CN109920509A - Medicine information recognition methods, device, computer equipment and storage medium - Google Patents
Medicine information recognition methods, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109920509A CN109920509A CN201910042928.9A CN201910042928A CN109920509A CN 109920509 A CN109920509 A CN 109920509A CN 201910042928 A CN201910042928 A CN 201910042928A CN 109920509 A CN109920509 A CN 109920509A
- Authority
- CN
- China
- Prior art keywords
- information
- drug
- text
- nomenclature
- broadcasted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000003814 drug Substances 0.000 title claims abstract description 248
- 229940079593 drug Drugs 0.000 title claims abstract description 176
- 238000000034 method Methods 0.000 title claims abstract description 37
- 239000000284 extract Substances 0.000 claims abstract description 22
- 238000005266 casting Methods 0.000 claims description 24
- 238000010606 normalization Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000033764 rhythmic process Effects 0.000 claims 1
- 238000012856 packing Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- RYYVLZVUVIJVGH-UHFFFAOYSA-N caffeine Chemical compound CN1C(=O)N(C)C(=O)C2=C1N=CN2C RYYVLZVUVIJVGH-UHFFFAOYSA-N 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 239000002775 capsule Substances 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- RZVAJINKPMORJF-UHFFFAOYSA-N Acetaminophen Chemical compound CC(=O)NC1=CC=C(O)C=C1 RZVAJINKPMORJF-UHFFFAOYSA-N 0.000 description 2
- QGZKDVFQNNGYKY-UHFFFAOYSA-N Ammonia Chemical compound N QGZKDVFQNNGYKY-UHFFFAOYSA-N 0.000 description 2
- 206010004542 Bezoar Diseases 0.000 description 2
- 241001672694 Citrus reticulata Species 0.000 description 2
- 230000000857 drug effect Effects 0.000 description 2
- 230000005802 health problem Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000004806 packaging method and process Methods 0.000 description 2
- 229960005489 paracetamol Drugs 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 239000011358 absorbing material Substances 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 229910021529 ammonia Inorganic materials 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000029142 excretion Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000003292 glue Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000002503 metabolic effect Effects 0.000 description 1
- 230000004060 metabolic process Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 230000035790 physiological processes and functions Effects 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
This application involves image identification technical fields, more particularly to a kind of medicine information recognition methods, device, computer equipment and storage medium, include: to obtain medicine information image to be identified, extracts text information and digital information from the medicine information image to be identified;The nomenclature of drug text in the text information is extracted, generates nomenclature of drug after combination, extracts the usage text information in the text information, obtains drug usage information after the usage text information and the digital information are combined;The text in preset sound bank is traversed, the nomenclature of drug and the corresponding voice to be broadcasted of the drug usage information are obtained, the voice to be broadcasted is broadcasted according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information.The application effectively realizes the automatic identification to medicine information and the speech habits according to different crowd are broadcasted.
Description
Technical field
This application involves image identification technical field more particularly to a kind of medicine information recognition methods, device, computers to set
Standby and storage medium.
Background technique
Drug is the substance to prevent, treat and diagnose the illness.Theoretically, drug, which refers to, all can influence biological organs
The chemical substance of physiological function and metabolic activity in cells belongs to the scope of drug.Drug physiological disposition varies with each individual, it is many because
Element can influence absorption, distribution, metabolism, the excretion of drug, to influence final drug effect.Some other factor can also influence drug
Effect, such as heredity, the interaction between drug, disease.
Currently, for the patient needing the acquisition of the title of drug and dosage information by doctor's advice or drug packet
Explanation in mounted box.And often drug has multiple titles, such as trade name, medical name and common first names, this just gives patient in drug
Very big puzzlement is caused when use, simultaneously because the text of doctor's advice is difficult to, and the usage on Key works Drug packing and metering text
Usual very little is typically only capable to be entrusted according to other people language to take medicine particularly with the elderly when determining how medication.In this way
There is great security risk, being easy to cause patient to wrongly take drug leads to injury to body.
Summary of the invention
Based on this, it is necessary to asking for title and usage and dosage can not be provided to patient at any time for due to drug information complexity
Topic provides a kind of medicine information recognition methods, device, computer equipment and storage medium.
A kind of medicine information recognition methods, includes the following steps:
Medicine information image to be identified is obtained, extracts text information and number from the medicine information image to be identified
Information;
The nomenclature of drug text in the text information is extracted, generates nomenclature of drug after combination, extracts the text letter
Usage text information in breath obtains drug usage letter after being combined the usage text information and the digital information
Breath;
The text in preset sound bank is traversed, the nomenclature of drug is obtained and the drug usage information is corresponding wait broadcast
Voice is reported, the voice to be broadcasted is broadcasted according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information.
It is described to obtain medicine information image to be identified in a wherein possible embodiment, from the drug to be identified
Text information and digital information are extracted in information image, comprising:
When it is any include medicine information images to be recognized to image acquisition device screen distance be less than preset distance
When threshold value, the images to be recognized is acquired to obtain original images to be recognized;
Color normalization images to be recognized is obtained after carrying out gray processing processing to the original images to be recognized;
The color normalizing is obtained after progressively scanning the color normalization images to be recognized according to preset character height
Change all characters in images to be recognized;
The scanning is obtained after the character that scanning recognition goes out is matched with the character information being pre-stored in database to know
Not Chu the corresponding text information of character or digital information.
In a wherein possible embodiment, the nomenclature of drug text extracted in the text information, after combination
Nomenclature of drug is generated, the usage text information in the text information is extracted, by the usage text information and the number
Information obtains drug usage information after being combined, comprising:
The recognition result of each text in the text information is matched with the word in word library, obtains constituting word
The recognition result of language;
The nomenclature of drug Feature Words in the recognition result for constituting word are extracted, is named and is advised according to preset nomenclature of drug
Then, the nomenclature of drug is obtained after nomenclature of drug Feature Words described in permutation and combination;
The usage text information in the recognition result for constituting word is extracted, the usage text information and number are believed
Breath obtains several drug usage information sentences after being combined;
The semantic confidence degree for calculating each drug usage information sentence, according to the semantic confidence degree, determine described in
Drug usage information.
In a wherein possible embodiment, the text traversed in preset sound bank obtains the drug name
Claim voice to be broadcasted corresponding with the drug usage information, it is corresponding according to the nomenclature of drug and the drug usage information
Prosodic parameter broadcasts the voice to be broadcasted, comprising:
The text in default sound bank is traversed, the nomenclature of drug and the corresponding several originals of the drug usage information are obtained
Beginning voice;
User language use information is obtained, determines a certain raw tone for institute according to the user language use information
State voice to be broadcasted;
The historical data that word is broadcasted in the nomenclature of drug and the drug usage information is obtained, according to the history number
According to original prosodic parameter is obtained, the voice to be broadcasted is broadcasted according to the original prosodic parameter, receives the voice to be broadcasted
Reflected acoustic wave, obtain final prosodic parameter after correcting the original prosodic parameter according to the reflected acoustic wave, according to it is described most
Whole prosodic parameter broadcasts the voice to be broadcasted.
In a wherein possible embodiment, the character that scanning recognition is gone out and the character being pre-stored in database
Information obtains the corresponding text information of character or digital information that the scanning recognition goes out after being matched, comprising:
The connected domain that each stroke of the character is constituted is obtained, determines the boundary rectangle of each connected domain;
The pixel value for obtaining each point in the boundary rectangle carries out piecemeal to the boundary rectangle according to the pixel value
After form several sub-blocks;
According to the ratio of width to height of preset characters, several sub-blocks are merged into block to be identified;
According to the stroke pixel value of preset character in the pixel value and database of each point in the block to be identified into
Row compares, and obtains stroke confidence level;
Stroke corresponding to the maximum value of the stroke confidence level of each block to be identified is obtained, these strokes are folded
Text corresponding to the character in the boundary rectangle or number are obtained after adding.
In a wherein possible embodiment, the text traversed in preset sound bank obtains the drug name
Claim voice to be broadcasted corresponding with the drug usage information, it is corresponding according to the nomenclature of drug and the drug usage information
Prosodic parameter casting it is described wait broadcast voice after, further include being modified according to field feedback to voice broadcast content
Step specifically includes:
The feedback information that user terminal is sent is received, problem information characteristic character included in the feedback information is extracted,
The corresponding question attributes of the feedback information are obtained according to described problem information characteristics character;
The prosodic parameter is adjusted according to described problem attribute, if the prosodic parameter adjusted reaches described
The user terminal demand for including in feedback information, the then prosodic parameter broadcasted prosodic parameter adjusted as medicine information are no
The voice messaging for then obtaining user's input corrects the voice to be broadcasted according to the voice messaging of user input, until reaching
To the demand of user terminal.
In a wherein possible embodiment, word in the acquisition nomenclature of drug and the drug usage information
The historical data of casting obtains original prosodic parameter according to the historical data, according to the original prosodic parameter casting
Voice to be broadcasted, receives the reflected acoustic wave of the voice to be broadcasted, and corrects the original prosodic parameter according to the reflected acoustic wave
After obtain final prosodic parameter, the voice to be broadcasted is broadcasted according to the final prosodic parameter, comprising:
The historical data for obtaining the casting of the nomenclature of drug and the word in the drug usage information, extracts each
The most pitch of word frequency of occurrence, the duration of a sound and volume carry out the most pitch of frequency of occurrence, the duration of a sound and volume at binaryzation
Reason, forms the original prosodic parameter after splicing;
Preset sentence casting speed is obtained, according to the sentence broadcasts speed and the original prosodic parameter is broadcasted
Voice to be broadcasted;
The primary reflection wave from the reflected voice to be broadcasted of preset sound reflection wall is received, the original is filtered
Beginning back wave obtains practical reflected acoustic wave, wherein the formula of back wave filtering are as follows:
In formula, NMSE indicates standard deviation, and x (n) indicates that primary reflection wavelength, y (n) indicate estimated value, will be described original anti-
Ejected wave length obtains the practical reflected acoustic wave after making the difference with the standard deviation;
The practical reflected acoustic wave and preset reflected acoustic wave threshold value are made the difference, according to difference and pitch, the duration of a sound and
The corresponding relationship of volume corrects the original prosodic parameter and obtains final prosodic parameter, is broadcasted according to the final prosodic parameter
The voice to be broadcasted.
A kind of medicine information identification device, including following module:
Information extraction modules are set as obtaining medicine information image to be identified, from the medicine information image to be identified
Extract text information and digital information;
Information identification module is set as extracting the nomenclature of drug text in the text information, and drug name is generated after combination
Claim, extracts the usage text information in the text information, the usage text information and the digital information are subjected to group
Drug usage information is obtained after conjunction;
Information broadcasting module is set as traversing the text in preset sound bank, obtains the nomenclature of drug and the medicine
The corresponding voice to be broadcasted of product usage information is broadcast according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information
Report the voice to be broadcasted.
A kind of computer equipment, including memory and processor are stored with computer-readable instruction in the memory, institute
When stating computer-readable instruction and being executed by the processor, so that the processor executes the step of above-mentioned medicine information recognition methods
Suddenly.
A kind of storage medium being stored with computer-readable instruction, the computer-readable instruction are handled by one or more
When device executes, so that the step of one or more processors execute above-mentioned medicine information recognition methods.
Compared with current mechanism, the application is had the advantages that
(1) by effectively being identified to medicine information, and voice broadcast is carried out to user using suitable mode, thus
Different user is set can correctly to obtain medicine information, so as to avoid health problem caused by drug improper use;
(2) by handling collected medicine information, the accuracy of medicine information identification is improved;
(3) by effectively obtaining nomenclature of drug and usage information to recognition result progress semantic analysis, so as to
Accurate instruction user uses drug;
(4) prosodic parameter is adjusted obtain meeting user speak habit medicine information casting by way of.
Detailed description of the invention
By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field
Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the application
Limitation.
Fig. 1 is a kind of overall flow figure of the medicine information recognition methods of the application in one embodiment;
Fig. 2 is the information extraction process signal in a kind of medicine information recognition methods of the application in one embodiment
Figure;
Fig. 3 is the information identification process signal in a kind of medicine information recognition methods of the application in one embodiment
Figure;
Fig. 4 is that the information in a kind of medicine information recognition methods of the application in one embodiment broadcasts process signal
Figure;
Fig. 5 is a kind of structure chart of the medicine information identification device of the application in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, and
It is not used in restriction the application.
Those skilled in the art of the present technique are appreciated that unless expressly stated, singular " one " used herein, " one
It is a ", " described " and "the" may also comprise plural form.It is to be further understood that being arranged used in the description of the present application
Diction " comprising " refer to that there are the feature, integer, step, operation, element and/or component, but it is not excluded that in the presence of or addition
Other one or more features, integer, step, operation, element, component and/or their group.
Fig. 1 is a kind of overall flow figure of the medicine information recognition methods of the application in one embodiment, such as Fig. 1 institute
Show, a kind of medicine information recognition methods, comprising the following steps:
S1 obtains medicine information image to be identified, extracted from the medicine information image to be identified text information and
Digital information;
Specifically, can be extracted by the way of taking pictures outside commercially available drug when obtaining medicine information image to be identified
Packaging image, can also be from medicine information database, as extracted medicine information figure in hospital database or pharmaceutical factory database
Picture.Extracted text information is mainly that " nomenclature of drug ", " usage ", " points for attention " etc. are believed concerning the text of drug user's health
Breath, and digital information is mainly the subsequent specific number of usage, for example, 1 day 3 times, in " 1 " and " 3 ".
S2 extracts the nomenclature of drug text in the text information, generates nomenclature of drug after combination, extracts the text
Usage text information in information obtains drug usage letter after being combined the usage text information and the digital information
Breath;
Specifically, when extracting nomenclature of drug text, because general drug is divided into medical name and common first names, by drug
After title text extracts, need to distinguish medical name and common first names.Usage text usually will appear in drug outer packing
Then " usage " this 2 words will search the detailed directions with number as long as searching this 2 words before and after it.Than
Such as, usage: 1 day 3 times, 1 time 1 etc..
S3 traverses the text in preset sound bank, obtains the nomenclature of drug and the drug usage information is corresponding
Voice to be broadcasted broadcasts the language to be broadcasted according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information
Sound.
It specifically, be stored with the pronunciation of all Chinese characters in sound bank, that is, include mandarin also include dialect, it can also be
Foreign language voice is added in sound bank.The deviation on understanding is generated to content since different speech intonations will use family,
Need to obtain nomenclature of drug information and drug usage information habitually speech intonation, that is, prosodic parameter.Prosodic parameter master
If the dwell interval between pitch and each sound.
The present embodiment by effectively being identified to medicine information, and carries out voice to user using suitable mode and broadcasts
Report, to make different user can correctly obtain medicine information, so as to avoid health problem caused by drug improper use.
Fig. 2 is the information extraction process signal in a kind of medicine information recognition methods of the application in one embodiment
Figure, as shown, the S1, obtains medicine information image to be identified, extracts text from the medicine information image to be identified
Word information and digital information, comprising:
S101, when it is any include medicine information images to be recognized to image acquisition device screen distance be less than it is preset
When distance threshold, the images to be recognized is acquired to obtain original images to be recognized;
Specifically, can be acquired twice using CCD camera for the acquisition of drug image information to be identified, then will
Drug image to be identified collected twice carries out pixel veritification, if the result veritified of pixel preset error range with
It is interior, then it carries out in next step, otherwise re-starting acquisition.
S102, color normalization images to be recognized is obtained after carrying out gray processing processing to the original images to be recognized;
Wherein, the color normalization, which refers to, carries out black whitening processing for colored image, is that color image becomes black and white
Two chromatic graph pictures.
S103, the color is obtained after progressively scanning the color normalization images to be recognized according to preset character height
Normalize all characters in images to be recognized;
Wherein, preset character height is determined according to the common text size of Key works Drug packing, can be according to drug
The size of packaging determines preset character height.
S104, described sweep is obtained after being matched the character that scanning recognition goes out with the character information being pre-stored in database
Retouch the corresponding text information of the character identified or digital information.
Specifically, can be compared to the stroke of character information when carrying out character information matching, only when scanning
Character information and the character information that is pre-stored in database, all strokes of the two are consistent, then both illustrate matching, otherwise
It is all to mismatch.
The present embodiment improves the accuracy of medicine information identification by handling collected medicine information.
Fig. 3 is the information identification process signal in a kind of medicine information recognition methods of the application in one embodiment
Figure, as shown, the S2, extracts the nomenclature of drug text in the text information, generate nomenclature of drug after combination, extract
Usage text information in the text information obtains medicine after being combined the usage text information and the digital information
Product usage information, comprising:
S201, the recognition result of each text in the text information is matched with the word in word library, is obtained
Constitute the recognition result of word;
Wherein, the recognition result for constituting word refers to two or above text can be obtained in word library it is corresponding
Word, for example, the recognition result of two texts of certain in text information are as follows: " use ", " method " then can be with the words in equivalent repertorie:
" usage ", and the recognition result of another two word is " glue ", " element " can not then obtain corresponding word in word library.
Nomenclature of drug Feature Words in S202, the extraction recognition result for constituting word, according to preset nomenclature of drug
Naming rule obtains the nomenclature of drug after nomenclature of drug Feature Words described in permutation and combination;
Specifically, nomenclature of drug Feature Words may include: medical name and trade name, such as: paracetamol caffein atificial cow-bezoar is medical
Name;Capsule for cold is trade name, when carrying out permutation and combination, needs for " ammonia coffee " and " Huang Min " to be combined, and cannot
" quick-acting " and " Huang Min " are combined.Meanwhile " cold quick acting capsule " can not be combined into according to nomenclature of drug naming rule
Such title.
S203, extract it is described constitute word recognition result in usage text information, by the usage text information with
Digital information obtains several drug usage information sentences after being combined;
Specifically, can all occur " usage " or " dosage " the two texts on Key works Drug packing, searched from recognition result
Rope goes out the two words, and it is attached to extract " usage " or " dosage " this 2 words for scanning sequency when then being identified according to OCR
Then text and number are combined to obtain several drug usage information sentences by close text and number.
S204, the semantic confidence degree for calculating each drug usage information sentence are determined according to the semantic confidence degree
The drug usage information.
Specifically, the used calculation formula when carrying out confidence calculations are as follows:
N=Z × (P × (1-P))/E;
N is confidence level, and E is the standard deviation of sample average, and Z is overall error, and P is that target semanteme quantity accounts for overall semantic quantity
Ratio.
If confidence level is greater than confidence threshold value, otherwise it is not drug usage information which, which is drug usage information,.
The present embodiment, by effectively obtaining nomenclature of drug and usage information to recognition result progress semantic analysis, from
And can accurate instruction user use drug.
Fig. 4 is that the information in a kind of medicine information recognition methods of the application in one embodiment broadcasts process signal
Figure obtains the nomenclature of drug and drug usage letter as shown, the S3, traverses the text in preset sound bank
Cease corresponding voice to be broadcasted, according to the nomenclature of drug and the drug usage information corresponding prosodic parameter casting it is described to
Broadcast voice, comprising:
S301, the text preset in sound bank is traversed, obtains the nomenclature of drug and the drug usage information is corresponding
Several raw tones;
Specifically, being stored with the corresponding voice messaging of different literals in sound bank, each text corresponds at least one
A voice messaging, for example, voice messaging corresponding to " eating " this text can be " chi " and " qia ".Inquiring default voice
It before text in library, needs to convert nomenclature of drug, for example common first names corresponding to medical entitled paracetamol caffein atificial cow-bezoar are
Capsule for cold often only knows common first names in patients, without knowing medical name, and in drug outer packing or doctor
Medical name is then often write in the prescription issued, this just needs that medical name is converted into common first names when carrying out voice broadcast.
S302, user language use information is obtained, a certain original language is determined according to the user language use information
Sound is the voice to be broadcasted;
Wherein, user language use information is primarily referred to as user using which kind of language and dialect, such as English, Chinese, and
It is using mandarin or Guangdong language in Chinese.The acquisition of user language use information can be according to historical data either basis
User input voice messaging, such as a user input " eating " this word pronunciation be " qia " then may determine that user makes
Language is the language near Hunan.
S303, the historical data for obtaining word casting in the nomenclature of drug and the drug usage information, according to described
Historical data obtains original prosodic parameter, broadcasts the voice to be broadcasted according to the original prosodic parameter, receives described wait broadcast
The reflected acoustic wave for reporting voice obtains final prosodic parameter after correcting the original prosodic parameter according to the reflected acoustic wave, according to
The final prosodic parameter broadcasts the voice to be broadcasted.
Specifically, prosodic parameter mainly includes, pitch, tone color, dwell interval.When carrying out reflected acoustic wave amendment, need
It is carried out under a closed room, only has a face in six faces in this room and be made of reflected acoustic wave material, other five faces
It is made of absorbing material, the sound wave of that face reflection of reflected acoustic wave material need to be only received in this way, then according to sound wave number
Value obtains practical prosodic parameter, and practical prosodic parameter is compared with original prosodic parameter, then utilizes error correction function
Original prosodic parameter is modified to obtain final prosodic parameter.
The present embodiment is broadcasted by being adjusted to obtain meeting the speak medicine information of habit of user to prosodic parameter
Mode.
In one embodiment, the S104, the character that scanning recognition is gone out and the character information that is pre-stored in database
The corresponding text information of character or digital information that the scanning recognition goes out are obtained after being matched, comprising:
The connected domain that each stroke of the character is constituted is obtained, determines the boundary rectangle of each connected domain;
Specifically, the connected domain that each stroke is constituted refers to, for example the connected domain of " Pie " and " one " two strokes composition is exactly
Region where " factory ", and the boundary rectangle of connected domain is exactly with " one " for horizontal edge, the length perpendicular to horizontal edge of " Pie " is vertical edge
Rectangle.
The pixel value for obtaining each point in the boundary rectangle carries out piecemeal to the boundary rectangle according to the pixel value
After form several sub-blocks;
Specifically, to boundary rectangle carry out sub-block segmentation when, first preset a sub-block size, by the boundary rectangle into
The segmentation of row homalographic, then counts pixel value in each sub-block, then expands the sub-block size that pixel value is less than average value,
The sub-block size that pixel value is greater than average value is reduced, until the pixel value in each sub-block is consistent, generates final sub-block.
According to the ratio of width to height of preset characters, several sub-blocks are merged into block to be identified;
Wherein, the ratio of width to height of text refers to the width of character area and the ratio of height, for example the ratio of width to height of " mouth " is 1,
And the ratio of width to height of " state " is 3 to 4, the block formed after sub-block is merged will guarantee that the ratio of width to height is consistent with preset the ratio of width to height.
According to the stroke pixel value of preset character in the pixel value and database of each point in the block to be identified into
Row compares, and obtains stroke confidence level;
Wherein, when progress pixel value compares, binary conversion treatment first can be done to two pixel values respectively, then carried out again
The comparison of pixel value, when obtaining stroke confidence level using following formula:
N=Z × (2 × S/d)/2;
N is confidence level, and S is the standard of any stroke pixel value in block to be identified in the pixel value and database of each point
Difference, Z are the pixel value of any stroke in database, and d is max value of error.Wherein, max value of error can be according to historical data
Statistics obtains, and general max value of error is 5%.
Stroke corresponding to the maximum value of the stroke confidence level of each block to be identified is obtained, these strokes are folded
Text corresponding to the character in the boundary rectangle or number are obtained after adding.
Wherein, when carrying out stroke superposition, take common portion as stroke to be used if two strokes have intersection.
The present embodiment, by carrying out the accuracy that stroke segmentation identification is able to ascend Text region to text information.
In one embodiment, the S3 traverses the text in preset sound bank, obtains the nomenclature of drug and described
The corresponding voice to be broadcasted of drug usage information, according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information
Casting it is described wait broadcast voice after, further include the steps that being modified voice broadcast content according to field feedback, have
Body includes:
The feedback information that user terminal is sent is received, problem information characteristic character included in the feedback information is extracted,
The corresponding question attributes of the feedback information are obtained according to described problem information characteristics character;
Wherein, problem information character be primarily referred to as " no ", negative literals or character or the journeys such as " small ", " big " such as " not having "
Spend adverbial word.These characters can indicate that user can not obtain drug content information according to current voice broadcast content.
Question attributes refer to the corresponding attribute of problem information, for example, " sound is small " corresponding question attributes be " volume ",
Corresponding " XX word is unclear " is the different prosodic parameters such as the tone of " XX word ".
The prosodic parameter is adjusted according to described problem attribute, if the prosodic parameter adjusted reaches described
The user terminal demand for including in feedback information, the then prosodic parameter broadcasted prosodic parameter adjusted as medicine information are no
The voice messaging for then obtaining user's input corrects the voice to be broadcasted according to the voice messaging of user input, until reaching
To the demand of user terminal.
Specifically, the demand of user terminal can be evaluated by field feedback, i.e., no longer go out in feedback information
The words such as existing " no ", " big " then think that broadcasting voice meets user demand.It can be targetedly right according to different problems attribute
A certain item parameter in prosodic parameter is adjusted.But there may be the situations of inaccuracy for the feedback information of user, at this time
The own voices for needing to acquire user's input, then according to the prosodic parameter of user's own voices to the medicine information voice of casting
It is modified.
The present embodiment, by user to the feedback of drug voice broadcast content, to casting prosodic parameter be adjusted to
The speech habits of different user are more in line with, obtain correct medicine information convenient for user.
In one embodiment, word casting in the S303, the acquisition nomenclature of drug and the drug usage information
Historical data, original prosodic parameter is obtained according to the historical data, it is described wait broadcast according to the original prosodic parameter casting
Voice is reported, the reflected acoustic wave of the voice to be broadcasted is received, is obtained after correcting the original prosodic parameter according to the reflected acoustic wave
To final prosodic parameter, the voice to be broadcasted is broadcasted according to the final prosodic parameter, comprising:
The historical data for obtaining the casting of the nomenclature of drug and the word in the drug usage information, extracts each
The most pitch of word frequency of occurrence, the duration of a sound and volume carry out the most pitch of frequency of occurrence, the duration of a sound and volume at binaryzation
Reason, forms the original prosodic parameter after splicing;
Specifically, pitch can be placed on to front when pitch, the duration of a sound and volume are spliced into original prosodic parameter,
The duration of a sound can be placed on to front, pitch, the duration of a sound and volume can be carried out to different labels when carrying out binary conversion treatment, so as to
In searching these parameters respectively in original prosodic parameter.
Preset sentence casting speed is obtained, according to the sentence broadcasts speed and the original prosodic parameter is broadcasted
Voice to be broadcasted;
Specifically, preset sentence casting speed is to be counted to obtain according to historical data, it is generally common according to the drug
The case where crowd, sets, for example, the main of certain drug uses crowd for 60 years old or more the elderly, then it is usual to broadcast word speed
It is relatively slow.
The primary reflection wave from the reflected voice to be broadcasted of preset sound reflection wall is received, the original is filtered
Beginning back wave obtains practical reflected acoustic wave, wherein the formula of back wave filtering are as follows:
In formula, NMSE indicates standard deviation, and x (n) indicates that primary reflection wavelength, y (n) indicate estimated value, will be described original anti-
Ejected wave length obtains the practical reflected acoustic wave after making the difference with the standard deviation;
The practical reflected acoustic wave and preset reflected acoustic wave threshold value are made the difference, according to difference and pitch, the duration of a sound and
The corresponding relationship of volume corrects the original prosodic parameter and obtains final prosodic parameter, is broadcasted according to the final prosodic parameter
The voice to be broadcasted.
Specifically, difference is carried out attribute subdivision, i.e. this difference is caused by pitch, the duration of a sound or volume, so
After targetedly corrected.
The present embodiment is effectively adjusted original casting prosodic parameter using reflected acoustic wave, so that medicine information
Casting more meets user demand.
In one embodiment it is proposed that a kind of medicine information identification device, as shown in figure 5, including following module:
Information extraction modules 51 are set as obtaining medicine information image to be identified, from the medicine information image to be identified
In extract text information and digital information;
Information identification module 52 is set as extracting the nomenclature of drug text in the text information, generates drug after combination
Title extracts the usage text information in the text information, and the usage text information and the digital information are carried out
Drug usage information is obtained after combination;
Information broadcasting module 53 is set as traversing the text in preset sound bank, obtains the nomenclature of drug and described
The corresponding voice to be broadcasted of drug usage information, according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information
Broadcast the voice to be broadcasted.
In one embodiment it is proposed that a kind of computer equipment, the computer equipment includes memory and processor,
Computer-readable instruction is stored in memory, when computer-readable instruction is executed by processor, so that processor execution is above-mentioned
The step of medicine information recognition methods in each embodiment.
In one embodiment it is proposed that a kind of storage medium for being stored with computer-readable instruction, this is computer-readable
When instruction is executed by one or more processors, so that one or more processors execute the drug in the various embodiments described above
The step of information identifying method.Wherein, the storage medium can be non-volatile memory medium.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of above-described embodiment is can
It is completed with instructing relevant hardware by program, which can be stored in a computer readable storage medium, storage
Medium may include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random
Access Memory), disk or CD etc..
Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality
It applies all possible combination of the technical characteristic in example to be all described, as long as however, lance is not present in the combination of these technical characteristics
Shield all should be considered as described in this specification.
The some exemplary embodiments of the application above described embodiment only expresses, wherein describe it is more specific and detailed,
But it cannot be understood as the limitations to the application the scope of the patents.It should be pointed out that for the ordinary skill of this field
For personnel, without departing from the concept of this application, various modifications and improvements can be made, these belong to the application
Protection scope.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of medicine information recognition methods characterized by comprising
Medicine information image to be identified is obtained, text information and number letter are extracted from the medicine information image to be identified
Breath;
The nomenclature of drug text in the text information is extracted, generates nomenclature of drug after combination, is extracted in the text information
Usage text information, obtain drug usage information after the usage text information and the digital information are combined;
The text in preset sound bank is traversed, the nomenclature of drug and the corresponding language to be broadcasted of the drug usage information are obtained
Sound broadcasts the voice to be broadcasted according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information.
2. medicine information recognition methods according to claim 1, which is characterized in that described to obtain medicine information figure to be identified
Picture extracts text information and digital information from the medicine information image to be identified, comprising:
When it is any include medicine information images to be recognized to image acquisition device screen distance be less than preset distance threshold
When, the images to be recognized is acquired to obtain original images to be recognized;
Color normalization images to be recognized is obtained after carrying out gray processing processing to the original images to be recognized;
Progressively scanned according to preset character height obtain after the color normalization images to be recognized color normalization to
Identify all characters in image;
The scanning recognition is obtained after the character that scanning recognition goes out is matched with the character information being pre-stored in database to go out
The corresponding text information of character or digital information.
3. medicine information recognition methods according to claim 2, which is characterized in that described to extract in the text information
Nomenclature of drug text generates nomenclature of drug after combination, extracts the usage text information in the text information, by the usage
Text information and the digital information obtain drug usage information after being combined, comprising:
The recognition result of each text in the text information is matched with the word in word library, obtains constituting word
Recognition result;
The nomenclature of drug Feature Words in the recognition result for constituting word are extracted, according to preset nomenclature of drug naming rule,
The nomenclature of drug is obtained after nomenclature of drug Feature Words described in permutation and combination;
Extract the usage text information in the recognition result for constituting word, by the usage text information and digital information into
Several drug usage information sentences are obtained after row combination;
The semantic confidence degree for calculating each drug usage information sentence determines the drug according to the semantic confidence degree
Usage information.
4. medicine information recognition methods according to claim 1, which is characterized in that in the preset sound bank of traversal
Text obtains the nomenclature of drug and the corresponding voice to be broadcasted of the drug usage information, according to the nomenclature of drug and institute
It states the corresponding prosodic parameter of drug usage information and broadcasts the voice to be broadcasted, comprising:
The text in default sound bank is traversed, the nomenclature of drug and the corresponding several original languages of the drug usage information are obtained
Sound;
Obtain user language use information, according to the user language use information determine a certain raw tone be it is described to
Broadcast voice;
The historical data that word is broadcasted in the nomenclature of drug and the drug usage information is obtained, is obtained according to the historical data
To original prosodic parameter, the voice to be broadcasted is broadcasted according to the original prosodic parameter, receives the anti-of the voice to be broadcasted
Sound wave is penetrated, final prosodic parameter is obtained after correcting the original prosodic parameter according to the reflected acoustic wave, according to the final rhythm
It restrains parameter and broadcasts the voice to be broadcasted.
5. medicine information recognition methods according to claim 2, which is characterized in that it is described by scanning recognition go out character with
The character information being pre-stored in database obtained after being matched the scanning recognition go out the corresponding text information of character or
Digital information, comprising:
The connected domain that each stroke of the character is constituted is obtained, determines the boundary rectangle of each connected domain;
The pixel value for obtaining each point in the boundary rectangle, according to the pixel value to shape after boundary rectangle progress piecemeal
At several sub-blocks;
According to the ratio of width to height of preset characters, several sub-blocks are merged into block to be identified;
Compared according to the stroke pixel value of preset character in the pixel value and database of each point in the block to be identified
Compared with acquisition stroke confidence level;
Stroke corresponding to the maximum value of the stroke confidence level of each block to be identified is obtained, after the superposition of these strokes
Obtain text corresponding to the character in the boundary rectangle or number.
6. medicine information recognition methods according to claim 4, which is characterized in that in the preset sound bank of traversal
Text obtains the nomenclature of drug and the corresponding voice to be broadcasted of the drug usage information, according to the nomenclature of drug and institute
State the corresponding prosodic parameter casting of drug usage information it is described wait broadcast voice after, further include according to field feedback to language
The step of sound casting content is modified, specifically includes:
The feedback information that user terminal is sent is received, problem information characteristic character included in the feedback information is extracted, according to
Described problem information characteristics character obtains the corresponding question attributes of the feedback information;
The prosodic parameter is adjusted according to described problem attribute, if the prosodic parameter adjusted reaches the feedback
Otherwise the user terminal demand for including in information, the then prosodic parameter broadcasted prosodic parameter adjusted as medicine information obtain
The voice messaging for taking family input corrects the voice to be broadcasted according to the voice messaging of user input, until reaching use
The demand at family end.
7. medicine information recognition methods according to claim 4, which is characterized in that described to obtain the nomenclature of drug and institute
The historical data that word is broadcasted in drug usage information is stated, original prosodic parameter is obtained according to the historical data, according to described
Original prosodic parameter broadcasts the voice to be broadcasted, and the reflected acoustic wave of the voice to be broadcasted is received, according to the reflected acoustic wave
Final prosodic parameter is obtained after correcting the original prosodic parameter, the language to be broadcasted is broadcasted according to the final prosodic parameter
Sound, comprising:
The historical data for obtaining the casting of the nomenclature of drug and the word in the drug usage information, extracts each word
The most pitch of frequency of occurrence, the duration of a sound and volume are carried out binary conversion treatment by the most pitch of frequency of occurrence, the duration of a sound and volume,
The original prosodic parameter is formed after splicing;
Preset sentence casting speed is obtained, speed is broadcasted according to the sentence and the original prosodic parameter casting is described wait broadcast
Report voice;
The primary reflection wave from the reflected voice to be broadcasted of preset sound reflection wall is received, is filtered described original anti-
Ejected wave obtains practical reflected acoustic wave, wherein the formula of back wave filtering are as follows:
In formula, NMSE indicates standard deviation, and x (n) indicates primary reflection wavelength, and y (n) indicates estimated value, by the primary reflection wave
The long and standard deviation obtains the practical reflected acoustic wave after making the difference;
The practical reflected acoustic wave and preset reflected acoustic wave threshold value are made the difference, according to difference and pitch, the duration of a sound and volume
Corresponding relationship, correct the original prosodic parameter and obtain final prosodic parameter, according to the final prosodic parameter casting
Voice to be broadcasted.
8. a kind of medicine information identification device characterized by comprising
Information extraction modules are set as obtaining medicine information image to be identified, extract from the medicine information image to be identified
Text information and digital information out;
Information identification module is set as extracting the nomenclature of drug text in the text information, and nomenclature of drug is generated after combination, takes out
The usage text information in the text information is taken out, after the usage text information and the digital information are combined
To drug usage information;
Information broadcasting module is set as traversing the text in preset sound bank, obtains the nomenclature of drug and the drug is used
The corresponding voice to be broadcasted of method information broadcasts institute according to the nomenclature of drug and the corresponding prosodic parameter of the drug usage information
State voice to be broadcasted.
9. a kind of computer equipment, which is characterized in that including memory and processor, being stored with computer in the memory can
Reading instruction, when the computer-readable instruction is executed by the processor, so that the processor executes such as claim 1 to 7
Any one of medicine information recognition methods described in claim the step of.
10. a kind of storage medium, which is characterized in that the storage medium is stored with computer-readable instruction, the storage medium
It can be read and write with device processed, when the computer-readable instruction is executed by one or more processors, so that at one or more
Device is managed to execute as described in any one of claims 1 to 7 claim the step of medicine information recognition methods.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910042928.9A CN109920509B (en) | 2019-01-17 | 2019-01-17 | Drug information identification method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910042928.9A CN109920509B (en) | 2019-01-17 | 2019-01-17 | Drug information identification method, device, computer equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109920509A true CN109920509A (en) | 2019-06-21 |
CN109920509B CN109920509B (en) | 2024-05-14 |
Family
ID=66960440
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910042928.9A Active CN109920509B (en) | 2019-01-17 | 2019-01-17 | Drug information identification method, device, computer equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109920509B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112053761A (en) * | 2020-08-12 | 2020-12-08 | 北京左医健康技术有限公司 | Medication guiding method and device based on image recognition and storage medium |
CN112967787A (en) * | 2021-01-28 | 2021-06-15 | 壹健康健康产业(深圳)有限公司 | Medicine information input method, device, medium and terminal equipment |
CN112989974A (en) * | 2021-03-02 | 2021-06-18 | 赵宏福 | Text recognition method and device for automatic word segmentation and spelling and storage medium |
CN113012783A (en) * | 2021-03-18 | 2021-06-22 | 深圳市瑞意博科技股份有限公司 | Medicine rechecking method and device, computer equipment and storage medium |
CN116168376A (en) * | 2023-04-18 | 2023-05-26 | 苏州大学 | Voice broadcast medicine box recognition device and medicine box recognition method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120330665A1 (en) * | 2011-06-03 | 2012-12-27 | Labels That Talk, Ltd | Prescription label reader |
US20170068798A1 (en) * | 2015-09-04 | 2017-03-09 | Walgreen Co. | Automated pharmacy translation engine for prescription medication instructions |
CN106652995A (en) * | 2016-12-31 | 2017-05-10 | 深圳市优必选科技有限公司 | Voice broadcasting method and system for text |
-
2019
- 2019-01-17 CN CN201910042928.9A patent/CN109920509B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120330665A1 (en) * | 2011-06-03 | 2012-12-27 | Labels That Talk, Ltd | Prescription label reader |
US20170068798A1 (en) * | 2015-09-04 | 2017-03-09 | Walgreen Co. | Automated pharmacy translation engine for prescription medication instructions |
CN106652995A (en) * | 2016-12-31 | 2017-05-10 | 深圳市优必选科技有限公司 | Voice broadcasting method and system for text |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112053761A (en) * | 2020-08-12 | 2020-12-08 | 北京左医健康技术有限公司 | Medication guiding method and device based on image recognition and storage medium |
CN112967787A (en) * | 2021-01-28 | 2021-06-15 | 壹健康健康产业(深圳)有限公司 | Medicine information input method, device, medium and terminal equipment |
CN112989974A (en) * | 2021-03-02 | 2021-06-18 | 赵宏福 | Text recognition method and device for automatic word segmentation and spelling and storage medium |
CN113012783A (en) * | 2021-03-18 | 2021-06-22 | 深圳市瑞意博科技股份有限公司 | Medicine rechecking method and device, computer equipment and storage medium |
CN116168376A (en) * | 2023-04-18 | 2023-05-26 | 苏州大学 | Voice broadcast medicine box recognition device and medicine box recognition method |
Also Published As
Publication number | Publication date |
---|---|
CN109920509B (en) | 2024-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109920509A (en) | Medicine information recognition methods, device, computer equipment and storage medium | |
US11449556B2 (en) | Responding to user queries by context-based intelligent agents | |
CN106874643A (en) | Build the method and system that knowledge base realizes assisting in diagnosis and treatment automatically based on term vector | |
CN109472207A (en) | Emotion identification method, apparatus, equipment and storage medium | |
WO2021114841A1 (en) | User report generating method and terminal device | |
Yu et al. | Searching for category-consistent features: A computational approach to understanding visual category representation | |
CN114898861A (en) | Multi-modal depression detection method and system based on full attention mechanism | |
Abdusalomov et al. | Improved feature parameter extraction from speech signals using machine learning algorithm | |
Mulder et al. | Tense and aspect in a Spanish literary work and its translations | |
CN113409771B (en) | Detection method for forged audio frequency, detection system and storage medium thereof | |
CN117493504A (en) | Medical event extraction method based on generated pre-training language model | |
Barotto | Exemplification and categorization: the case of Japanese | |
CN110147791A (en) | Character recognition method, device, equipment and storage medium | |
Bulut et al. | Comprehension of Turkish relative clauses: Evidence from eye-tracking and corpus analysis | |
Daland | Word segmentation, word recognition, and word learning: A computational model of first language acquisition | |
CN111599363B (en) | Voice recognition method and device | |
Xu et al. | Gabor based lipreading with a new audiovisual mandarin corpus | |
Dose | Patterns of growing standardisation and interference in interpreted German discourse | |
CN116842168B (en) | Cross-domain problem processing method and device, electronic equipment and storage medium | |
Choudhury et al. | Relevance of Language-Specific Training on Image Caption Synthesis for Low Resource Assamese Language | |
US11960840B2 (en) | Selecting a meaning of a word of a phrase | |
Gimeno-Gómez et al. | Comparing speaker adaptation methods for visual speech recognition for continuous spanish | |
WO2023238722A1 (en) | Information creation method, information creation device, and moving picture file | |
US20220261538A1 (en) | Skipping natural language processor | |
US20220383148A1 (en) | Processing a contradiction in a knowledge database |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |