CN110338748A - Quickly method, storage medium, terminal and the sight tester of positioning eyesight value - Google Patents
Quickly method, storage medium, terminal and the sight tester of positioning eyesight value Download PDFInfo
- Publication number
- CN110338748A CN110338748A CN201910513319.7A CN201910513319A CN110338748A CN 110338748 A CN110338748 A CN 110338748A CN 201910513319 A CN201910513319 A CN 201910513319A CN 110338748 A CN110338748 A CN 110338748A
- Authority
- CN
- China
- Prior art keywords
- information
- sighting target
- data information
- matrix data
- training
- 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
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B3/00—Apparatus for testing the eyes; Instruments for examining the eyes
- A61B3/02—Subjective types, i.e. testing apparatus requiring the active assistance of the patient
- A61B3/028—Subjective types, i.e. testing apparatus requiring the active assistance of the patient for testing visual acuity; for determination of refraction, e.g. phoropters
- A61B3/032—Devices for presenting test symbols or characters, e.g. test chart projectors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Animal Behavior & Ethology (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Ophthalmology & Optometry (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Surgery (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Heart & Thoracic Surgery (AREA)
- Eye Examination Apparatus (AREA)
Abstract
The invention discloses method, storage medium, terminal and the sight testers of a kind of quickly positioning eyesight value；It solves and indicates that sighting target could be completed to detect step by step, the low problem of whole detection efficiency, its key points of the technical solution are that, the current triggering information of active user is obtained, the triggering information includes that sighting target judges information；Sighting target is judged that information is compared to each other analysis with corresponding display sighting target information to form judging result information；It calls trained neural network model in advance and the current sighting target at current time is judged that information and current display sighting target information feed back to the neural network model to analyze to be formed and show sighting target information next time, the present invention shows sighting target information according to the neural network model obtained by training to be formed next time, it avoids successively showing that sighting target allows examinee to judge step by step as far as possible, improves whole detection efficiency.
Description
Technical field
The present invention relates to sight testers, in particular to quickly position method, storage medium, terminal and the eyesight of eyesight value
Detector.
Background technique
Most traditional and common eyesight detection device has lamp box visual chart, eyesight projector, comprehensive optometry instrument, these tradition
Eyesight detection device, drawback is: one, the sighting target in visual chart and its arrangement mode are fixed, and ordinary people is through too short
A possibility that observation of time can write down sighting target and its arrangement mode, and there are cheatings；Two, medical staff is needed aside
Sighting target is given directions using indicating bar, the sighting target opening direction seen is fed back to medical staff by examinee, using medical worker
Multilevel iudge, finally obtains examinee's eyesight value, this brings no small workload to medical staff, also difficult in the detection process
Exempt to malfunction, or even in order to complete task as early as possible, surveys several sighting targets less and deal with and get over, medical staff is to the judgment criteria that gives a test of one's eyesight
Difference, certain influence is also brought to the accuracy of testing result.
With the emergence and development of electronic visual chart, self-service eye test gradually becomes generally, when testing eyesight
It only needs tested personnel oneself that operation can be completed, greatly reduces human cost, examinee is facilitated oneself to check at any time.
After usually there is sighting target on visual chart, examinee indicates direction by pressing the key on remote controler, if same shelves sighting target
By multiple instruction it is correct after, as soon as then sighting target becomes smaller grade, otherwise become larger one grade.
But this mode inefficiency, it needs to complete the detection of examinee's eyesight by multiple instruction, influences to use
Family experience, especially tested personnel are more, when time requirement is stringenter, influence the completion of inspection task.
Summary of the invention
The first object of the present invention is to provide a kind of method of quickly positioning eyesight value, is capable of fast fast reading positioning examinee's
Eyesight value is to improve detection efficiency.
Above-mentioned technical purpose of the invention has the technical scheme that
A kind of method of quick positioning eyesight value, comprising:
The current triggering information of active user is obtained, the triggering information includes that sighting target judges information；
Sighting target is judged that information is compared to each other analysis with corresponding display sighting target information to form judging result information；
It calls trained neural network model in advance and the current sighting target at current time is judged into information and current display sighting target
Information feeds back to the neural network model to analyze to be formed and show sighting target information next time.
Using the above scheme, according to trained neural network model, corresponding parameter factors are input to neural network
In model, i.e., current sighting target is judged that information and current display sighting target information feed back to the neural network model, be based on nerve net
The training learning process of network model more can accurately feed back one and show sighting target information next time, so that this shows next time
Show that sighting target information can substantially reduce the number of test relatively close in the eyesight value of current examinee, improves detection effect
Rate.
Preferably, as follows about the method analyzed by neural network model:
It obtains current sighting target and judges information and current display sighting target information；And current sighting target is judged into information and current display
The corresponding data of sighting target information difference form relation factor matrix data information；
According to default activation primitive to carry out data processing to relation factor matrix data information to form level-one matrix function
It is believed that breath；Random zero setting is carried out to form diode matrix data information to level-one matrix data information；
According to default activation primitive to carry out data processing to diode matrix data information to form three-level matrix data letter
Breath；
According to default normalization exponential function to carry out data processing to three-level matrix data information to form current display
Sighting target information.
Using the above scheme, by the setting of multilayer neural network model, so that neural network model is in the training process,
The study that can data fill with part rejects some external influence factors as far as possible to improve finally formed current display view
Mark the accuracy of information；The current display sighting target information with the eyesight value most proximity of current examinee is obtained by prediction, is reduced
The number of test improves testing efficiency.
Preferably, as follows about the method that activation primitive carries out data processing to relation factor matrix data information:
The the first weight matrix data information and relation factor matrix data information for being obtained training according to Sigmoid function carry out
Mapping is to form the first primary mapping data information；
The the second weight matrix data information and relation factor matrix data information for being obtained training according to Sigmoid function carry out
Mapping is to form the second primary mapping data information；
Second primary mapping data information and last display sighting target information are subjected to data processing to weaken last display and regard
Mark information simultaneously forms reduction display sighting target information；
Reduction display sighting target information and sighting target are judged that the corresponding data of information difference form reduction matrix data information；
The third weight matrix data information of acquisition will will be trained to map with reduction matrix data information according to tanh function
To form internal output data information；
It is aobvious to the last time being sequentially allocated different weights according to weighted data corresponding to the first primary mapping data information
Show sighting target information and internal output data information and is formed when previous stage matrix data information and/or three-level matrix data information.
Preferably, forming the first primary mapping data information and the second primary mapping data by Sigmoid function
During information, by default random zero setting matrix data information respectively to the first weight matrix data information and second
After weight matrix data information is handled, then mapped with relation factor matrix data information.
Preferably, being formed when the formula of previous stage matrix data information and/or three-level matrix data information is as follows:
Zt=σ (M*Wz*[h(t-1),x(t-1)])；
rt=σ (M*Wr*[h(t-1),x(t-1)])；
ht'=tanh (M*W* [rt*h(t-1),x(t-1)])；
ht=(1-Zt)*h(t-1)+Zt*ht′；
Wherein, ZtFor the first primary mapping data information；
rtFor the second primary mapping data information；
ht' it is internal output data information；
htFor the display sighting target information of t moment；
WzFor training the first weight matrix data information obtained；
WrFor training the second weight matrix data information obtained；
W is training third weight matrix data information obtained；
h(t-1)For the display sighting target information at t-1 moment；
x(t-1)Sighting target for the t-1 moment judges information；
M is random zero setting matrix data information.
Using the above scheme, finally formed during training neural network model is exactly weight to different data
Proportion, can form three weighted datas according to different needs in the training process, in data processing, by using
Sigmoid function and tanh function complete the mapping of data, and then export the data that processing is completed, and wait subsequent processing；
The the first weight matrix data information and the second weight matrix data letter that training is obtained by random zero setting matrix data information
Breath carries out random zero setting to keep the diversity of data, avoids training data from belonging to same type and causes to train obtained knot
Fruit excessively tends to identity, improves the accuracy of Neural Network model predictive.
Preferably, the learning method about neural network model is as follows:
Required training data information is generated according to the predicted method by half of visual chart；
It obtains training data information and the filtering of invalid data is carried out to form filtering data information to training data information；It is described
Filtering data information includes current sighting target training of judgement information and current display sighting target training information；
Obtain current sighting target training of judgement information and current display sighting target training information；And by current sighting target training of judgement information
And data corresponding to current display sighting target training information difference form relation factor matrix data information；
According to default activation primitive to carry out data processing to relation factor matrix data information to form level-one matrix function
It is believed that breath；Random zero setting is carried out to form diode matrix data information to level-one matrix data information；
According to default activation primitive to carry out data processing to diode matrix data information to form three-level matrix data letter
Breath；
According to default normalization exponential function to carry out data processing to three-level matrix data information to form current display
Sighting target training information.
Using the above scheme, predicted method belongs to a kind of method more positioned rapidly by half, but it is still not accurate enough and
And data formed it is excessively single, therefore using this kind of form as trained data, by these training datas come to neural network into
Row training, so that, based on the principle of predicted method by half, being allowed after training by neural network model during final prediction
The accuracy of detection is higher.
Preferably, as follows about the method that training data information is formed:
Define i-th inspection is recorded as (ti, si, pi), tiThe time interval in direction, s are indicated to there is sighting target to examineei
For eyesight value and correspond to display sighting target information, piFor instruction correctness and correspond to judging result information；
If piCorrectly, then the predictive visual acuity value that i+1 time checks are as follows:
If piMistake, then the predictive visual acuity value that i+1 time checks are as follows:
The detection of institute's preset times is defined through as primary complete detection；
If there is each p in the detection process of institute's preset timesiIt is mistake, then by all detection records as training
Data information；
If in the detection process of institute's preset times, wherein arbitrarily once there is piTo be correct, then by residue without detection
Data are supplied at random and detection record after testing are combined to be used as training data information.
Using the above scheme, the number of detection is defined, and not the obtained data of foot time detection is mended at random
Foot avoids computer from being instructed in reading data to ensure that computer can read relevant data smoothly to be corresponded to
During practicing neural network model, the case where misprogramming occur and be unable to run, stability is improved.
The second object of the present invention is to provide a kind of computer readable storage medium, can store corresponding program, can
The eyesight value of fast fast reading positioning examinee is to improve detection efficiency.
Above-mentioned technical purpose of the invention has the technical scheme that
A kind of computer readable storage medium, including that can be realized as described in the claims when processor load and execution
The quickly program of the method for positioning eyesight value.
Using the above scheme, according to trained neural network model, corresponding parameter factors are input to neural network
In model, i.e., current sighting target is judged that information and current display sighting target information feed back to the neural network model, be based on nerve net
The training learning process of network model more can accurately feed back one and show sighting target information next time, so that this shows next time
Show that sighting target information can substantially reduce the number of test relatively close in the eyesight value of current examinee, improves detection effect
Rate.
The third object of the present invention is to provide a kind of terminal, is capable of the eyesight value of fast fast reading positioning examinee to improve detection
Efficiency.
Above-mentioned technical purpose of the invention has the technical scheme that
A kind of terminal including memory, processor and is stored in the journey that can be run on the memory and on the processor
Sequence, which can be realized the quick positioning eyesight value as described in the claims method when processor load and execution.
Using the above scheme, according to trained neural network model, corresponding parameter factors are input to neural network
In model, i.e., current sighting target is judged that information and current display sighting target information feed back to the neural network model, be based on nerve net
The training learning process of network model more can accurately feed back one and show sighting target information next time, so that this shows next time
Show that sighting target information can substantially reduce the number of test relatively close in the eyesight value of current examinee, improves detection effect
Rate.
The fourth object of the present invention is to provide a kind of sight tester, is capable of the eyesight value of fast fast reading positioning examinee to mention
High detection efficiency.
Above-mentioned technical purpose of the invention has the technical scheme that
A kind of sight tester, including memory, processor and be stored on the memory and can transport on the processor
Capable program, the program can be by quick positioning eyesight value of the realization as described in the claims when processor load and execution
Method.
Using the above scheme, according to trained neural network model, corresponding parameter factors are input to neural network
In model, i.e., current sighting target is judged that information and current display sighting target information feed back to the neural network model, be based on nerve net
The training learning process of network model more can accurately feed back one and show sighting target information next time, so that this shows next time
Show that sighting target information can substantially reduce the number of test relatively close in the eyesight value of current examinee, improves detection effect
Rate.
In conclusion the invention has the following advantages: being formed according to the neural network model obtained by training
Sighting target information is shown next time, is avoided successively showing that sighting target allows examinee to judge step by step as far as possible, is improved whole detection efficiency.
Detailed description of the invention
Fig. 1 is the flow diagram of the quickly method of positioning eyesight value；
Fig. 2 is the flow diagram for the method analyzed by neural network model；
Fig. 3 is the flow diagram for carrying out the method for data processing to relation factor matrix data information about activation primitive；
Fig. 4 is the flow diagram of the learning method about neural network model.
Specific embodiment
Below in conjunction with attached drawing, invention is further described in detail.
This specific embodiment is only explanation of the invention, is not limitation of the present invention, those skilled in the art
Member can according to need the modification that not creative contribution is made to the present embodiment after reading this specification, but as long as at this
All by the protection of Patent Law in the scope of the claims of invention.
The embodiment of the present invention provides a kind of method of quickly positioning eyesight value, comprising: obtains the current triggering of active user
Information, the triggering information includes that sighting target judges information；Sighting target is judged that information is compared to each other with corresponding display sighting target information
Analysis is to form judging result information；It calls trained neural network model in advance and judges the current sighting target at current time
Information and current display sighting target information feed back to the neural network model to analyze to be formed and show sighting target information next time.
In the embodiment of the present invention, according to trained neural network model, corresponding parameter factors are input to nerve net
In network model, i.e., current sighting target is judged that information and current display sighting target information feed back to the neural network model, based on nerve
The training learning process of network model more can accurately feed back one and show sighting target information next time, so that this is next time
Show that sighting target information can substantially reduce the number of test, improve detection relatively close in the eyesight value of current examinee
Efficiency.
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
In addition, the terms "and/or", only a kind of incidence relation for describing affiliated partner, indicates may exist
Three kinds of relationships, for example, A and/or B, can indicate: individualism A exists simultaneously A and B, these three situations of individualism B.Separately
Outside, character "/" herein typicallys represent the relationship that forward-backward correlation object is a kind of "or" unless otherwise specified.
The embodiment of the present invention is described in further detail with reference to the accompanying drawings of the specification.
The embodiment of the present invention provides a kind of method of quickly positioning eyesight value, and the main flow of the method is described as follows.
As shown in Figure 1:
Step 1000: obtaining the current triggering information of active user, the triggering information includes that sighting target judges information.
Wherein, currently obtaining for triggering information can obtain in such a way that mechanical key triggers, can also be by virtual
The mode of key triggering obtains；The mode of mechanical key triggering, can be by pressing default mechanical key, for example, remote control
Key etc. up and down on key up and down, keyboard in device, after can also choosing corresponding region by using mouse
By clicking mouse to obtain current triggering information；In the present embodiment, it is preferred to use the form of remote controler obtains current triggering letter
Breath；The mode of virtual key triggering, can be by pressing relevant virtual triggering key in the interface of corresponding software to realize
It obtains.
Step 2000: sighting target is judged that information is compared to each other analysis with corresponding display sighting target information to form judging result
Information.
Wherein, differentiated according to preset decision procedure, i.e. the sighting target of examinee's feedback judges information and current display
Whether sighting target information is identical, if they are the same, then illustrates that examinee can see current sighting target, if not identical, illustrates tested
Person can not see current sighting target clearly.
Step 3000: call in advance trained neural network model and by the current sighting target at current time judge information and
Current display sighting target information feeds back to the neural network model to analyze to be formed and show sighting target information next time.
Wherein, according to trained neural network model, corresponding parameter factors are input in neural network model, i.e.,
Current sighting target is judged that information and current display sighting target information feed back to the neural network model, the instruction based on neural network model
Practice learning process, more can accurately feed back one and show sighting target information next time, so that this shows sighting target information next time
It can be relatively close in the eyesight value of current examinee.
As shown in Fig. 2, as follows about the method analyzed by neural network model:
Step 3100: obtaining current sighting target and judge information and current display sighting target information；And by current sighting target judge information with
And data corresponding to current display sighting target information difference form relation factor matrix data information.
Wherein, current sighting target judges that information is that current examinee is sentenced by one that the sighting target of observation display is fed back later
It is disconnected；And currently show the sighting target that sighting target information is as shown on the display apparatus.Phase is formed after the two factors are associated
The relation factor matrix data information of pass；It can be set as [h(t-1),x(t-1)], h(t-1)Believe for the display sighting target at t-1 moment
Breath；x(t-1)Sighting target for the t-1 moment judges information.
Step 3200: according to default activation primitive to carry out data processing to relation factor matrix data information with shape
At level-one matrix data information.
Wherein, as shown in figure 3, carrying out the method for data processing such as to relation factor matrix data information about activation primitive
Under:
Step 3210: the first weight matrix data information and relation factor matrix function for being obtained training according to Sigmoid function
It is believed that breath is mapped to form the first primary mapping data information.
Step 3220: the second weight matrix data information and relation factor square for being obtained training according to Sigmoid function
Battle array data information is mapped to form the second primary mapping data information.
Wherein, Sigmoid function is a common S type function in biology, also referred to as S sigmoid growth curve.Believing
In breath science, since singly properties, the Sigmoid function such as increasing and the increasing of inverse function list are often used as the threshold value letter of neural network for it
Number, by variable mappings to 0, between 1.
In the process for forming the first primary mapping data information and the second primary mapping data information by Sigmoid function
In, by default random zero setting matrix data information respectively to the first weight matrix data information and the second weight matrix number
It is believed that after breath is handled, then mapped with relation factor matrix data information.
Relevant processing formula is as follows:
Zt=σ (M*Wz*[h(t-1),x(t-1)])。
rt=σ (M*Wr*[h(t-1),x(t-1)])。
Wherein, ZtFor the first primary mapping data information；
rtFor the second primary mapping data information；
σ is Sigmoid function；
M is random zero setting matrix data information；
WzFor training the first weight matrix data information obtained；
WrFor training the second weight matrix data information obtained；
h(t-1)For the display sighting target information at t-1 moment；
x(t-1)Sighting target for the t-1 moment judges information.
Step 3230: the second primary mapping data information and last time display sighting target information are subjected to data processing to weaken
Last time display sighting target information simultaneously forms reduction display sighting target information.
Step 3240: reduction display sighting target information and sighting target are judged that the corresponding data of information difference form reduction square
Battle array data information.
Step 3250: the third weight matrix data information and reduction matrix data of acquisition will will be trained according to tanh function
Information is mapped to form internal output data information.
Wherein, through the above steps obtained in the second primary mapping data information to show sighting target information to the last time
It is weakened, is forming corresponding reduction matrix data information after completing preliminary data processing, it can be set as [rt*
h(t-1),x(t-1)]。
Tanh function is one in hyperbolic functions, and tanh () is tanh.In mathematics, tanh " tanh " is
From by hyperbolic sine and hyperbolic cosine, both basic hyperbolic functions are derived.Function: y=tanh x；Domain: R, codomain:
(-1,1).Y=tanh x is an odd function, and functional image was origin and the strictly monotone increasing for passing through I, III quadrant
Curve, image are limited between two horizontal asymptote y=1 and y=-1.
Relevant processing formula is as follows:
ht'=tanh (M*W* [rt*h(t-1),x(t-1)])；
Wherein, ht' it is internal output data information；
htFor the display sighting target information of t moment；
M is random zero setting matrix data information；
W is training third weight matrix data information obtained；
rtFor the second primary mapping data information；
h(t-1)For the display sighting target information at t-1 moment；
x(t-1)Sighting target for the t-1 moment judges information.
Step 3260: according to weighted data corresponding to the first primary mapping data information successively to divide different weights
The dispensing last time, which shows sighting target information and internal output data information and formed, works as previous stage matrix data information and/or third order moment
Battle array data information.
Wherein, the level-one matrix data information required for it can be formed through the above steps, can also form following
Required three-level matrix data information in step.
Relevant processing formula is as follows:
ht=(1-Zt)*h(t-1)+Zt*ht′；
htFor the display sighting target information of t moment；
ZtFor the first primary mapping data information；
h(t-1)For the display sighting target information at t-1 moment；
ht' it is internal output data information.
Finally formed during training neural network model is exactly to match to the weight of different data, is being trained
Cheng Zhonghui forms three weighted datas according to different needs, in data processing, by using Sigmoid function with
And tanh function completes the mapping of data, and then exports the data that processing is completed, and waits subsequent processing；Pass through random zero setting square
Battle array data information carries out random zero setting to trained the first obtained weight matrix data information and the second weight matrix data information
To keep the diversity of data, the result for avoiding training data from belonging to same type and training being caused to obtain excessively tends to be same
Property, improve the accuracy of Neural Network model predictive.
Step 3300: random zero setting is carried out to form diode matrix data information to level-one matrix data information.
Step 3400: according to default activation primitive to carry out data processing to diode matrix data information to form three
Grade matrix data information.
Wherein, which can be obtained by the step identical as level-one matrix data information is obtained
It takes, can also be obtained by other activation primitives.
Step 3500: according to default normalization exponential function with to three-level matrix data information carry out data processing with
Form current display sighting target information.
Wherein, normalization indicator function uses softmax function, is a kind of popularization of logical function.It can contain one
The K dimensional vector of any real number is tieed up in real vector " compressed " to another K, so that the range of each element is between, and
All elements and be 1.
By the setting of multilayer neural network model so that neural network model is in the training process, can to data into
Row fills the study of part, rejects some external influence factors as far as possible to improve the accurate of finally formed current display sighting target information
Property；It obtains reducing the number of test with the current display sighting target information of the eyesight value most proximity of current examinee, mentioning by prediction
High testing efficiency.
As shown in figure 4, the learning method about neural network model is as follows:
Step 4100: required training data information is generated according to the predicted method by half of visual chart.
Wherein, the method about the formation of training data information is as follows:
Define i-th inspection is recorded as (ti, si, pi), tiThe time interval in direction, s are indicated to there is sighting target to examineei
For eyesight value and correspond to display sighting target information, piFor instruction correctness and correspond to judging result information；
If piCorrectly, then the predictive visual acuity value that i+1 time checks are as follows:
If piMistake, then the predictive visual acuity value that i+1 time checks are as follows:
The detection of institute's preset times is defined through as primary complete detection；
If there is each p in the detection process of institute's preset timesiIt is mistake, then by all detection records as training
Data information；
If in the detection process of institute's preset times, wherein arbitrarily once there is piTo be correct, then by residue without detection
Data are supplied at random and detection record after testing are combined to be used as training data information.
The number of detection is defined, and not the obtained data of foot time detection are supplied at random, to ensure to calculate
Machine can read relevant data smoothly to be corresponded to, and computer is avoided to be trained neural network model in reading data
During, the case where misprogramming occur and be unable to run, improve stability.
Step 4200: obtaining training data information and the filtering of invalid data is carried out to form filtering to training data information
Data information；The filtering data information includes current sighting target training of judgement information and current display sighting target training information；
Step 4300: obtaining current sighting target training of judgement information and current display sighting target training information；And current sighting target is sentenced
Data corresponding to disconnected training information and current display sighting target training information difference form relation factor matrix data information；
Step 4400: according to default activation primitive to carry out data processing to relation factor matrix data information to form one
Grade matrix data information；
Step 4500: random zero setting is carried out to form diode matrix data information to level-one matrix data information；
Step 4600: according to default activation primitive to carry out data processing to diode matrix data information to form third order moment
Battle array data information；
Step 4700: according to default normalization exponential function to carry out data processing to three-level matrix data information to be formed
Current display sighting target training information.
Predicted method belongs to a kind of method more positioned rapidly by half, but still not accurate enough and data are formed excessively
It is single, therefore using this kind of form as the data of training, neural network is trained by these training datas, so that most
Eventually prediction during, based on the principle of predicted method by half, allowed after training by neural network model detection accuracy more
It is high.
The embodiment of the present invention provides a kind of computer readable storage medium, including can be by processor load and execution when realize
Such as Fig. 1-Fig. 4.Each step described in process.
The computer readable storage medium for example, USB flash disk, mobile hard disk, read-only memory (Read-Only
Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. is various to deposit
Store up the medium of program code.
Based on the same inventive concept, the embodiment of the present invention provides a kind of terminal, including memory, processor and is stored in institute
The program that can be run on memory and on the processor is stated, which can be realized such as Fig. 1-when processor load and execution
Fig. 4.The method of quick positioning eyesight value described in process.
Based on the same inventive concept, the embodiment of the present invention provides a kind of sight tester, including memory, processor and deposits
The program that can be run on the memory and on the processor is stored up, which realizes when can be by processor load and execution
Such as Fig. 1-Fig. 4.The method of quick positioning eyesight value described in process.
It is apparent to those skilled in the art that for convenience and simplicity of description, only with above-mentioned each function
The division progress of module can according to need and for example, in practical application by above-mentioned function distribution by different function moulds
Block is completed, i.e., the internal structure of device is divided into different functional modules, to complete all or part of function described above
Energy.The specific work process of the system, apparatus, and unit of foregoing description, can be with reference to corresponding in preceding method embodiment
Journey, details are not described herein.
In several embodiments provided herein, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the module or
The division of unit, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units
Or component can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, institute
Display or the mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, device or unit
Indirect coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, the technical solution of the application is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
It is each that equipment (can be personal computer, server or the network equipment etc.) or processor (processor) execute the application
The all or part of the steps of embodiment the method.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory,
The various media that can store program code such as random access memory, magnetic or disk.
The above, above embodiments are only described in detail to the technical solution to the application, but the above implementation
The explanation of example is merely used to help understand method and its core concept of the invention, should not be construed as limiting the invention.This
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by those skilled in the art, should all cover
Within protection scope of the present invention.
Claims (10)
1. a kind of method of quickly positioning eyesight value, characterized in that include:
The current triggering information of active user is obtained, the triggering information includes that sighting target judges information；
Sighting target is judged that information is compared to each other analysis with corresponding display sighting target information to form judging result information；
It calls trained neural network model in advance and the current sighting target at current time is judged into information and current display sighting target
Information feeds back to the neural network model to analyze to be formed and show sighting target information next time.
2. the method for quick positioning eyesight value according to claim 1, characterized in that about by neural network model into
The method of row analysis is as follows:
It obtains current sighting target and judges information and current display sighting target information；And current sighting target is judged into information and current display
The corresponding data of sighting target information difference form relation factor matrix data information；
According to default activation primitive to carry out data processing to relation factor matrix data information to form level-one matrix function
It is believed that breath；
Random zero setting is carried out to form diode matrix data information to level-one matrix data information；
According to default activation primitive to carry out data processing to diode matrix data information to form three-level matrix data letter
Breath；
According to default normalization exponential function to carry out data processing to three-level matrix data information to form current display
Sighting target information.
3. the method for quick positioning eyesight value according to claim 2, characterized in that about activation primitive to relation factor
The method that matrix data information carries out data processing is as follows:
The the first weight matrix data information and relation factor matrix data information for being obtained training according to Sigmoid function carry out
Mapping is to form the first primary mapping data information；
The the second weight matrix data information and relation factor matrix data information for being obtained training according to Sigmoid function carry out
Mapping is to form the second primary mapping data information；
Second primary mapping data information and last display sighting target information are subjected to data processing to weaken last display and regard
Mark information simultaneously forms reduction display sighting target information；
Reduction display sighting target information and sighting target are judged that the corresponding data of information difference form reduction matrix data information；
The third weight matrix data information of acquisition will will be trained to map with reduction matrix data information according to tanh function
To form internal output data information；
It is aobvious to the last time being sequentially allocated different weights according to weighted data corresponding to the first primary mapping data information
Show sighting target information and internal output data information and is formed when previous stage matrix data information and/or three-level matrix data information.
4. the method for quick positioning eyesight value according to claim 3, characterized in that formed by Sigmoid function
During first primary mapping data information and the second primary mapping data information, pass through default random zero setting matrix function
It is believed that breath the first weight matrix data information and the second weight matrix data information are handled respectively after, then with relation factor
Matrix data information is mapped.
5. the method for quick positioning eyesight value according to claim 4, characterized in that formed when previous stage matrix data is believed
The formula of breath and/or three-level matrix data information is as follows:
Zt=σ (M*Wz*[h(t-1),x(t-1)])；
rt=σ (M*Wr*[h(t-1),x(t-1)])；
ht'=tanh (M*W* [rt*h(t-1),x(t-1)])；
ht=(1-Zt)*h(t-1)+Zt*ht′；
Wherein, ZtFor the first primary mapping data information；
rtFor the second primary mapping data information；
ht' it is internal output data information；
htFor the display sighting target information of t moment；
WzFor training the first weight matrix data information obtained；
WrFor training the second weight matrix data information obtained；
W is training third weight matrix data information obtained；
h(t-1)For the display sighting target information at t-1 moment；
x(t-1)Sighting target for the t-1 moment judges information；
M is random zero setting matrix data information.
6. the method for quick positioning eyesight value according to claim 1, characterized in that
Learning method about neural network model is as follows:
Required training data information is generated according to the predicted method by half of visual chart；
It obtains training data information and the filtering of invalid data is carried out to form filtering data information to training data information；It is described
Filtering data information includes current sighting target training of judgement information and current display sighting target training information；
Obtain current sighting target training of judgement information and current display sighting target training information；And by current sighting target training of judgement information
And data corresponding to current display sighting target training information difference form relation factor matrix data information；
According to default activation primitive to carry out data processing to relation factor matrix data information to form level-one matrix function
It is believed that breath；Random zero setting is carried out to form diode matrix data information to level-one matrix data information；
According to default activation primitive to carry out data processing to diode matrix data information to form three-level matrix data letter
Breath；
According to default normalization exponential function to carry out data processing to three-level matrix data information to form current display
Sighting target training information.
7. the method for quick positioning eyesight value according to claim 6, characterized in that
The method formed about training data information is as follows:
Define i-th inspection is recorded as (ti, si, pi), tiThe time interval in direction, s are indicated to there is sighting target to examineeiFor
Eyesight value and correspond to display sighting target information, piFor instruction correctness and correspond to judging result information；
If piCorrectly, then the predictive visual acuity value that i+1 time checks are as follows:
If piMistake, then the predictive visual acuity value that i+1 time checks are as follows:
The detection of institute's preset times is defined through as primary complete detection；
If there is each p in the detection process of institute's preset timesiIt is mistake, then by all detection records as training number
It is believed that breath；
If in the detection process of institute's preset times, wherein arbitrarily once there is piTo be correct, then by residue without the number of detection
Training data information is used as according to being supplied and being recorded at random in conjunction with detection after testing.
8. a kind of computer readable storage medium, characterized in that including that can be realized when processor load and execution as right is wanted
The program of the method for quick positioning eyesight value described in asking any one of 1 to 7.
9. a kind of terminal, characterized in that including memory, processor and be stored on the memory and can be in the processor
The program of upper operation, the program can be realized quick as described in any one of claims 1 to 7 when processor load and execution
The method for positioning eyesight value.
10. a kind of sight tester, it is characterized in that: including memory, processor and being stored on the memory and can be in institute
The program run on processor is stated, which can be realized when processor load and execution such as any one of claims 1 to 7 institute
The method for the quick positioning eyesight value stated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910513319.7A CN110338748B (en) | 2019-06-13 | 2019-06-13 | Method for quickly positioning vision value, storage medium, terminal and vision detector |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910513319.7A CN110338748B (en) | 2019-06-13 | 2019-06-13 | Method for quickly positioning vision value, storage medium, terminal and vision detector |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110338748A true CN110338748A (en) | 2019-10-18 |
| CN110338748B CN110338748B (en) | 2022-03-08 |
Family
ID=68181976
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910513319.7A Active CN110338748B (en) | 2019-06-13 | 2019-06-13 | Method for quickly positioning vision value, storage medium, terminal and vision detector |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110338748B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111265182A (en) * | 2020-01-21 | 2020-06-12 | 李小丹 | AI remote optometry service platform and optometry equipment |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2002039754A1 (en) * | 2000-11-08 | 2002-05-16 | Andrzej Czyzewski | Visual screening tests by means of computers |
| CN102599879A (en) * | 2012-02-23 | 2012-07-25 | 天津理工大学 | Self-adaptive eyesight test intelligent system and eyesight test method |
| US20140211164A1 (en) * | 2013-01-31 | 2014-07-31 | Nidek Co., Ltd. | Optometric apparatus |
| US9277857B1 (en) * | 2014-11-06 | 2016-03-08 | Bertec Corporation | System for testing and/or training the vision of a subject |
| CN106060142A (en) * | 2016-06-17 | 2016-10-26 | 杨斌 | Mobile phone capable of checking eyesight, and method for checking eyesight by using mobile phone |
| CN106537290A (en) * | 2014-05-09 | 2017-03-22 | 谷歌公司 | Systems and methods for biomechanically-based eye signals for interacting with real and virtual objects |
| CN106778597A (en) * | 2016-12-12 | 2017-05-31 | 朱明� | Intellectual vision measurer based on graphical analysis |
| CN107198505A (en) * | 2017-04-07 | 2017-09-26 | 天津市天中依脉科技开发有限公司 | Visual function detecting system and method based on smart mobile phone |
| CN107358036A (en) * | 2017-06-30 | 2017-11-17 | 北京机器之声科技有限公司 | A kind of child myopia Risk Forecast Method, apparatus and system |
| CN107411700A (en) * | 2017-04-07 | 2017-12-01 | 天津大学 | A kind of hand-held vision inspection system and method |
| CN109285602A (en) * | 2017-07-19 | 2019-01-29 | 索尼公司 | Main module, system and method for self-examination eyes of user |
| US20190096055A1 (en) * | 2017-09-27 | 2019-03-28 | Fanuc Corporation | Inspection device and inspection system |
| US20190125183A1 (en) * | 2017-10-31 | 2019-05-02 | Welch Allyn, Inc. | Visual acuity examination |
-
2019
- 2019-06-13 CN CN201910513319.7A patent/CN110338748B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2002039754A1 (en) * | 2000-11-08 | 2002-05-16 | Andrzej Czyzewski | Visual screening tests by means of computers |
| CN102599879A (en) * | 2012-02-23 | 2012-07-25 | 天津理工大学 | Self-adaptive eyesight test intelligent system and eyesight test method |
| US20140211164A1 (en) * | 2013-01-31 | 2014-07-31 | Nidek Co., Ltd. | Optometric apparatus |
| CN106537290A (en) * | 2014-05-09 | 2017-03-22 | 谷歌公司 | Systems and methods for biomechanically-based eye signals for interacting with real and virtual objects |
| US9277857B1 (en) * | 2014-11-06 | 2016-03-08 | Bertec Corporation | System for testing and/or training the vision of a subject |
| CN106060142A (en) * | 2016-06-17 | 2016-10-26 | 杨斌 | Mobile phone capable of checking eyesight, and method for checking eyesight by using mobile phone |
| CN106778597A (en) * | 2016-12-12 | 2017-05-31 | 朱明� | Intellectual vision measurer based on graphical analysis |
| CN107198505A (en) * | 2017-04-07 | 2017-09-26 | 天津市天中依脉科技开发有限公司 | Visual function detecting system and method based on smart mobile phone |
| CN107411700A (en) * | 2017-04-07 | 2017-12-01 | 天津大学 | A kind of hand-held vision inspection system and method |
| CN107358036A (en) * | 2017-06-30 | 2017-11-17 | 北京机器之声科技有限公司 | A kind of child myopia Risk Forecast Method, apparatus and system |
| CN109285602A (en) * | 2017-07-19 | 2019-01-29 | 索尼公司 | Main module, system and method for self-examination eyes of user |
| US20190096055A1 (en) * | 2017-09-27 | 2019-03-28 | Fanuc Corporation | Inspection device and inspection system |
| US20190125183A1 (en) * | 2017-10-31 | 2019-05-02 | Welch Allyn, Inc. | Visual acuity examination |
Non-Patent Citations (1)
| Title |
|---|
| SVEN P. HEINRICH ET AL.: ""The dynamics of practice effects in an optotype acuity task"", 《BASIC SCIENCE》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111265182A (en) * | 2020-01-21 | 2020-06-12 | 李小丹 | AI remote optometry service platform and optometry equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110338748B (en) | 2022-03-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103528617B (en) | A kind of cockpit instrument identifies and detection method and device automatically | |
| Beesley et al. | Pre-exposure of repeated search configurations facilitates subsequent contextual cuing of visual search. | |
| CN110322104A (en) | Performance indicators in interactive computer simulation | |
| CN110322099A (en) | The training activity that assessment user carries out in interactive computer simulation | |
| CN110264444A (en) | Damage detecting method and device based on weak segmentation | |
| CN109978870A (en) | Method and apparatus for output information | |
| CN109948740A (en) | A kind of classification method based on tranquillization state brain image | |
| CN114343577B (en) | Cognitive function evaluation method, terminal device, and computer-readable storage medium | |
| CN114155397B (en) | Small sample image classification method and system | |
| CN109935337A (en) | A kind of medical record lookup method and system based on similarity measurement | |
| CN107705231A (en) | A kind of computer assisted method to go over files, device and computer-readable recording medium | |
| CN115713256A (en) | Medical training assessment and evaluation method and device, electronic equipment and storage medium | |
| CN110322098A (en) | S.O.P. feedback during interactive computer simulation | |
| CN109978868A (en) | Toy appearance quality determining method and its relevant device | |
| Banno et al. | The processing speed of scene categorization at multiple levels of description: The superordinate advantage revisited | |
| CN111612865A (en) | MRI (magnetic resonance imaging) method and device for generating countermeasure network based on conditions | |
| CN111429547A (en) | Abnormal color vision test chart synthesis method based on false homochromy search | |
| CN110338748A (en) | Quickly method, storage medium, terminal and the sight tester of positioning eyesight value | |
| CN116350203B (en) | Physical testing data processing method and system | |
| CN110276802A (en) | Illness tissue localization method, device and equipment in medical image | |
| KR20150061206A (en) | Device and method of diagnosing mental state using diagram and color | |
| CN110956226A (en) | Handwriting track abnormity detection method based on deep learning | |
| Borovcnik | Informal and “informal” inference | |
| CN116309465A (en) | Tongue image detection and positioning method based on improved YOLOv5 in natural environment | |
| CN110096708A (en) | A kind of determining method and device of calibration collection |
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 |