Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It needs
Illustrate, some or all of in any number of technical solutions provided below technical characteristic in the absence of conflict,
It can be used in combination, form new technical solution.Based on the embodiment in the application, those of ordinary skill in the art are not making
Every other embodiment obtained under the premise of creative work out, shall fall in the protection scope of this application.
Term "and/or" in the embodiment of the present application, a kind of only incidence relation for describing affiliated partner, expression can be with
There are three kinds of relationships, for example, A and/or B, can indicate: individualism A exists simultaneously A and B, these three feelings of individualism B
Condition.In addition, character "/" herein, typicallys represent the relationship that forward-backward correlation object is a kind of "or".If explanation is not added, herein
In " multiple " refer to two or more.
In the embodiment of the present application, " illustrative " or " such as " etc. words for indicate make example, illustration or explanation.This
Application embodiment in be described as " illustrative " or " such as " any embodiment or design scheme be not necessarily to be construed as comparing
Other embodiments or design scheme more preferably or more advantage.Specifically, use " illustrative " or " such as " etc. words purport
Related notion is being presented in specific ways.
It should be noted that in the embodiment of the present application, unless otherwise indicated, the meaning of " plurality " is refer to two or two with
On.
It should also be noted that, in the embodiment of the present application, " (English: of) ", " corresponding (English:
Corresponding, relevant) " it sometimes can be mixed with " corresponding (English: corresponding) ", it should be pointed out that
It is that, when not emphasizing its difference, meaning to be expressed is consistent.
The basic principle of technical solution provided by the embodiment of the present application are as follows: be directed to coin identification method in the prior art
It is not suitable for the problem of obtaining money information in blind person's daily life, passing through the image for obtaining coin in the embodiment of the present application, so
The classification for obtaining coin according to the image of coin afterwards obtains corresponding money information further according to coin classification;Finally by voice
Money information is broadcasted.Since blind person can obtain money information by the sense of hearing, and in the identification process of coin
Operation is simple, so the embodiment of the present application, which can be adapted for assisting blind, obtains money information.
Based on above content, the embodiment of the present application provides a kind of blind guiding system.
Illustratively, the blind guiding system in the embodiment of the present application may include: wear-type blind-guide device, glasses for guiding blind, hand
The terminal devices such as machine, portable computer, pocket computer, handheld computer, Digital Frame, palm PC, navigator,
Or blind guiding system also may include: terminal device, cordless communication network and network server.
Shown in referring to Fig.1, blind guiding system 10 provided by the embodiments of the present application includes: image capture device 11, processing mould group
12 and voice-output device 13;Image capture device 11 and voice-output device 12 are coupled to processing mould group 12.
Processing mould group 12 obtains coin image to be detected for controlling image capture device 11, according to coin image to be detected
It obtains the classification of coin and corresponding money information is obtained according to the classification of coin;Processing module 11 is also used to control voice defeated
Equipment 13 broadcasts money information by voice out.
Illustratively, above-mentioned image capture module 11 can be the first-class imaging sensor of monocular cam, binocular camera shooting
One of or it is a variety of.Processing mould group 12 can be processor, be also possible to the general designation of multiple processing elements.For example, centre
Manage device (English name: central processing unit, abbreviation: CPU).Handling mould group 12 can also be other general places
Manage device, digital signal processor (English name: digital signal processing, abbreviation: DSP), dedicated integrated electricity
Road, field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware group
Part etc. may be implemented or execute various illustrative logic blocks, module and electricity in conjunction with described in present disclosure
Road.General processor can be microprocessor or the processor is also possible to any conventional processor etc..Handle mould group 12
It can also be application specific processor, which may include at least one in baseband processing chip, radio frequency processing chip etc.
It is a.Processor is also possible to realize the combination of computing function, such as combines comprising one or more microprocessors, DSP and Wei Chu
Manage the combination etc. of device.Voice-output device 13 can be the audio output apparatus such as loudspeaker, power amplifier, speaker, earphone.
It should be noted that the coin in the embodiment of the present application can be bank note, or coin, it can be with other classes
The coin of type.In addition, the currency type of the coin in the embodiment of the present application can for RMB, dollar, sterling, Euro etc..
It should also be noted that, money information information may include coin: currency type (such as: RMB, dollar, sterling,
Euro etc.), denomination (such as: 1 yuan, 2 yuan, 5 yuan, 10 yuan, 100 yuan etc.), version (such as: the 5th set of RMB, third cover people
People's coin, new edition dollar etc.) etc. information.
Illustratively, during processing mould group 12 controls image capture device 11 and obtains coin image to be detected: place
Reason mould group can control image capture device from a visual angle to coin carry out Image Acquisition, or control image capture device from
Multiple visual angles carry out Image Acquisition to coin.In addition, same visual angle can also carry out multiple Image Acquisition to coin, it is then based on
Clarity, integrity degree of the image of acquisition etc. choose therein one or more and are used as coin image to be detected.
In addition, when carrying out Image Acquisition to coin from multiple visual angles, handling mould group 12 and image capture device 11 can be with
It is realized by performing the following operations:
Coin is placed within the acquisition range of image collecting device by user, and then coin is moved and/or turned over
Turn, processing mould group 12 controls image collecting device 11 during user moves and/or overturns to coin repeatedly to money
Coin carries out Image Acquisition, carries out Image Acquisition to coin to realize from multiple visual angles.Or setting image capture device 11 wraps
Include it is multiple can be with the device of independent acquisition image, and multiple different positions can be respectively arranged at the device of independent acquisition image
It sets.When carrying out Image Acquisition, processing mould group 12 controls multiple image collecting devices for being set to different location respectively from one
Visual angle carries out Image Acquisition to coin, carries out Image Acquisition to coin to realize from multiple visual angles.Such as: image capture device
Including two opposite cameras, when carrying out Image Acquisition acquisition to coin, coin is placed between two cameras by user,
Two cameras acquire the image of coin head and reverse side respectively.
The blind guiding system that embodiments herein provides includes: that image capture device, processing mould group and voice output are set
It is standby;Image capture device and voice-output device are coupled to processing mould group, wherein processing mould group can control image capture device
Coin image to be detected is obtained, the classification of coin is obtained according to coin image to be detected and correspondence is obtained according to the classification of coin
Money information;Processing module is also used to control voice-output device and is broadcasted by voice to money information, due to this Shen
Please the blind guiding system that provides of embodiment when carrying out Image Acquisition and obtaining coin image to be detected, be not necessarily to multipair coin placement position
The influence of intensity of illumination, light position etc. when setting and limited, and not having to consider Image Acquisition, so the identification to coin
Operation is simple in journey, and money information is obtained suitable for blind person's daily life, i.e. the embodiment of the present application can assist blind
People obtains money information.
Optionally, processing mould group 12 obtains the classification of coin according to coin image to be detected and is obtained according to the classification of coin
Take corresponding money information, comprising: processing mould group 12 intercepts detection zone on coin image to be detected;Detection zone is with to be checked
It surveys centered on the center of coin image;Processing is carried out to the image for the detection zone being truncated to and obtains second processing image;Second
Processing image is that the pixel size of gray scale image and second processing image is preset value;It is special that detection is extracted from second processing image
Sign;Calculate separately detection feature and each coin feature templates Euclidean distance (English name: Euclidean Metric,
Also known as: euclidean metric);Obtain matching characteristic template;Matching characteristic template is the smallest with the Euclidean distance of detection feature
Feature templates;The classification of coin is obtained according to matching characteristic template and corresponding money information is obtained according to the classification of coin.
That is, processing mould group 12 can be realized by executing following steps according to coin image to be detected referring to shown in Fig. 2
It obtains the classification of coin and corresponding money information is obtained according to the classification of coin.
S21, detection zone is intercepted on coin image to be detected.
Wherein, detection zone is centered on the center of coin image to be detected.
Illustratively, referring to shown in Fig. 3, the center of coin image 31 to be detected is 310, the center of detection zone 32 with to
The center 310 for detecting coin image 31 is overlapped.It should be noted that with coin image 31 to be detected and detection zone 32 in Fig. 3
It is to be illustrated for rectangle, but it's not limited to that for the embodiment of the present application, coin image 31 to be detected and detection zone 32
It may be other shapes.
S22, processing acquisition second processing image is carried out to the image for the detection zone being truncated to.
Wherein, it is preset value that second processing image, which is the pixel size of gray scale image and second processing image,.
Illustratively, processing is carried out to the image for the detection zone being truncated to and obtains second processing image concretely: is first
The image that first will test region is converted to gray scale image, and then gray scale image is normalized.It can certainly pass through
The image of other image procossing mode detection zones is handled, do not limited in the embodiment of the present application to the image of detection zone into
The mode of row image procossing, gray scale image of the image procossing in region as pixel size for preset value can be will test.
Illustratively, preset value is specifically as follows 100 pixel *, 100 pixel.
S23, detection feature is extracted from second processing image.
Illustratively, detection feature is specifically as follows: size dimensions invariant features convert (English name: Scale-
Invariant Feature Transform, referred to as: SIFT) feature.
S24, the Euclidean distance for calculating separately detection feature with the template characteristic of each coin.
S25, matching characteristic template is obtained.
Wherein, matching characteristic template is the smallest feature templates of Euclidean distance with detection feature.
S26, corresponding coin letter is obtained according to the classification of matching characteristic template acquisition coin and according to the classification of coin
Breath.
Further, mould group 12 is handled in above-mentioned steps S24 needs the template of calculating detection feature and each coin special
The Euclidean distance of sign, therefore processing mould group 12 also needs the feature templates to each coin to extract simultaneously before step S24
It saves, is calculated in order to read the feature templates of each coin in step s 24.Therefore processing mould group 12 is also used to obtain
The template image of each coin;Processing is carried out to the template image of each coin and obtains corresponding first processing image;At first
Reason image is gray scale image and the first pixel size for handling image is preset value;Feature is extracted from each first processing image
Template simultaneously saves.
That is, processing mould group 12 extracts the feature templates of each coin and is saved and can be led to referring to shown in Fig. 4
Execution following steps are crossed to realize.
S41, the template image for obtaining each coin.
Specifically, the template image for obtaining each coin, which is specifically as follows, obtains various coin heads, reverse side original scale
Template image of the clear image of original text as coin.Such as: the template image for obtaining dollar is specifically as follows: respectively obtain 1 dollar,
2 dollars, 5 dollars, 10 dollars, 20 dollars, 50 dollars and 100 dollars fronts, reverse side original scale the clear image of original text, and will
Template image of 14 images obtained as coin.
S42, the corresponding first processing image of processing acquisition is carried out to the template image of each coin.
Wherein, the first processing image is gray scale image and the first pixel size for handling image is preset value.
Illustratively, processing is carried out to the template image of each coin and obtains the first processing image concretely: first will
The template image of each coin is converted to gray scale image, and then gray scale image is normalized.Equally, the application is implemented
The mode that image procossing is carried out to the template image of each coin is not limited in example, with can be by the template image of each coin yet
It is subject to the gray scale image of preset value that processing, which is pixel size,.
S43, template characteristic is extracted from each first processing image and is saved.
Illustratively, processing mould group 12 can save the template characteristic of each coin in the local database, can also be with
The template characteristic of each coin is stored in remote server.
It should be noted that detection feature and template characteristic are the feature of same type.Such as: template characteristic is SIFT special
Sign, then detecting feature is also SIFT feature.
It should also be noted that, above-mentioned steps S41, S42 and S43 are characterized template extraction and save process, in some realities
It applies in example, may not need and execute above-mentioned S41, S42 and S43.Such as: template characteristic has been extracted and preserved when using for the first time
In local data base, next time by implementation is shown in Fig. 2 to obtain money information when, can be directly from local data base
In transfer template characteristic, without repeating above-mentioned steps S41, S42 and S43.
It should also be noted that, if template can be enriched by being performed both by above-mentioned steps S41, S42 and S43 to each coin
Feature, and then identification classification can be carried out to each coin when obtaining the classification of coin, but the world is existing at present
More than 1200 kinds of coins, coin is many kinds of, if being performed both by above-mentioned steps S41, S42 and S43 to each coin, needs more
Computing resource, storage resource, and most of template characteristic takes less than in blind person's daily life, results in waste of resources, therefore
Above-mentioned steps S41, S42 and S43 only can be executed to a part of coin in actual use, and user can be according to practical need
It asks increase or deletes segment template feature.
Further, processing mould group 12 is also used to judge the minimum value of the Euclidean distance of detection feature and each feature templates
Whether threshold value is less than or equal to;If so, using with detection feature the smallest feature templates of Euclidean distance as matching characteristic template;
If it is not, whether judgement is equal to preset times to the number of coin image rotation to be detected;If being revolved to coin image to be detected
The number turned is less than preset times, intercepts detection zone to coin image rotation predetermined angle to be detected and again;
Processing mould group 12 is for intercepting detection zone on coin image to be detected, comprising: is getting coin to be detected
Detection zone is intercepted after image on coin image to be detected and after rotating coin image to be detected, in coin figure to be detected
As upper interception detection zone.
That is, referring to Figure 5, in above-mentioned steps S25, processing mould group 12 obtains matching characteristic template can be by executing such as
Lower step is realized.
S281, judge whether the minimum value for detecting the Euclidean distance of feature and each feature templates is less than or equal to threshold value.
In step S281, if the minimum value of detection feature and the Euclidean distance of each feature templates is less than or equal to threshold
Value, thens follow the steps S282;If the minimum value for detecting the Euclidean distance of feature and each feature templates is greater than threshold value, step is executed
Rapid S283.
S282, using with detection feature the smallest feature templates of Euclidean distance as matching characteristic template.
Whether S283, judgement are equal to preset times to the number of coin image rotation to be detected.
Illustratively, preset times can for 1 time, 2 times, 3 times, it is 10 inferior.
In step S283, if preset times are less than to the number of coin image rotation to be detected, such as: preset times
It is 1 time, rotation not yet was carried out to coin image to be detected;Again for example: preset times be 10 times, to coin image to be detected into
It went 5 times and rotated;It thens follow the steps in S284 and return step S21 and re-execute the steps S21 to S283;If to money to be detected
The number of coin image rotation is equal to preset times, then explanation fails to will test feature and template characteristic successful match, terminates detection.
" coin is not detected " or " detection failure " can be exported at this time.
S284, by coin image rotation predetermined angle to be detected.
Illustratively, it referring to shown in Fig. 6, is carried out for being 45 degree by the angle a of coin image rotation to be detected in Fig. 6
Explanation.It should be noted that predetermined angle can also be other angles, and such as: 10 degree, 30 degree, 90 degree etc..In addition, rotation is pre-
If angle can be to rotate clockwise predetermined angle, or rotate predetermined angle counterclockwise.
It should be noted that in the above-described embodiments, if preset times setting is bigger, step S21 to S283 is weighed
The number executed again is more, it is longer to calculate time-consuming, but it is bigger to will test a possibility that feature is with template characteristic successful match.Instead
It, if preset times setting is smaller, the number that step S21 to S283 is repeatedly executed is fewer, it is shorter to calculate time-consuming, but will examine
It is smaller to survey a possibility that feature is with template characteristic successful match.Therefore the size of preset times can be set according to actual needs
It is fixed.
Optionally, processing mould group 12 obtains the classification of coin according to coin image to be detected and is obtained according to the classification of coin
Take corresponding money information, comprising: processing mould group extracts default feature in coin image to be detected;Default feature is divided
Class obtains the classification of coin and obtains corresponding money information according to the classification of coin.
That is, processing mould group 12 can also obtain the classification and root of coin by executing following steps referring to shown in Fig. 7
Corresponding money information is obtained according to the classification of coin.
S71, default feature is extracted in coin image to be detected.
Optionally, the default feature in above-described embodiment can be the feature of manual setting, such as: Haar feature (English
Title: Haar-like features), histograms of oriented gradients (English name: Histogram of Oriented
Gradient, referred to as: HOG) feature, local binary patterns (English name: Local Binary Pattern, abbreviation: LBP) spy
Sign.Default feature in above-described embodiment is also possible to through machine learning method, for example is learned automatically by deep learning method
Practise the characteristics of image obtained.
S72, classify to default feature obtains the classification of coin and obtains corresponding coin according to the classification of coin
Information.
Specifically, the classification method classified in above-described embodiment to default feature can be with are as follows: use support vector machines
(English name: Support Vector Machine, referred to as: SVM) classified, iteration (English name: adaboost) point
Class device carries out classification or k-nearest neighbor is classified, and then obtains corresponding money information according to the classification of coin.
Optionally, processing mould group 12 obtains the classification of coin according to coin image to be detected and is obtained according to the classification of coin
Take corresponding money information, comprising: processing mould group is specifically used for through deep neural network classification to coin image to be detected
Classification is carried out to obtain the classification of coin and obtain corresponding money information according to the classification of coin.
That is, directly classified by preparatory trained deep neural network classifier to coin image to be detected, from
And it obtains the classification of coin and corresponding money information is obtained according to the classification of coin.
Optionally, referring to shown in Fig. 8, processing mould group 12 include: first processing units 121, the second processing unit 122 and
Network communication unit 123;
First processing units 121 obtain coin figure to be detected for controlling image capture device
Picture, and acquired image is sent to the second processing unit 122 by network communication unit 123;
The second processing unit 122 is used to obtain the classification of coin according to coin image to be detected and according to the classification of coin
Corresponding money information is obtained, and first processing units are back to by the money information that network communication unit 123 will acquire
121。
That is, above-mentioned obtain the classification of coin according to coin image to be detected and obtain corresponding money according to the classification of coin
The realization process of coin information can be completed inside terminal device, can also assist to complete by remote service apparatus.Work as basis
Coin image to be detected obtains the classification of coin and is passed through according to the process that the classification of coin obtains corresponding money information remote
When journey service equipment assists to complete, the network communication unit 123 of processing mould group 12 obtains first processing units 121 to be detected
Coin image is sent in the second processing unit 122, and the second processing unit 122 obtains the class of coin according to coin image to be detected
And according to the classification of coin corresponding money information, and the coin letter that will acquire by network communication unit 123 are not obtained
Breath is back to first processing units 123.
Optionally, referring to shown in Fig. 9, blind guiding system further includes the helmet 900;Image capture device 11, processing mould group 12 with
And voice-output device 13 is arranged on the helmet 900.
Optionally, referring to Fig.1 shown in 0, blind guiding system further include: voice-input device 14;
Processing mould group 12 is also used to acquire phonetic order by voice-input device 14;It collects specifically for working as referring to
When showing the instruction for carrying out coin identification, control image capture device 11 obtains coin image to be detected, according to coin figure to be detected
As obtaining the classification of coin and obtaining corresponding money information according to the classification of coin.In this way, it enables to lead
Blind system enters the process of coin image acquisition according to the triggering of blind person, to be carried out using the template of corresponding coin identification
Identification can effectively improve the efficiency and precision of coin identification in this way.
Illustratively, voice-input device 14 can be microphone.
Those skilled in the art are it will be appreciated that in said one or multiple examples, function described herein
It can be realized with hardware, software, firmware or their any combination.It when implemented in software, can be by these functions
Storage in computer-readable medium or as on computer-readable medium one or more instructions or code transmitted.
Computer-readable medium includes computer storage media and communication media, and wherein communication media includes convenient for from a place to another
Any medium of one place transmission computer program.Storage medium can be general or specialized computer can access it is any
Usable medium.
The above, the only specific embodiment of the application, but the protection scope of the application is not limited thereto, it is any
Within the technical scope of the present application, any changes or substitutions that can be easily thought of by those familiar with the art, all answers
Cover within the scope of protection of this application.Therefore, the protection scope of the application should be subject to the protection scope in claims.