 7 is a flowchart for explaining a vote identification processing operation;

FIG. 8 is a flowchart for explaining a hash table creation processing operation.

FIG. 9 is a diagram showing a specific example of a hash table based on the center coordinates of a ruled line frame.

FIG. 10 is a flowchart for explaining a form identification processing operation using a hash table.

FIG. 11 is a diagram specifically showing a search area on a hash table.

FIG. 12 is a diagram showing an example of a registration form search result using a hash table.

FIG. 13 is a view showing an example of a similar form in which a part of the feature is different.

FIG. 14 is a diagram showing an area in a document to be recognized by a character.

FIG. 15 is a view schematically showing format information of a registration form.

FIG. 16 is a diagram showing an example of a result of classifying registered forms by color, ruled lines, and preprinted characters.

FIG. 17 is a flowchart for explaining a processing operation when selecting a feature amount used for identification according to the type of an image of a form to be identified;

FIG. 18 is a flowchart illustrating an example of an unregistered form registration processing operation.

FIG. 19 is a diagram showing an example of an unregistered form and a registered form similar thereto.

FIG. 20 is a diagram showing an example of component registration when characteristic elements of a form are converted into components.

FIG. 21 is a diagram showing an example of registration of a form when the feature amount of the form is represented using the parts shown in FIG. 20;

FIG. 22 is a flowchart illustrating a color determination processing operation.

FIG. 23 is a diagram illustrating a color determination process.

FIG. 24 is a diagram illustrating a color determination process.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 image input unit 2 form registration unit 3 input data memory 4 format information storage unit 5 form identification unit 6 display device 7 instruction device 8 storage unit

Claims (10)

[Claims] When a plurality of features are extracted from an input image of a form to be identified and one of them is compared with a feature of a registered form, the type of the form to be identified cannot be identified. And a feature amount of the registered form by comparing another feature amount of the plurality of feature amounts with a feature amount of the registered form. 2. A method for extracting at least one of a color, a ruled line, and a character string for specifying a type of a form from an input form image as a feature amount, and a plurality of registered forms common to the extracted forms. Of the forms to be identified, and among the feature quantities extracted from the input image of the form to be identified, a feature quantity used to identify the type of the form to be identified is selected. A form identification method characterized by identifying a type of a form to be identified based on a feature amount of a registered form classified by a feature amount. 3. The form identification method according to claim 2, wherein a feature amount to be extracted from the form image is selected according to the type of the input form image of the form to be identified. 4. When a form matching a feature amount extracted from an image of the form to be identified is not registered, the form is registered using a feature amount of a registered form most similar to the feature amount. 3. The form identification method according to claim 2, wherein: 5. The form identification method according to claim 2, wherein the color of each pixel is determined based on the color of peripheral pixels when extracting a color as the feature amount of the form. 6. An extracting means for extracting a plurality of feature amounts from an input image of a form to be identified, and comparing one of the plurality of feature amounts extracted by the extracting means with a feature amount of a registered form. First identifying means for identifying the type of the form to be identified; and when the type of the form to be identified cannot be identified by the first identifying means, A second identification means for comparing the characteristic amount with the characteristic amount of the registered form to identify the type of the form to be identified. 7. An extracting means for extracting at least one of a color, a ruled line, and a character string for specifying a type of a form from an input form image as a feature amount, and a plurality of registered forms are commonly used for these. Classifying means for classifying based on the feature amount to be selected; and a feature amount used when identifying the type of the form to be identified among the feature amounts extracted from the input image of the form to be identified. A form identification device, comprising: identification means for identifying the type of the form to be identified based on the characteristic amount of the registered form classified by the selected characteristic amount. 8. The form identification apparatus according to claim 7, wherein a feature amount extracted from the form image is selected according to the type of the input form image of the form to be identified. 9. When a form matching a feature amount extracted from an image of the form to be identified is not registered, the form is registered using a feature amount of a registered form most similar to the feature amount. 8. The form identification device according to claim 7, wherein: 10. An extracting means for extracting at least one of a color, a ruled line, and a character string for identifying a type of a form from an input form image as a feature amount, and a plurality of registered forms are commonly used for these. Classifying means for classifying by the feature amount to be selected; and a feature amount used for identifying the type of the form to be identified among the feature amounts extracted from the input image of the form to be identified. An identification means for identifying the type of the form to be identified based on the characteristic amount of the registered form classified by the selected characteristic amount; and a machine-readable storage medium recording a program for executing the following.