JP2824372B2 - Report recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention cuts out a recognition target such as a character, a symbol, or a mark written in a predetermined portion from image data read by an image scanner or image data transmitted by a facsimile machine. The present invention relates to a form used for processing and inputting necessary information to a computer, and a recognition device for the form.

[0002]

2. Description of the Related Art Conventionally, an OCR (optical character reader) has been known as a device for inputting necessary data to a computer. In OCR, a recognition target such as a character, a symbol, or a mark written in a predetermined position of a form is recognized by character recognition processing or position detection processing, and is converted into predetermined data that can be input to a computer. In recent years, in addition to directly inputting a form to an OCR, a system that uses a facsimile apparatus or the like to cut out a recognition target from image data of a form transmitted from a remote place and perform a recognition process has been used. Was.

[0003] By the way, O using a facsimile machine
In a CR system, a transmitted form image has vertical and horizontal displacements, inclinations, distortions, and the like due to general characteristics of the facsimile machine. Therefore, conventionally, as shown in FIG. 10, fiducial marks 3 are provided on both sides of each read line 2 of the form 1 and the coordinates of each fiducial mark 3 in the transmitted image data 4 are detected. The position of the read line is correctly detected by correcting the positional deviation (for example, JP-A-3-189783).

[0004]

However, in the above-described conventional form reading system, when the form is inserted in a facsimile apparatus and the insertion direction of the form is different from the expected direction, the position of the recognition target is correctly set. There is a problem that it cannot be detected. In particular, as the spread of OCR advances and the number of forms to be handled increases, and the number of form senders using facsimile machines becomes unspecified, the number of cases where forms are not sent in a predetermined direction is expected to increase more and more. Is done. Further, even in a system in which a form is directly read by an OCR device, the same problem occurs when the number of handled forms increases.

The present invention has been made in order to solve such a problem, and it is possible to always correctly cut out a recognition target even if the form is read in different directions, and to execute a correct recognition process. It is an object of the present invention to provide a form and a recognition device thereof. Another object of the present invention is to provide a form capable of detecting a position of a recognition target in a form in real time and executing a recognition process at a high speed, and a recognition device therefor.

[0006]

[0007]

[0008]

According to the present invention, there is provided a form reading apparatus comprising: a reference mark provided at each of four corners of a form which can identify at least a reading direction from image data obtained by reading the form by raster scanning; Fiducial mark detecting means for detecting only a fiducial mark arranged near a side to be scanned earlier of the form to identify at least a reading direction of the form, and reading a form identified from image data of the form Cut-out processing means for cutting out image data to be recognized from the image data of the form based on direction information, and a recognition processing unit for recognizing the recognition target from the image data to be recognized cut out by the cut-out processing means Wherein the cut-out processing means is
Without storing the image data of the ledger,
From the image data of the vote, the image data of the recognition target
It is characterized by cutting out .

[0009]

[0010]

According to the form of the present invention, since reference marks at least at the four corners for identifying the reading direction are provided, the orientation of the form can be detected based on the reference marks.
The position of the recognition target can be cut out correctly.

In particular, 2 which is located near each side of the form
If the combination pattern of one reference mark is different from the combination pattern of the two reference marks located near any of the other sides, the preceding two reference marks are extracted from the image data obtained by raster scanning. It is possible to identify the form reading direction only by detecting. For this reason, the orientation of the form can be known only from the image data at the upper end of the form, the position of the recognition target can be correctly identified in real time, and the recognition target can be cut out at high speed.

Further, according to the form reading apparatus of the present invention, the reference mark detecting means detects the reference mark from the image data of the form and identifies the reading direction of the form, and the cut-out processing means performs the processing of the form. Since the form recognition target is cut out based on the reading direction information, the position of the form recognition target can always be correctly identified regardless of the form reading direction, that is, the form orientation.

In this case, similarly to the above, the form reading direction is detected by the two previously inputted fiducial marks among the fiducial marks provided at the four corners of the form, so that the input image data is detected. It becomes possible to cut out image data to be recognized in real time. Therefore, the image data can be cut out at high speed, and a memory for temporarily storing the entire image data is not required.

[0014]

Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 shows an OC according to an embodiment of the present invention.
It is a figure showing the report for R. The form 11 is provided with fiducial marks 12, 13, 14, and 15 at four corners, and a recognition target 16 such as a character, a symbol, or a mark is arranged at a predetermined position with respect to the fiducial marks 12 to 15. The reference marks 12 to 15 are formed of, for example, black circles whose constant width is always detected even when the form 11 is read with inclination, and among the reference marks 12 to 15, the reference mark 12 located near the upper side is used. , 13 have a larger diameter than the reference marks 14 and 15 located near the lower side.

When such a form 11 is transmitted by, for example, a facsimile machine, it is expected that the form 11 will be read in four reading directions d1, d2, d3 and d4 in the figure. In this case, the relationship between the two reference marks transmitted in advance and the reading directions d1 to d4 is as shown in Table 1 below.

[0016]

[Table 1]

Therefore, by examining the magnitude relationship between the two reference marks 12 to 15 transmitted in advance, the orientation of the form 11 can be identified by relatively early data of the image data.

FIG. 2 is a functional block diagram showing a configuration of such a form 11 reading device. The image data D of the form 11 read by the image scanner or the image data D of the form 11 transmitted via the facsimile machine is first input to the reference mark detection unit 21. The fiducial mark detecting unit 21 detects two preceding fiducial marks among the fiducial marks 12 to 15 described above from the image data, identifies a reading direction di of the form 11 from a combination pattern of the two fiducial marks, From the coordinates of the two fiducial marks, the positional deviation X1, X2 of the form 11 and the rotation angle θ
And the expansion / contraction ratio S are detected. These pieces of detection information are supplied to the cutout processing unit 22 together with the image data D. The cutout processing section 22 cuts out the image data Da of the recognition target 16 from the image data D based on the detection information. The cut-out image data Da of the recognition target 16 is supplied to the recognition processing unit 23 and recognized. When the recognition target 16 is a character or a symbol, the image data Da is converted into its code information Dc, and the recognition target 16 stores the information in the position. In the case of a possessed mark, it is converted into mark position information.

FIG. 3 is a functional block diagram of the reference mark detecting section 21. The image data D is first supplied to the upper right area reference mark detection unit 31 and the upper left area reference mark detection unit 32. The detection units 31 and 32 cut out the image data of the upper right quarter and the upper left quarter by, for example, a predetermined mask process to detect the reference mark pattern. The detected reference marks are sent to the size discriminating units 33 and 34.
Is supplied to each of them to determine the size. Note that the expansion / contraction ratio S is also involved in the size determination of the reference mark, so the expansion / contraction ratio S may be used for the size determination. The information of the magnitude relation is given to the document direction detection unit 35, and the reading direction di is identified. Information on the detected fiducial marks is supplied to the coordinate detectors 36 and 37, respectively, and the center coordinates of the upper right and upper left fiducial marks are obtained.

As shown in FIG. 4, the coordinates of an arbitrary point in the coordinate system on the form 11 are (x, y), and the coordinates of an arbitrary point in the coordinate system on the image data D are [X,
Y], and the coordinates of the four reference marks 12 to 15 of the form 11 are (0, 0), (x1, 0), (0, y1), respectively.
), (X1, y1), the origin on the image data D is [0, 0], the coordinates of the upper left reference mark on the image data D are [X1, Y1], and the upper right reference mark on the image data D. Is [X2, Y2], the displacement of the form 11 is X1, Y1, the rotation angle θ and the expansion / contraction ratio S
Is determined by the following equation (1).

[0021]

[Number 1] θ = tan-1 [(Y1 -Y2) / (X2 -X1)] [ For reading direction d1, d3] S = √ [ (X1 -X2) 2 + (Y1 -Y2) 2] / x1 [for reading direction d2, d4] S = √ [ (X1 -X2) 2 + (Y1 -Y2) 2] / y1

The rotation angle θ and expansion / contraction ratio S are calculated by the above-mentioned operation by the rotation angle / expansion detection unit 38.

As shown in FIG. 5, when the reading direction of the form 11 is d1, d2, d3, d4, the coordinates [X, Y] on the image data D are expressed by the following equation (2). Is represented by

[0024]

(1) Reading direction d1 X = Sx / cosθ + (Y−Y1) tanθ + X1 Y = Sy / cosθ− (X−X1) tanθ + Y1 (2) Reading direction d2 X = Sy / cosθ + (Y−Y1) tan θ + X1 Y = S · (x1−x) / cos θ− (X−X1) tanθ +
Y1 (3) Reading direction d3 X = S ・ (x1−x) / cos θ + (Y−Y1) tan θ +
X1 Y = S ・ (y1−y) / cos θ− (X−X1) tan θ +
Y1 (4) reading direction d4 X = S ・ (y1−y) / cos θ + (Y−Y1) tan θ +
X1 Y = S ・ x / cos θ- (X-X1) tan θ + Y1

Therefore, the coordinates [X, Y] on the image data D are obtained on the form 11 as shown in the following Expression 3.

[0026]

(1) Reading direction d1 x = [(XX1) cosθ- (YY1) sinθ] / Sy = [(Y−Y1) cosθ + (XX1) sinθ] / S (2) Reading direction d2 x = x1 − [(Y−Y1) cos θ + (X−X1) sin
θ] / Sy = [(X−X1) cos θ− (Y−Y1) sin θ] / S (3) Reading direction d3 x = x1 − [(X−X1) cosθ− (Y−Y1) sin
θ] / S y = y 1 − [(Y−Y 1) cos θ + (X−X 1) sin
θ] / S (4) Reading direction d4 x = [(Y−Y1) cos θ + (X−X1) sin θ] / Sy = y1 − [(X−X1) cos θ− (Y−Y1) sin
θ] / S

From these facts, it is clear that: (1) the coordinates [X1, Y1 of the two preceding fiducial marks;
], [X2, Y2], the arbitrary coordinates (x, y) on the form 11 can be converted to the coordinates [X, Y] on the image data D. (2) The coordinates [X, Y] on the image data D are the coordinates [X1, Y1], [X2, Y] of the two preceding reference marks.
2] can be converted to the coordinates (x, y) on the form 11 when [2] is known. That is the point.

Based on the above two points, the cut-out processing section 22 executes the cut-out processing of the recognition target in real time.
The cutout processing unit 22 is configured, for example, as shown in FIG. The cutout area detection unit 41 replaces the area to be recognized on the form 11 with cutout area information represented by coordinate axes on the image data D based on the detection information di, θ, S, X1, and Y1. More specifically, information is generated by converting the region 17 to be recognized indicated by oblique lines in FIG. 7 into an [X, Y] coordinate system and stored. On the other hand, the clock signal CK synchronized with the image data D is input to the address counter 42, and the coordinates [X, Y] of the data are generated in synchronization with the image data D sequentially input by the raster scan. The coordinates [X, Y] are determined by the cut-out area detection unit 4.
1 is supplied. The cut-out area detecting unit 41 determines whether or not the coordinates [X, Y] are within the stored cut-out area.
3 and the cut-out information memory 44.

The coordinates [X, Y] from the address counter 42 are also supplied to a coordinate conversion unit 43. The coordinate conversion unit 43 receives the write command W from the cutout area detection unit 41.
When E is given, the coordinates [X,
Y] is converted into coordinates (x, y) on the form 11 based on the detection information di, θ, S, X1, Y1. Then, the offset xo, yo from the form origin of the recognition target area 17 shown in FIG. 7 is subtracted from the transformed coordinates (x, y), and this is supplied to the cutout information memory 44 as the address AD. The extraction information memory 44 stores the image data D at a given address. When this is sequentially repeated, as shown in FIG. 8, when the raster scan of the cutout area is completed, the image data D of the area to be recognized is stored in the cutout information memory 44 at a regular position and size. Will be. Thereafter, in the recognition processing unit 23, the form 11
, Standardized data that is completely unaffected by the reading direction, the rotation angle, the displacement, and the degree of expansion / contraction can be recognized.

As described above, according to the apparatus of this embodiment, image data sequentially input by raster scanning can be processed in real time, and all image data are stored in the apparatus until the cut-out processing is completed. It is not necessary to save the memory capacity, and the memory capacity can be saved and high-speed processing can be performed.

In the above embodiment, the large and small circles are used as the reference marks. However, the form of the reference marks is, for example, as shown in FIG.
(B) A round reference mark 52 having all different diameters, (c)
Various forms such as a reference mark 53 using a circle and a double circle can be used. In each of these reference marks 51 to 53, the reading direction of the form 11 can be identified only by the preceding two reference marks.

[0032]

As described above, according to the present invention, at least four reference marks which can identify the reading direction are provided at the four corners of a form, and the orientation of the form can be detected based on the reference marks. Even if the reading directions are different, the recognition target can always be cut out correctly, and correct recognition processing can be executed.

[Brief description of the drawings]

FIG. 1 is a diagram showing a form according to an embodiment of the present invention.

FIG. 2 is a functional block diagram of a recognition device that recognizes the form;

FIG. 3 is a functional block diagram of a reference mark detection unit in the device.

FIG. 4 is a diagram for explaining a reference mark detection process in the apparatus.

FIG. 5 is a diagram for explaining a coordinate conversion process in the device.

FIG. 6 is a functional block diagram of a cut-out processing unit in the same device.

FIG. 7 is a diagram for explaining a cut-out process in the apparatus.

FIG. 8 is a diagram for explaining a cutout process in the same device.

FIG. 9 is a diagram for explaining a cut-out process in the same device.

FIG. 10 is a diagram for explaining a conventional form displacement correction method.

[Description of Signs] 1,11: Form, 3, 12 to 15, 51 to 53: Reference mark, 4, D: Image data, 16: Recognition target, 21
... Reference mark detector, 22... Cut-out processor, 23... Recognition processor, 31... Upper right area reference mark detector, 32.
... Document direction detectors, 36 and 37 coordinate detectors, 38 rotation angle / expansion detectors, 41 cutout area detectors, 42.
Address counter, 43: coordinate conversion unit, 44: cut-out information memory.

Claims (1)

(57) [Claims] 1. A reference mark provided at each of four corners of a form, which can be identified at least in a reading direction from image data obtained by reading the form by raster scanning, and is arranged near a previously scanned side of the form. Reference mark detecting means for detecting at least the reference mark identified and identifying at least the reading direction of the form; and a recognition target from the image data of the form based on information on the reading direction of the form identified from the image data of the form. And a recognition processing unit for recognizing the recognition target from the image data of the recognition target cut out by the extraction processing means, wherein the cut-out processing means includes: Image data
The recognition from the image data of the form without storing
A form recognition apparatus for extracting image data to be recognized.