JPH0423184A - Optical character reader - Google Patents

Abstract

PURPOSE:To read a colored general document by invalidating a picture image sorted into the color of a picture part designated by position information obtained by a designating means, and binarizing successively the picture image of a remaining color in a labeling part. CONSTITUTION:A scanning part 1 reads document as a color picture image, and a sheet buffer 2 holds the read color picture image. Then, a display part 3 displays the color picture image in the buffer 2. The labeling part 5 sorts the color picture image into the picture of every color, and holds the picture image of every color, and a binarizing part 6 invalidates the picture image sorted into the color of the picture part designated by the position information obtained by an operating part 4, and binarizes successively the picture image of the remaining color in the labeling part 5. A character recognizing part 7 executes successively the character recognition of the picture image binarized by the binarizing part 6. Thus, the multi-color-printed general document can be read.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Field of Industrial Application) The present invention relates to an optical character reading device for reading color images.

(Prior art) Conventional optical character reading system FI! (hereinafter referred to as OCR)
The black-and-white image is read as a black-and-white image, with black as valid and white as invalid. For this reason, in the scanning section that scans the OCR image and performs photoelectric conversion, the obtained image data is binarized into either white or black data by distinguishing the obtained image data using a certain threshold value.

In addition, by using a color that appears white to the OCR scanning unit (called dropout) as the color used for the handwritten character frame of the document to be read, the character frame is not read and the text is not read within the character frame. There were times when I tried to read only sloppy handwriting. However, the dropout color is determined by the wavelength of the light source used in the scanning section.

When you want to read a general document printed in multiple colors using various colors, such as a magazine, using conventional OCR, if the document has text written in the color that drops out, the text will be lost when it is binarized. There were problems in that the text would drop out and be unreadable, or the background and text would both become effective and they could not be separated during binarization, resulting in the text being unreadable. Therefore, conventional OCR has the disadvantage that it cannot read general documents that are colored.

(Problems to be Solved by the Invention) As mentioned above, since conventional OCR scans black and white images, it has the disadvantage that it cannot read general documents printed in multiple colors.

SUMMARY OF THE INVENTION The present invention aims to eliminate the above-mentioned drawbacks and provides an optical character reading device capable of reading general documents printed in multiple colors.

[Structure of the invention]

(Means for Solving the Problems) The optical character reading device of the present invention includes a reading means for reading a form as a color image, a holding means for holding the color image read by the reading means, and a holding means for holding the color image read by the reading means. display means for displaying a color image in the means; designation means for specifying positional information of any one point of the color image displayed on the display means; a labeling means for classifying into images and holding an image for each color; and a labeling means for disabling the image classified into the color of the image portion specified by the position information obtained by the specifying means, and disabling the remaining color in the labeling section. The present invention has a configuration including a binarization means for sequentially binarizing the image images, and a character recognition means for sequentially performing character recognition of the image images binarized by the binarization means.

(Operation) In the optical character reading device of the present invention, the reading means reads the form as a color image.

The holding means holds the color image read by the reading means. Display means displays the color image image within said holding means. The specifying means specifies positional information of any one point in the color image displayed on the display means. The labeling means classifies the color images in the holding means into images for each color and holds the image for each color. The binarization means invalidates the image classified into the color of the image portion designated by the position information obtained by the designation means, and sequentially binarizes the remaining color images in the labeling section.

The character recognition means sequentially performs character recognition of the image image binarized by the binarization means.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the optical character reading device of the present invention. 1 is a scanning unit that scans an input multicolor printing form to obtain a color image; 2 is a sheet buffer that stores color image data obtained from scanning unit 1 for one form; 3 is stored in sheet buffer 2. A display section displays the input color image in color; 4 is an operation section for specifying any one point on the display section 3 and inputting it to the control section 1; 5 is a display section for each color of the color image stored in the sheet buffer 2; A labeling section 6 stores an image for each label after labeling the connectivity of the pixels as a group; 6 is a labeling section that stores an image for each label; 5 is a binarization unit that performs binarization by invalidating the image classified by the label of the same color as the color at the given position, and 7 is a binarization unit output from the binarization unit 6. A character recognition unit 8 that recognizes characters from the image pattern outputs the color image stored in the sheet buffer 2 to the display unit 3 for display, and also outputs the position information from the operation unit 4 at least twice or more. This is a control unit that inputs the position information and provides it to the binarization unit 6 to control character recognition processing.

Here, the sheet buffer 2 serves as a holding means, the display section 3 serves as a display means, the operation section 4 and the control section 8 serve as specifying means, the labeling section 5 serves as a labeling means, and the binarization section 6 serves as a binarization means. , the character recognition section 7 constitutes character recognition means.

Next, the operation of this embodiment will be explained. First, the document to be read is photoelectrically converted by the scanning unit 1 to become color image data (input color image), and this color image data is stored in the sheet buffer 2 . The control unit 8 reads out the color image data stored in the sheet buffer 2 and supplies it to the display unit 3, thereby causing the display unit 3 to display the color image stored in the sheet buffer 2.

At the same time, the labeling unit 5 examines the connectivity of pixels for each color of the color image stored in the sheet buffer 2, labels each set of connected pixels, and labels each image with this labeling. Performs storage operation for each label.

That is, for example, when a color image consisting of red, blue, and yellow parts is stored in the sheet buffer 2, the labeling unit 3 labels each of red, blue, and yellow (in this example, three labels are attached). , the image of the red part, the image of the blue part, and the image of the yellow part are separated and stored.

On the other hand, the operator looks at the color image displayed on the display section 3, uses a mouse, cursor, etc., to specify, for example, a point in the colored part that will become the background of the characters, and notifies the control section 8 of this positional information. For example, if the background has multiple colors, one point on the red background and one point on the blue background are designated, for example. Thereby, the operation unit 4 outputs to the control unit 8 the position information of one point on the red background and one point on the blue background specified by the operator. Control unit 8
provides the thus input position information to the binarization section 6. As described above, the binarization unit 6 invalidates the label information corresponding to the color of the position information part given from the control unit 8 (for example, makes the label number negative), and validates only the image of the remaining label. and convert it into a binary value.

Here, set the red label number to (1) and the blue label number to (1).
2), if the yellow label number is (3), the binarization unit 6
In the above example, the red and blue label numbers are made negative and invalid, so only the yellow label number is valid. Therefore,
The binarization unit 6 converts the labels stored in the labeling unit 5 (
Binarization is performed only on the yellow image formed by the connection of yellow pixels in step 3), and the resulting binarized pattern is sent to the character recognition section 7. The character recognition section 7 performs character recognition on the binarized pattern sent from the binarization section 6, and notifies the control section 8 of the result. Here, if there are multiple label numbers to be binarized by the binarization unit 6, the binarization unit 6 binarizes each label and sends the binarized pattern to the character recognition unit 7. , the character recognition unit 7 performs character recognition processing multiple times for each image corresponding to a valid label.

According to this embodiment, when reading general documents such as magazines printed in multiple colors, the read images are separated and stored for each color, images of colors that you do not want to read are specified, and these images are removed. Since it is possible to perform character recognition by binarizing only the remaining image, dropouts do not occur and both the background and characters are prevented from being valid.
Color text can be read accurately.

〔Effect of the invention〕

As described above, according to the optical character reading device of the present invention, general documents printed in multiple colors can be read.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the optical character reading device of the present invention. 1...Scanning section...Sheet buffer 3...
・Display section 4...Operation section 5...Labeling section 6...Binarization section 7...Character recognition section 8.
...Control Department Agent Patent Attorney Hon 1) Takashi

Claims (1)

[Claims] a reading means for reading the form as a color image;
A holding means for holding the color image read by the reading means, a display means for displaying the color image in the holding means, and a display means for displaying the color surface image displayed on the display means. a specifying means for specifying positional information; a labeling means for classifying the color images in the holding means into images for each color and holding the image for each color; binarization means for invalidating the image classified into the color of the image portion and sequentially binarizing the remaining color images in the labeling section; What is claimed is: 1. An optical character reading device comprising character recognition means for sequentially performing character recognition.