JP2714003B2 - Address area detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention] (Industrial application field) The present invention can efficiently detect an address entry area in a letter mail written in various formats and use it for address reading. The present invention relates to a destination area detection device.

(Prior Art) The format (form) of a form provided for character reading by an optical character reading device (OCR) is generally defined, and the character reading device uses character information described in a predetermined area of the form. Is configured to be read only. Therefore, conventionally, a form in which a character entry area is specified by a frame or the like exclusively according to a predetermined format is prepared.

In the case of a character reading device for reading a form in various formats, a form in which an ID number or an ID code for specifying the format is preprinted at a predetermined position is prepared. Is read, the format is known, and the area to be read is set.

However, for example, in a character reading apparatus for reading the address information of a letter mail, the format (format) of the address in the letter mail is not fixed, so the area to be read is detected in advance. It is necessary. In addition, on the surface where the address information of the letter mail is written, there is a communication message, advertising message, sender's address name, stamp, postmark, postal code, etc., and the address writing area is detected without misjudging these information. It is necessary to do. However, there is a problem in that a great deal of processing load is involved in detecting the address description area separately from these pieces of information, and the processing efficiency is extremely poor.

(Problems to be Solved by the Invention) As described above, in the conventional character reading device, the address information of the letter / mail having an undefined format is read without referring to the ID information for specifying the format. There is a problem that it is difficult to target the area, and it takes a large processing load to detect the address description area in the letter mail and distinguish it from other information to specify the area to be read.

The present invention has been made in view of such circumstances,
The purpose is to easily and accurately detect the address description area of a letter / mail with an undefined format without using ID information etc. to specify the format, and to read and input the address information. An object of the present invention is to provide a destination area detecting device that can be provided.

[Configuration of the Invention] (Means for Solving the Problems) A destination area detecting apparatus according to the present invention inputs image information of a surface on which the destination information of a letter mail is described, and reads the destination description area from the input image. Means for detecting an input image and dividing the input image into predetermined unit block areas to obtain image feature information in each unit block area. 2 in which each unit block area is binarized and expressed as one unit according to the feature information of the image
Means for determining a value image, means for determining a projecting feature of the block binary image, and a charge stamp / marker in the input image based on the projecting feature and knowledge information on the letter mail.
Means for detecting the postal code section and the advertisement / communication sentence respectively; means for removing the charge stamp face / postal code field and the advertisement / communication sentence from the input image; and Means for detecting a destination addressing area in the input image from the projection feature.

(Operation) The present invention divides the captured input image of the letter mail into predetermined unit block areas, captures each unit block area as one unit, and captures the block as a block binary image expressed in binary according to the image feature information. According to the projection characteristics of the binary image and the knowledge of the letter / mail, the charge stamp / postal code section and the advertisement / communication sentence in the input image are respectively detected. Then, these detection areas are removed from the input image, and the address description area is detected in the remaining area according to the projection of the block binary image.

Therefore, block 2 merely indicates whether each unit block area plays an important role as an information recording area.
It becomes possible to globally and efficiently identify various information describing areas of the letter mail from the value image, and to effectively determine the type of the information describing area according to knowledge information on the letter mail. Then, the information description area in which the type of the information is clear is removed, and the information description area in the remaining area is detected as the address description area. It is possible to easily detect the address description area to be entered.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of an apparatus according to an embodiment, and reference numeral 1 denotes a letter mail to be read by the apparatus. An image input unit 2 composed of an image scanner or a TV camera is for inputting image information of a surface on which address information and the like of the letter mail 1 are described. For example, the image input image is binarized by a predetermined threshold value. The image data is stored as a binary digital image in an image memory and subjected to a character reading process (destination region detecting process) described below.

The feature creating unit 3 performs the input image (binary digital image)
Is divided into predetermined unit block areas, and the feature information of the entire image is detected and created. That is, the feature creator 3a measures the appearance frequency of black pixels in the block area for each unit block area, and obtains the measured value as feature information of the unit block area. A black pixel frequency in each unit block area is compared with a predetermined threshold value, and a block is formed by binary representation of whether the unit block area is entirely represented by a black pixel component or a white pixel component. It comprises a block binary image creation unit 3b for obtaining a binary image, and a projection feature creation unit 3c for obtaining projections of the block binary image in the X and Y directions as feature information.

In other words, the feature creating unit 3 assigns each unit block of the input image divided into the predetermined unit block area to “1” or “0” according to the frequency of occurrence of black pixels in the block area.
Are obtained, and the projected components in the horizontal direction (X direction) and the vertical direction (Y direction) of the block binary image are obtained. Note that the detection of the projected component is appropriately performed according to the result of the area determination described later.

On the other hand, the knowledge dictionary 4 stores general knowledge of information described in the letter mail 1. This knowledge is
For example, the postal code entry field in the letter mail 1 is usually located at the upper right of the mail, the area of the postage stamp face where the stamp is attached is the upper left corner of the mail, The lower area is composed of information such as a message or an advertisement sentence.

The address area detection processing of the letter mail 1 in the present apparatus is as follows.
This is performed as follows based on the above-described block binary image, its projection information, and the knowledge information stored in the knowledge dictionary 4.

The fee stamp / postal code column determination unit 5 and the advertisement / correspondence / sender name determination unit 6 determine the input image based on the block binary image and its projection information and the knowledge information stored in the knowledge dictionary 4. It detects a unit block area of data "1" at the top and bottom as a charge stamp face / zip code field area and an advertisement text / message text / sender area, respectively. Respectively removes these detection areas from the input image. The region detection and the process of removing the detected region from the input image may be performed step by step, or may be performed in parallel.

As described above, the address area determination unit 9 projects the block binary image in the area where the input image obtained by removing the charge amount stamp face / postal code field area and the advertising text / correspondence / sender name area is left. According to the components, a unit of the unit block area of the data “1” is detected as the address description area.

That is, here, the block 2 obtained as described above is used.
According to the projection characteristics of the value image, the address description area is determined from the global viewpoint of whether or not the unit block area is an information description area and how the unit block area of data "1" is organized. The other areas are removed, and the address description area to be detected is simply and efficiently detected.

The above-described addressing area detection processing will be specifically described. Information on the addressing side of the letter mail 1 to be processed is given, for example, as shown in FIG. Such information is captured and input by the image input unit 2 as a binary digital image.

Thus, the black pixel frequency feature creator in the feature creator 3
In step 3a, the input image as shown in FIG. 2 is divided vertically and horizontally into predetermined unit block areas. This division is performed, for example, by dividing an input image of 400 pixels in the horizontal (horizontal) direction and 800 pixels in the vertical (vertical) direction into 16 in each of the vertical and horizontal directions. By this division, a (16 × 16) block in which a (25 × 50) pixel area is a unit block area is set.

Then, the black pixel frequency feature creation unit 3a measures the appearance frequency of the black pixel in each of these unit block areas, and obtains the measurement result as shown in FIG. The block binary image creation unit 3b discriminates the black pixel appearance frequency of each unit block area shown in FIG.
The unit block area is binarized into information of “1” and “0” indicating whether or not the unit block area is a description area of information when viewed globally. As a result, a block binary image as shown in FIG. 4 is created, and the input image is expressed as (16 × 16) pixel data.

Projection feature creation unit for such a block binary image
In 3c, the horizontal and vertical projection components in the arbitrary rectangular area are obtained as feature information of the block binary image.

FIG. 5 shows the projection characteristics of such a block binary image,
It is a figure which shows typically the concept of the area | region detection process performed using the knowledge information regarding letter mail.

That is, the fee stamp / postal code column determining unit 7 determines the data “upstream portion of the block binary image” according to the foresight knowledge that the fee stamp / postal code column in the letter mail 1 exists in the information portion. A unit block area of "1" is detected and all the areas above the detected area are set to the fifth area.
As shown in FIG. 7A, the charge amount is detected as the charge stamp / postal code field area α. The determination area deletion unit 7 deletes the area detected as the charge stamp / postal code field area α from the block binary image, and sets the data of the detected unit block area to “0”.

From the block binary image from which the area detected as the fee stamp face / postal code field area α is deleted as described above, the advertisement text / correspondence / sender name etc. According to the a priori knowledge that the information is written in the lower area of the letter mail, the horizontal projection feature is examined from the lower end side as shown in FIG. 5 (b), and the advertisement text / correspondence / sender name area β Is detected. The advertisement text / correspondence / sender name area β thus detected is also deleted from the block binary image by the determination area deletion unit 8.

As described above, the information recording areas α and β other than the address area
Is deleted from the block binary image, the address area determination unit 9 performs a process of detecting the address description area γ. In the detection processing of the address description area γ, the horizontal and vertical projection characteristics of the block binary image from which the information recording areas α and β other than the address area are deleted as described above are obtained, and the data “1” is obtained. Is obtained by obtaining a rectangular area formed by the unit block areas as the address writing area γ.

As a result, the address description area γ of the letter mail is
From the 6th to 13th in the X direction and from the 3rd in the Y direction of the value image
It will be determined as the twelfth rectangular area.

As described above, according to the present apparatus, the information amount of the input image is treated as a greatly compressed block binary image, and the address information area γ is detected from a global viewpoint by obtaining the characteristic information. It is possible to detect the address area of the letter mail at high speed. Moreover, it is possible to accurately detect only the address description area from the input image of the letter mail containing various information other than the address information. As a result, it is possible to achieve a great effect in practical use, for example, it is possible to improve the reliability in the reading and input processing of the address information.

Note that the present invention is not limited to the above-described embodiment. Here, the input image was converted to a (16 × 16) block binary image and the address detection processing was performed. The number of block divisions and the threshold for obtaining the block binary image are determined by the letter mail to be processed. What is necessary is just to determine according to the specification of an object. In addition, here, the characteristic information is obtained from the frequency of appearance of black pixels in the unit block area. However, the frequency of appearance of white pixels can be naturally obtained as the characteristic information. Alternatively, it is also possible to block the multi-valued input image as it is and to detect the feature by obtaining the density histogram. Others
The knowledge information used for detecting various regions is not limited to the above-described example. In short, the present invention can be variously modified and implemented without departing from the gist thereof.

[Effects of the Invention] As described above, according to the present invention, a captured binary image of a letter mail is converted into a block binary image based on a predetermined unit block area, and this information amount compressed block binary image is converted. Since the address description area is detected from the feature information of the image,
It is possible to simply and efficiently detect the address description area in the letter mail accurately, and to achieve a great effect in practical use, such as improvement in the reliability of the address information reading and input processing.

[Brief description of the drawings]

FIG. 1 shows an embodiment of a destination area detecting apparatus according to the present invention. FIG. 1 is a schematic configuration diagram of the embodiment apparatus, and FIG. 2 is a format example of a letter mail and the concept of blocking the input image. FIG. 3 is a diagram showing an example of information on the appearance frequency of black pixels in a unit block region, FIG. 4 is a diagram showing an example of a block binary image, and FIG. 5 is a region detection process from a block binary image. It is a figure which shows typically. 1 ... letter mail (to be read), 2 ... image input unit,
3 ... Character creation unit, 3a ... Black pixel frequency feature creation unit, 3b ...
... Block binary image creation unit, 3c ... Projection feature creation unit, 4
…… Knowledge dictionary, 5… Charge amount stamp / postal code field judgment part, 6
…… advertising text / correspondence / sender name judging unit 7,8 …… judgment area deletion unit 9 …… address area judging unit

Claims (1)

(57) [Claims] An address area detecting device for inputting information of a face on which address information of a letter mail is described, detecting an address description area from the input image, and performing character reading, wherein the input image is input to a predetermined area. Means for obtaining image feature information in each unit block area by dividing each unit block area, and binarizing each unit block area as one unit in accordance with the image feature information obtained for each unit block area Means for obtaining the block binary image thus obtained, means for obtaining the projection feature of the block binary image, and a charge stamp / postal code field in the input image based on the projection feature and the knowledge information on the letter mail. Department and advertising
Means for respectively detecting the message portion, means for removing the charge stamp face / postal code column portion and the advertisement / message portion from the input image, and the projection feature in the area removed by the removal means in the input image. Means for detecting an address entry area.