JPS6244895A - Reading method for image - Google Patents

Abstract

PURPOSE:To read out a two-dimensional image or a character image over a wide range by synthesizing two images adjacent in time out of plural reading images obtained by two or more frame scannings based on their correlation and reading out one image. CONSTITUTION:An image inputted from a two-dimensional image sensor 5 is inputted to a temporary buffer memory 12 and an image memory 14 through a sensor controller 11 and an image inputted after the passage of several frames is inputted to a buffer memory 13. The relative position data of the images stored in the buffer memories 12, 13 are obtained by a correlating circuit 15. On the basis of the obtained relative position data, the image in the buffer memory 13 is synthesized with the image of the memory 14 and the synthesized image is inputted. The image of the buffer memory 13 is inputted to the buffer memory 12 and used as a reference image for the succeeding frame.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to an image reading method for obtaining photoelectric conversion signals from characters, figures, etc. written on a manuscript by manual feeding.

[Conventional technology]

FIG. 5 is a diagram illustrating a conventional image reading method using manual feeding operation. In this figure, 1 is a document on which the image to be read is written, 2 is the characters and figures written on the document 1, and 3 is a light source for illuminating the document, which is a solid-state light source such as a tungsten lamp or LED. . 4 is a men's optical system that uses a spherical lens and a 7V focusing fiber. 5 is a two-dimensional image sensor, which is a CdS thin film photoelectric conversion element, CODM
Elements such as O8 are used. This two-dimensional image sensor 5 passes through the lens optical system 4t to display the characters and characters praised on the manuscript 1.
The figure 2 is read and moved in the direction indicated by the arrow 6 by manual feeding or the like, thereby sequentially inputting the position of each character/figure 2.

6(a), (b), and (c) show the reading state in terms of the positional relationship between the character @A'' and the two-dimensional image sensor 5. Time i, -+i depending on the moving direction.

→t, and when the state reaches time t, if the output of the two-dimensional image sensor 5 is taken out, a complete character 1A'' can be read.

This condition is detected when there are spaces on both sides of the character II A n.
It shows that it is located at the center of the dimensional image sensor 5.

[Problem that the invention attempts to solve]

However, such conventional image reading methods are effective only when reading images of a specific size, such as small characters, and are not suitable for inputting large screens containing general characters, figures, etc. There is a problem that if the image sensor 5 is small, it cannot be used, and if a two-dimensional image sensor 5 that is sufficiently large and has high resolution is made, it will be very expensive.

This invention has been made to solve the above problems, and aims to provide an image reading method that can use a small two-dimensional image sensor to obtain two-dimensional images and character images over a wide range. purpose.

[Means for solving problems]

In the image reading method according to the present invention, a two-dimensional image sensor performs reading by scanning the frame twice or more to create a plurality of read images;
between the read images by finding the correlation between temporally adjacent images or by finding the correlation between one of the plurality of read images and a composite image obtained by the read image before that read image, Or detect the relative position between the read image and the composite image.

The images are combined based on this relative position information and read as a single image.

[Effect]

In this invention, a single image is synthesized by correlating temporally adjacent images among a plurality of read images obtained by two or more frame scans with a two-dimensional image sensor. Do a read.

〔Example〕

FIG. 3 is a diagram illustrating an embodiment of the present invention.

This embodiment uses a reading optical system having the same optical system and illumination system as those used in the conventional example shown in FIG.
Illustrations are omitted. However, the two-dimensional image sensor 5
The number of pixels in one hand feed direction is smaller than the one in FIG. 5, and a fence-like structure may be used.

This figure shows an example of the position where the two-dimensional image sensor 5 reads on the surface of the original 1 by moving the two-dimensional image sensor 5, and time t is LHa*.
(a), (b), and (c) lc respectively indicate the position 1 to be read as the value changes from @+LH.

Also, at this time, the input images are each 6-1.6-2.6
-3, and the black pixel part v2-1.2-2.2
-3. At each point in time, the two-dimensional image sensor 5
Based on the image 6-1.6-2.6-3 inputted to the image 6-1.6-2.6-3, black pixels 2-1.2-2.
Image 6-1.6-2.6 using the correlation from the information in 2-3
Perform screen composition of -3.

FIG. 4 shows the synthesis method. Image 6-1
Then, a two-dimensional correlation with the image 6-2 is determined, and the position where the black pixel 2-1 matches the pixel 2-2 is determined. At this time, images 6-1 and 6-2 are combined from the obtained relative position to obtain one combined image 1-1. Next, the correlation with image 6-3 for image @6-2 is determined using the same method as described above, and at this time, based on the relative position f''k of images 6-2 and 6-3 obtained, .

Composite image 7-1 and image 6-2 are combined to create 1 composite image 7-2
get. By repeating the above process, it is possible to use the two-dimensional image sensor 5 to obtain a composite image 7-3 having a field of view greater than that obtained by the image sensor 5.

FIG. 1 is an example of a block diagram of a circuit implementing the image reading method of the present invention. In this figure, 1 to 5 are the same as in Figure 5, 11 is sensor control O-5, 12, 13 is buffer memory, 14 is image memory, 15 is correlation circuit,
16 is a memory controller. Note that solid lines indicate data movement, and dotted lines indicate control relationships.

Next, the operation will be explained.

First, the screen input from the two-dimensional image sensor 5 is sent through the sensor controller 11 to a temporary buffer 12.
and input into the image memory 14.

Next, after a few frames, the input screen is saved to buffer 7 memory 1.
Enter 3. Relative position data is obtained by the correlation circuit 15 for the images in the buffer memories 12 and 13. Based on the obtained relative position data, the image in the buffer memory 13 is combined with the image in the image memory 14 and input. The image in the buffer memory 13 is input to the buffer memory 12, and the image is input to the next image (.

and use it as the reference image. child! The correlation circuit 15 is constituted by a drossic circuit, a microcomputer, a ROM memory, and the like.

Due to the above-described operation, as the two-dimensional image sensor 5 moves in the direction of the arrow 6, the input images are synthesized one after another, and the images can be input into one reading screen (7).

Also, for the second and subsequent manual feeding operations, correlation can be established with the image of the first manual feeding, and vertical composition can be performed.

FIG. 2 is a block diagram of a circuit implementing another embodiment of the invention. In this embodiment, the 1-image memory 14 also operates as a buffer memory, and the correlation circuit 15
4 and the image in the backup memory 13 are combined to form a composite image while the contents of the image memory 14 are successively rewritten.

[Effect of light]

As explained above, the present invention relates to a method for capturing an image using a hand-feeding two-dimensional image sensor.

A plurality of read images are created by two or more frame scans, and the correlation between temporally adjacent images is determined, or one of the plurality of images and the read image are (2) Detect the relative position between the read images or between the read image and the combined image by finding a correlation with the composite image obtained from the read images, and then combine the images based on this to read a single image. As a result, the image input range can be made wider than the field of view, and an image without distortion can be obtained.

Furthermore, as an image reading method in a hand scanner device that inputs only one image, one advantage is that the two-dimensional image sensor part can be made smaller and cheaper, and the head part can be made smaller and the size of the head part can be reduced, such as K/% Ndos uniformity.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a circuit that implements an embodiment of this invention, FIG. 2 is a block diagram of a circuit that implements another embodiment of this invention, and FIG. 3 is for explaining one embodiment of this invention. FIG. 4 is a diagram showing the composition of read images for explaining one embodiment of the present invention, and FIG. 5 is a diagram showing the conventional reading and the optical system used in this embodiment. FIG. 6 is a perspective view showing an overview of the system, and is a diagram showing a reading position 1k of a two-dimensional image sensor for explaining a conventional reading method. In the figure, 1 is a document, 2 is a character/figure, 5 is a two-dimensional image sensor, 11 is a sensor controller, 12.13
is the buffer memory, and 14 is the image memory.

Claims (1)

[Claims] In an image reading method that reads characters and figures drawn on a document by hand-feeding, the characters and figures are read by two or more frame scans using a two-dimensional image sensor to create a plurality of read images. to find a correlation between temporally adjacent images among the read images, or to find a correlation between one of the plurality of read images and a composite image obtained by a read image before that read image. hand,
An image reading method characterized by detecting relative positions between read images or between a read image and a composite image, and based on this relative position information, images are combined and read as one image.