JPH08106510A - Character reader and character recognition device - Google Patents

Abstract

PURPOSE: To always quickly execute the character recognition processing with a high recognition rate independently of the size of characters as the read object. CONSTITUTION: An image pickup part which is provided with a zoom lens 24 having plural focal lengths and an area CCD 21 to pick up the image of a picture as the object and a DSP 11 which makes the image pickup part pick up the image again after variably setting the focal length of the zoom lens 24 in accordance with the character string picture in a specific position of the image obtained by the image pickup part are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character reading device and a character recognition device for picking up an image having a character and recognizing the character in the image.

[0002]

2. Description of the Related Art Conventionally, a character string in a captured image is extracted and character recognition processing is performed to obtain character information.
CR (optical character reader) is widely used. In this type of device, the range in which the image is captured is constant, and the character is cut out and the cut out character is recognized regardless of the size of the character in the captured image.

[0003]

As described above, since characters are read from an image captured within a certain range, the range to be captured is set in advance assuming a large character in the image. However, there is a problem in that the memory for image storage is wasted by taking in unnecessary portions and the operation processing time becomes long. Further, particularly when the characters in the image are small, the resolution is lowered, and reading defects called "blurring" and "crushing" are likely to occur, which lowers the recognition rate.

The present invention has been made in view of the above situation, and an object of the present invention is to always perform a rapid character recognition process with a high recognition rate regardless of the size of a character to be read. It is to provide a character reading device and a character recognition device capable of performing the above.

[0005]

That is, the present invention relates to an image pickup unit having a lens having a plurality of focal lengths, for example, a zoom lens, and a solid-state image pickup device, and for picking up an image of interest. The focal length of the zoom lens is variably set in correspondence with the character string image at the specific position in the image obtained in step 1, and the image is captured again.

[0006]

With the above-mentioned configuration, it is possible to always perform the character recognition processing quickly with a high recognition rate corresponding to the size of the character to be read.

[0007]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of the functional circuit. Reference numeral 11 denotes a DSP (Digital Signal) for controlling the entire operation.
al Processor). This DSP11
Control of each circuit connected to the system bus 14 is executed by sequentially reading operation instructions from the instruction ROM 13 in accordance with the operation clock supplied from the pulse generator (PG) 12.

The system bus 14 has a work RAM 15 for temporarily storing and holding various data during control, a recognition ROM 16 for storing a character recognition program for recognizing a read character, and a word based on the recognized character string. A dictionary ROM 17 for searching for, an LCD display unit 18 for displaying operation contents for the user, character recognition results, dictionary search results, etc.
Key input section 1 with various keys for performing pointing operations
9. Image RAM 20 for storing the captured image, CCD driver 22 for sending and driving a clock and a signal to the area CCD 21 which is a solid-state image sensor, and an analog image captured by the area CCD 21 driven by this CCD driver 22. Image signal processing circuit 23 for sampling and A / D converting data by a clock from CCD driver 22, area C
A motor driver 26 that controls a drive motor 25 that drives a zoom lens 24, which is an optical system member for making the image of the document surface incident on the CD 21, and continuously changes its focal length, is provided to illuminate the document surface. LED array 27a, 27b
An LED driver 28 for driving to light up is connected.

Next, the operation of the above embodiment will be described. 2 and 3 exemplify a series of processing contents from the image capturing to the display of the recognition result, and the operation control is mainly performed by the DSP 11.

At the beginning of the process, however, the key input unit 19
It waits for the operation of the capture key for instructing the capture of the image provided in the above (step S01). That is, the take-in key starts the taking-in with the scanner portion including the area CCD 21, the zoom lens 24, the drive motor 25, and the LED arrays 27a and 27b for taking in the image being in contact with the document surface to be taken in the image. It is the one that is operated accordingly.

When it is determined that the capture key is operated, the DSP 11 causes the LED driver 28 to operate the LED array 27.
The area CCD 21 is driven by the CCD driver 22 at the same time as a and 27b are turned on and the image data for one screen is taken in (step S02). At this time the zoom lens
The focal length of 24 is set to, for example, an intermediate (standard) position by driving the drive motor 25 under the control of the motor driver 26.

The image data for one screen captured by the area CCD 21 is sequentially digitized (binarized) by the image signal processing circuit 23 in synchronization with the drive clock of the CCD driver 22,
It is stored in the image RAM 20. FIG. 4A exemplifies the image data thus captured in the image RAM 20, and each point pixel forming the image is binarized in black and white.

When the image acquisition is completed, the DSP 11 obtains the horizontal black spot distribution of the image data stored in the image RAM 20 and cuts out the lines (step S03). FIG.
(B) illustrates an example of the image data in which line cutting has been performed in this way, and the region indicated by hatching in the drawing is a line region that is cut out.

After that, it is judged whether or not line cutting is possible (step S04). If line cutting is possible as shown in FIG. 4 (b), there is no problem. Since the zoom lens 24 is in a state of zooming up too much, the motor driver 26 drives the zoom lens 24 to a wider angle side by a certain width (zoom down) by the drive motor 25 (step S05), and again the above-mentioned. The process is restarted from the image capture in step S02.

FIG. 5A shows an example of the state of the optical system when the zoom lens 24 is driven to the wide-angle side, and 31 is the original surface to which an image is to be captured. From the figure, it can be seen that the range to be captured is wide.

When the image data having the line cutouts thus obtained is obtained, the vertical blackpoint distribution is further obtained for each of the cutout line regions, and the word cutout is performed (step S06). FIG. 4C exemplifies the image data in which the words are cut out in this way, and the hatched area in the drawing is the cut-out word area.

Thereafter, it is judged whether or not the word cutout has been completed (step S07). If the word cutout can be carried out as shown in FIG. 4C, there is no problem. Since it means that the zoom lens 24 has zoomed in too much, the motor driver 26
Drive motor 25 drives zoom lens 24 to a wider angle side by a certain width (zoom down) (step S0
5) Then, the process is restarted from the image capturing in step S02.

When the image data in which the words are cut out is obtained in this manner, the word region (here, "freedom") closest to the center of the image is set as the target word, and the reading range is determined from the coordinates of the four corners. The maximum zoom ratio is calculated without protruding (step S08).

In this case, the aspect ratio of the reading range is compared with the aspect ratio of the target word area, and if the target word area is horizontally long, the horizontal coordinate of the target word area is used as a reference, and if it is vertically long, the purpose is set. The zoom ratio is calculated based on the vertical coordinate of the word area.

Next, depending on whether or not the calculated zoom ratio is the same as the zoom ratio of the zoom lens 24 currently driven,
It is determined whether or not the zoom position of the zoom lens 24 can be left unchanged (step S09).

If the calculated zoom ratio happens to be the same as the zoom ratio of the zoom lens 24 that is currently being driven, there is no problem. However, if the zoom ratio is different, the motor driver 26 is set to the calculated zoom ratio. The drive motor 25 drives the zoom lens 24 toward the telephoto side (zoom up) (step S10), and the LED driver 28 is again positioned in a state where the target word area is maximized without extending beyond the reading range. The LED arrays 27a and 27b are driven to be driven by the LED driver, and at the same time, the CCD driver 22 drives the area CCD 21 to capture one screen of image data (step S).
11).

Image data for one screen captured by the area CCD 21 is sequentially digitized by the image signal processing circuit 23,
It is stored in the image RAM 20. FIG. 4D exemplifies image data including the area of the target word thus captured in the image RAM 20. In this case, the target word "fr
As a result of comparing the aspect ratio of “eedom” with the aspect ratio of the reading range, the target word region is horizontally long, and therefore the zoom ratio was calculated based on the horizontal coordinate of the target word region. Part of "i" and "f" in the word "economic" located below "freedom"
actors. Part of "f" of "" is the target word "fre
You can see that it has been taken in with "edom".

When the image capturing is completed, the DSP 11 obtains a horizontal black dot distribution of the image data including the area of the target word stored in the image RAM 20 and performs line segmentation (step S12), and then according to the zoom ratio. A tilt angle for tilt correction is calculated (step S13).

After that, the vertical black point distribution is obtained for the line regions that are continuously cut out, and the word is cut out (step S14), and the tilt correction is executed for the word region image using the calculated tilt angle. Yes (step S15).

Next, the image data of the obtained word area, whose inclination has been corrected, is decomposed into individual characters by character segmentation (step S16), and parameters of each character are extracted (step S17). Character recognition is performed according to a program stored in the recognition ROM 16 (step S
18).

Then, the dictionary ROM 17 is used by using the character strings each of which has the highest similarity as a result of recognition.
Is searched (step S19), the search result is converted into display data and displayed and output on the LCD display unit 18 (step S2).
0) As above, this series of processing is ended, and the process again shifts to the next operation waiting state (step S01).

Here, "f", "r", "e", and "e" are the characters with the highest similarity obtained as a result of recognition.
It is assumed that “d”, “o”, and “m” are obtained, and as a result of searching the dictionary ROM 17 using the character strings, the word “freedom” and its meaning are displayed on the LCD display unit 18.

In the above embodiment, the zoom lens 24 is used as a lens for continuously changing the focal length of the image picked up by the area CCD 21, and is driven by the drive motor 25 under the control of the motor driver 26. However, the present invention is not limited to this, and by using, for example, a bifocal lens that selectively switches the focal length instead of the zoom lens 24, a solenoid instead of the drive motor 25, and a solenoid driver instead of the motor driver 26, Although it is not possible to obtain a recognition result with such a high recognition rate, on the other hand, since the configuration and the control method are simpler, the cost can be reduced and high-speed processing can be performed.

[0029]

As described above in detail, according to the present invention, a character reading device and a character reading device capable of rapidly performing a character recognition process with a high recognition rate always corresponding to the size of a character to be read. A recognition device can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional circuit configuration according to an embodiment of the present invention.

FIG. 2 is a flowchart showing the contents of operation processing according to the embodiment.

FIG. 3 is a flowchart showing operation processing contents according to the embodiment.

FIG. 4 is a view for explaining the operation according to the embodiment.

FIG. 5 is a view for explaining the operation according to the embodiment.

[Explanation of symbols]

11 ... DSP, 12 ... Pulse generator (PG), 13 ... Instruction ROM, 14 ... System bus, 15 ... Work RAM, 16 ... Recognition ROM, 17 ... Dictionary ROM, 18 ... L
CD display section, 19 ... Key input section, 20 ... Image RAM, 21 ... Area CCD, 22 ... CCD driver, 23 ... Image signal processing circuit, 24 ... Zoom lens, 25 ... Drive motor, 26 ... Motor driver, 27a, 27b ... LED array, 28 ... LED driver, 31 ... Document surface.

Claims (6)

[Claims] 1. An image pickup device having a lens having a plurality of focal lengths and a solid-state image pickup device for picking up a target image, and a character string image at a specific position in the image obtained by the image pickup device. And setting means for variably setting the focal length of the lens of the captured image, and control means for re-capturing an image by the imaging means whose focal length of the lens is variably set by the setting means. Character reading device. 2. The character reading device according to claim 1, wherein the lens of the image pickup means is a zoom lens, and the setting means continuously and variably sets the focal length of the zoom lens. 3. The character according to claim 1, wherein the lens of the image pickup means is a multifocal lens, and the setting means selectively and variably sets the focal length of the multifocal lens to any one. Reader. 4. An image pickup device having a lens having a plurality of focal lengths and a solid-state image pickup device for picking up a target image, and a character string image at a specific position in the image obtained by the image pickup device. Then, setting means for variably setting the focal length of the lens of the captured image, control means for re-capturing an image by the imaging means for which the focal length of the lens is variably set by the setting means, and the control means are provided. A character recognition device comprising: a character recognition means for performing character recognition using a character string image in the image; and an output means for outputting a recognition result obtained by the character recognition means. 5. The character recognition device according to claim 4, wherein the lens of the image pickup means is a zoom lens, and the setting means continuously and variably sets the focal length of the zoom lens. 6. The character according to claim 4, wherein the lens of the imaging means is a multifocal lens, and the setting means selectively and variably sets the focal length of the multifocal lens to any one. Recognition device.