CN101009095A

CN101009095A - Fully-automatic intelligent blind reader

Info

Publication number: CN101009095A
Application number: CNA2007100668427A
Authority: CN
Inventors: 蒋清晓
Original assignee: Individual
Current assignee: Individual
Priority date: 2007-01-24
Filing date: 2007-01-24
Publication date: 2007-08-01

Abstract

This invention discloses one automatic intelligent blind reader, which comprises the following inner module connection parts: read control unit separately connected with scan and input unit, image process unit, image process unit and font identification unit, intelligent judgment unit, sound integration unit and sound unit, wherein, the memory unit is connected to scan input unit, image process unit, intelligent judgment unit and sound integration unit. This invention can help blind to read papers and files to avoid complex operations.

Description

Fully-automatic intelligent blind reader

Technical field

Overall design of the present invention is used for the blind person and the vision disorder personage is used for the fully-automatic intelligent arrangement for reading that text-to-speech transforms.Belonging to the information electronic technology field, is the clog-free disabled person's appurtenance of information.

Background technology

Blind person and vision disorder personage are the disadvantaged group of society, and the disappearance of visual capacity makes them obtain information in mode the most intuitively.Most in the world Word message is all expressed in the mode of vision now, and the blind person can only discern literal by the mode of braille and voice and obtain information.Yet in the actual life, Word message data more than 99% all is to occur with the paper counterparts form that the vision disorder personage can't read, the blind person is auxiliary the normal personage of anopsia to be to obtain these information fully down, paper counterparts information becomes the huge obstacle that blind person and vision disorder personage participate in real social activities, has also caused blind person and the low relatively serious consequence of vision disorder crowd educational level in the actual life.

Now blind person and the vision disorder personage mode of obtaining Word message mainly contains: 1. the form by CASE(Computer Aided Software Engineering) is converted into information that 2. voice be converted into tangible perception by utility appliance with written historical materials made of paper to the electronics Word message and 3. by utility appliance written historical materials information made of paper is converted into every kind of mode of voice messaging its relative merits are arranged, first kind of mode can't be handled for paper counterparts such as periodical newspaper file or the like, second way user need pass through complexity and the training of system.The third mode can directly be converted into voice messaging with written historical materials information, even for illiterate blind person and vision disorder personage, also can use, and is present the most outstanding information obtain manner.

Yet all there is following problem in present most of text-to-speech conversion equipment: at first, operation is quite complicated, and use is required great effort very much for twenty-twenty personage, and the people of obstacle is arranged for eyesight still more; Secondly, for the blind person, because the direction of their text printout on can't the perception paper counterparts, therefore in reading process wrong situation can often take place to place, current text-to-speech conversion equipment can't the intelligent decision paper counterparts the placement situation, use to the blind person and to have caused very big obstacle; At last, in reading process, these equipment all can't provide good reading control system, have caused very big difficulty for actual the use.

Therefore, be sought after a kind of can assisting blind and the vision disorder personage in the technology that does not have also can realize very easily under normal person's help situation written historical materials information reading made of paper.

Summary of the invention

The purpose of this invention is to provide a kind of fully-automatic intelligent blind reader.

The internal module annexation of fully-automatic intelligent blind reader is: read control module and join with input scan cell, graphics processing unit, word recognition unit, intelligent distinguishing unit, phonetic synthesis unit, pronunciation unit respectively, storage unit is joined with input scan cell, graphics processing unit, intelligent distinguishing unit, phonetic synthesis unit respectively.

Described reading control module comprises reading controller module and system control module, and reading controller module and system control module internal module annexation are: reader keyboard and USB keyboard controller chip, first USB port, the second USB port system flow automatic control module, voice suggestion control module are joined.

The reader keyboard has 8, is respectively to start to read aloud key, read aloud Pause key, read next key, read a key, change of voice key, to read aloud accelerator key, read aloud deceleration key and reset key.

Word recognition unit internal module annexation is: image cutting module and character feature extraction module, output module, standard feature library module join as a result.

Intelligent decision unit internal module annexation is: standard words library module and context intelligent decision are searched module, statistics discrimination module, output module join.

Graphics processing unit internal module annexation is: the denoising module is joined with luminance contrast adjustment module, image rotary module.

Phonetic synthesis unit and pronunciation unit internal module annexation are: text-to-speech modular converter and control command voice storage module and change of voice modified tone module, loudspeaker join.

The present invention can carry out the reading of written historical materialss made of paper such as books and periodicals, newspaper, file by full automatic assisting blind, avoided the blind person because of cannot see the situation that to carry out complex operations, can under can't distinguishing the situation of file placement direction made of paper and angle, the blind person accurately read, more have and read control function easily, the more convenient and high-level efficiency that the blind person is read carry out.

Description of drawings

Fig. 1 is the circuit block diagram of fully-automatic intelligent blind reader;

Fig. 2 is the circuit block diagram of reading controller of the present invention;

Fig. 3 is a reader keyboard synoptic diagram of the present invention;

Fig. 4 is the circuit block diagram based on FPGA of the present invention;

Fig. 5 is phonetic synthesis of the present invention unit and pronunciation cellular construction figure.

Embodiment

As shown in Figure 1, fully-automatic intelligent blind reader comprises input scan cell, reads control module, word recognition unit, intelligent decision unit, graphics processing unit, phonetic synthesis unit, storage unit and pronunciation unit.Read control module 2 and join 8 with input scan cell 1, graphics processing unit 5, word recognition unit 3, intelligent distinguishing unit 4, phonetic synthesis unit 6, pronunciation unit respectively, storage unit 7 is joined with input scan cell 1, graphics processing unit 5, intelligent distinguishing unit 4, phonetic synthesis unit 6 respectively.

Input scan cell 1 mainly is made up of optical imagery scanner head, mechanical transmission mechanism and control and A/D conversion processing circuit.Wherein, the optical imagery scanner head is made up of strip light spot source, three stripe-shape plane catoptrons, condenser lens (lens combination) and CCD charge-coupled image sensors.Bar shaped fluorescent tube and stripe-shape plane catoptron along continuous straight runs on scanner head is placed.The parallel rays that the bar shaped fluorescent tube sends during work after condenser lens (or lens combination) enters CCD, is converted to light signal the analog electrical signal that with light intensity be directly proportional by CCD through paper counterparts, stripe-shape plane mirror reflects.Mechanical transmission mechanism is made up of stepper motor, transmission gear, driving belt.Scanner head is supported by the circular support slide bar, is stuck on the driving belt, is driven along supporting slide bar by driving belt and moves.The A/D conversion processing circuit is made up of A/D conversion chip and corresponding external circuit, and the analog electrical signal of changing through CCD sends storage unit (7) to by the digital signal that the A/D conversion processing circuit becomes expression paper counterparts imaging gray scale.

As shown in Figure 2, read control module and be made up of reading controller module 21 and system control module 22, reading controller module 21 and system control module 22 internal module annexations are: reader keyboard 211 joins with USB keyboard controller chip 212, first USB port 213, second USB port, 223 system flow automatic control modules 222, voice suggestion control module 221.

Reading controller module wherein, its major function is to accept user's instruction, and this instruction is sent in the system control module by first USB port goes.The reading controller module mainly is made up of reader keyboard, USB keyboard controller chip and first USB port.Because user of the present invention much is the blind person, so in the configuration design of reader keyboard, the button that uses profile to differ greatly.

As shown in Figure 3, in the present embodiment, the button of reader keyboard has 8,8 orders of control of reading, the corresponding button of each order, be respectively to start to read aloud key 2111, read aloud Pause key 2112, read next key 2113, read a key 2114, change of voice key 2115, to read aloud accelerator key 2116, read aloud deceleration key 2117 and reset key 2118, USB keyboard controller chip 212 has adopted the AT43USB324 of American ATMEL.Any control command of user in reading process is all by button, and the coding through USB keyboard controller chip is sent to system control module by first USB port.System control module is made up of system flow automatic control module 222 and voice suggestion control module 221.The system flow automatic control module is responsible for controlling the operation of total system, and sends current system running state to the voice suggestion control module.The voice suggestion control module is connected with phonetic synthesis unit 6, and its major function is the voice of the synthetic current system running state in control phonetic synthesis unit, thereby which state is the current system of prompting user run to.Voice suggestion is very important for blind person user.In the present embodiment, system flow automatic control module and voice suggestion control module all realize based on FPGA

As shown in Figure 4 in the present embodiment, word recognition unit 3, intelligent decision unit 4, graphics processing unit 5 and system flow automatic control module 222 and voice suggestion control module 221 all realize based on FPGA.

Described word recognition unit 3 internal module annexations are: image cutting module 31 is with character feature extraction module 32, output module 34, standard feature library module 33 join as a result.Intelligent decision unit 4 internal module annexations are: standard words library module 41 and context intelligent decision are searched module 42, statistics discrimination module 43, output module 44 and are joined.Graphics processing unit 5 internal module annexations are: denoising module 51 is joined with luminance contrast adjustment module 52, image rotary module 53.

In the present embodiment, FPGA adopts the FPGAXC4VLX100 of the Virtex-4 series of Xilinx company product, and storage unit 7 is realized by SRAM storer K7N163601M.The flow process of system flow automatic control module control total system is the core of native system.After the user supressed start key by the reading controller module, this module can each module in the automatic control system be operated under need not the situation of user intervention in order, makes the Word message of paper counterparts be converted to voice signal.And, adjust operating process according to the control command of user when reading, satisfy customer requirements.Operate in order in system, when being in different states, the system flow automatic control module is passed to the voice suggestion control module with the information of current state and the information of user key-press, the voice suggestion control module produces corresponding voice suggestion control signal control phonetic synthesis module, reaches prompting user's current state and user key-press result's purpose.

Graphics processing unit has comprised denoising module 51, luminance contrast adjustment module 52 and image rotary module 53.The denoising module functions is to read the view data of input scan cell storage from storage unit, and removes the noise spot in the image, to improve the accuracy of literal identification.The major function of luminance contrast adjustment module is with the image enhancing contrast ratio behind the removal noise, and according to the adaptive adjustment brightness of the characteristic of image itself, is in order to increase discrimination equally.The major function of image rotary module is to export after image is rotated the angle of appointment, this angle can be that any one integer-valued angle and intelligent decision module cooperate correct the identifying of written historical materials image made of paper that can place with the angle of inclination from 0～360 °.Word recognition unit 3 mainly comprises image cutting module 31, character feature extraction module 32, standard feature library module 33 and output module 34 as a result.Image cutting module functions is that the image after handling through graphics processing unit is carried out cutting according to the zone of single literal, and image information just is separated into the some little pictures that contain single Word message like this.The character feature extraction module extracts the characteristic information that contains in the little picture of single Word message about character according to set algorithm, and the standard feature information in this information and the standard feature library module compared, the standard character of selecting the most approaching characteristic information place is as recognition result.Output module is integrated the recognition result information in the entire image as a result, and sends the intelligent decision unit to.

Intelligent decision unit 4 comprises standard words library module 41, and the context intelligent decision is searched module 42 and statistics discrimination module 43 and output module 44.The dictionary that contains a large amount of Chinese and englishes in the standard words library module is to search the standard that module provides intelligent decision for the context intelligent decision.The context intelligent decision is searched module each is searched all speech by the recognition result literal that word recognition unit identification draws in the standard words library module, and with in fact hereinafter comparing of this character in the recognition result, if in fact hereinafter all not the having of recognition result can find corresponding word in the standard words library module, so just think that this word can not become speech, and this result is passed to the statistics discrimination module.The one-tenth speech rate of all characters in the statistics discrimination module statistical recognition result full text.Through a large amount of facts have proved, if in the passage all characters add assembly speech quantity divided by total number of characters of article less than certain numerical value, can think that so this section literal is insignificant mess code.In reading process, if the user is mistaken with written historical materials placement direction made of paper accidentally, though can draw the identification conclusion so, but nonsensical mess code.Through the judgement of statistical recognition module, effectively whether the result that can draw this time scanning identification conclusion.If the result is effective, then give the phonetic synthesis unit by output module output recognition result.Otherwise, then image is rotated operation, and carries out identification process again by system's automatic flow control module, rotate to the tram of the document up to image, the position that can be identified just, this result will export the phonetic synthesis unit to.

As shown in Figure 5, phonetic synthesis unit 6 mainly comprises text-to-speech modular converter 61 and control command voice storage module 62.The major function of text-to-speech modular converter is to be converted to voice signal and to send the pronunciation unit to by the correct Word message after the intelligent decision.The control command voice storage module read control module control, its storage inside the prompt tone of each button on each flow process of system and the reading controller, read the control module move instruction, then control voice storage module and will send the corresponding prompt tone signal, send to the pronunciation unit after the signal of this signal and the output of text-to-speech modular converter is superimposed.In the present embodiment, the text-to-speech modular converter has used the OSYN06188 chipspeech, and the control command voice storage module has been used the AP89043 speech chip.

Pronunciation unit 8 is made up of change of voice modified tone module 81 and loudspeaker 82, and the main effect of change of voice modified tone module is that the sound of phonetic synthesis unit output is carried out real-time processing according to user's needs, as carries out the processing or the like of switching and modify tone of men and women's sound.Change of voice modified tone module has adopted the SD771D single-chip processor of changing voice in real time.

Although in conjunction with a limited number of embodiment the present invention has been described, those skilled in the art obviously know many modifications and variant in view of the above.Accompanying Claim is intended to comprise the modification and the variant of these true spirit according to the invention and scope.

Claims

1. fully-automatic intelligent blind reader, it is characterized in that, read control module (2) and join (8) with input scan cell (1), graphics processing unit (5), word recognition unit (3), intelligent distinguishing unit (4), phonetic synthesis unit (6), pronunciation unit respectively, storage unit (7) is joined with input scan cell (1), graphics processing unit (5), intelligent distinguishing unit (4), phonetic synthesis unit (6) respectively.

2. a kind of fully-automatic intelligent blind reader according to claim 1, it is characterized in that, described reading control module (2) comprises reading controller module (21) and system control module (22), and reading controller module (21) and system control module (22) internal module annexation are: reader keyboard (211) joins with USB keyboard controller chip (212), first USB port (213), second USB port (223) system flow automatic control module (222), voice suggestion control module (221).

3. a kind of fully-automatic intelligent blind reader according to claim 2, it is characterized in that, described reader keyboard (211) button has 8, is respectively to start to read aloud key (2111), read aloud Pause key (2112), read next key (2113), read a key (2114), change of voice key (2115), to read aloud accelerator key (2116), read aloud deceleration key (2117) and reset key (2118).

4. a kind of fully-automatic intelligent blind reader according to claim 1, it is characterized in that described word recognition unit (3) internal module annexation is: image cutting module (31) and character feature extraction module (32), output module (34), standard feature library module (33) join as a result.

5. a kind of fully-automatic intelligent blind reader according to claim 1, it is characterized in that described intelligent decision unit (4) internal module annexation is: standard words library module (41) and context intelligent decision are searched module (42), statistics discrimination module (43), output module (44) and are joined.

6. a kind of fully-automatic intelligent blind reader according to claim 1, it is characterized in that described graphics processing unit (5) internal module annexation is: denoising module (51) is joined with luminance contrast adjustment module (52), image rotary module (53).

7. a kind of fully-automatic intelligent blind reader according to claim 1, it is characterized in that described phonetic synthesis unit (6) and pronunciation unit (8) internal module annexation are: text-to-speech modular converter (61) and control command voice storage module (62) are joined with change of voice modified tone module (81), loudspeaker (82).