CN108717542A - Identify the method, apparatus and computer readable storage medium of character area - Google Patents
Identify the method, apparatus and computer readable storage medium of character area Download PDFInfo
- Publication number
- CN108717542A CN108717542A CN201810367675.8A CN201810367675A CN108717542A CN 108717542 A CN108717542 A CN 108717542A CN 201810367675 A CN201810367675 A CN 201810367675A CN 108717542 A CN108717542 A CN 108717542A
- Authority
- CN
- China
- Prior art keywords
- image
- region
- probability
- character area
- images
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Character Discrimination (AREA)
- Character Input (AREA)
Abstract
The disclosure is directed to a kind of method, apparatus and computer readable storage medium of identification character area.Using this method, first, the characteristic information of images to be recognized is input in character area identification model, obtain the first probability and the second probability of each first image-region in the images to be recognized, then, according to the first probability of each first image-region, filter out include word image-region, then, filtering out on the basis of including the image-region of word, further analyze the second probability of the image-region, judge whether mutually to merge image-region image-region adjacent thereto, finally, according to the image-region after merging, determine the character area in images to be recognized.Therefore, by determining the first probability and the second probability of each first image-region in images to be recognized, character area can be directly calculated, the character area recognition methods that a kind of accuracy is high and recognition speed is fast is provided.
Description
Technical field
This disclosure relates to which the method, apparatus and computer of image processing field more particularly to a kind of identification character area can
Read storage medium.
Background technology
With the rapid development of Internet technology, the picture number on internet is more and more, wants to user's acquisition
Picture brings inconvenience.Also include some words due to including not only picture in picture, people are intended to pass through search graph
Word in piece filters out desired picture, under the promotion of this trend, OCR (Optical Character
Recognition, optical character identification) technology comes into being.
OCR refers to the technology that optical character is identified by image procossing and mode identification technology, is generally comprised
Two stages:Character area identifies and Text region, wherein it is residing in the picture that character area identification goes out word for identification
Position, Text region go out the word in character area for identification.
Under normal conditions, identification character area is the method based on Adaboost that uses mostly, and this method mainly utilizes
The feature of engineer carries out character area identification, due to the error of artificial design features so that adopt this method into style of writing
The accuracy of word region recognition is not high.
Invention content
To overcome the problems in correlation technique, the disclosure provides a kind of method, apparatus and meter of identification character area
Calculation machine readable storage medium storing program for executing.
According to the first aspect of the embodiments of the present disclosure, a kind of method of identification character area is provided, including:
The characteristic information of images to be recognized is inputted into character area identification model, is obtained each the in the images to be recognized
The first probability and the second probability of one image-region, the first probability characterization described first image region is the general of character area
Rate, the probability that the second probability characterization described first image region is connected with adjacent image-region, wherein the literal field
The domain identification model character area in image for identification;
The described first image region that first probability is more than to the first probability threshold value is determined as the second image-region;
Two adjacent second image-regions that the second probability threshold value is all higher than to the second probability merge;
According to the image-region after merging, the character area in the images to be recognized is determined.
Optionally, it is determined as the second image district in the described first image region that the first probability is more than to the first probability threshold value
After domain, the method further includes:
Second image-region is inputted into the character area identification model, obtains each institute in the images to be recognized
State the position offset of the second image-region;
According to the position offset of each second image-region, the position of second image-region is adjusted;
Two adjacent second image-regions that the second probability threshold value is all higher than to the second probability merge, packet
It includes:
To the second probability be all higher than the second probability threshold value two adjacent and position adjustment after the second image-region into
Row merges.
Optionally, the image-region according to after merging determines the character area in the images to be recognized, including:
Determine the minimum enclosed rectangle of the image-region after merging;
Region where the minimum enclosed rectangle is determined as the character area in the images to be recognized.
Optionally, the method further includes:
According to the character area in the characteristic information of sample image and the sample image, convolutional neural networks are carried out
Training, obtains the character area identification model.
According to the second aspect of the embodiment of the present disclosure, a kind of device of identification character area is provided, including:
Probability obtains module, is configured as the characteristic information of images to be recognized inputting character area identification model, obtain
The first probability and the second probability of each first image-region in the images to be recognized, the first probability characterization described first
Image-region is the probability of character area, and the second probability characterization described image region is connected general with adjacent image-region
Rate, wherein the character area identification model character area in image for identification;
First determining module, the described first image region for being configured as the first probability being more than the first probability threshold value determine
For the second image-region;
Merging module is configured as being all higher than the second probability two adjacent second images of the second probability threshold value
Region merges;
Second determining module is configured as determining the word in the images to be recognized according to the image-region after merging
Region.
Optionally, described device further includes:
Offset obtains module, is configured as second image-region inputting the character area identification model, obtain
To the position offset of each second image-region in the images to be recognized;
Module is adjusted, the position offset according to each second image-region is configured as, to second image district
The position in domain is adjusted;
The merging module includes:
Merge submodule, is configured as being all higher than the second probability the two adjacent and position adjustment of the second probability threshold value
The second image-region afterwards merges.
Optionally, the merging module includes:
First determination sub-module is configured to determine that the minimum enclosed rectangle of the image-region after merging;
Second determination sub-module is configured as the region where the minimum enclosed rectangle being determined as the figure to be identified
Character area as in.
Optionally, described device further includes:
Training module is configured as according to the character area in the characteristic information of sample image and the sample image,
Convolutional neural networks are trained, the character area identification model is obtained.
According to the third aspect of the embodiment of the present disclosure, a kind of computer readable storage medium is provided, is stored thereon with calculating
Machine program instruction realizes the side for the identification character area that disclosure first aspect is provided when the program instruction is executed by processor
The step of method.
In the embodiments of the present disclosure, first, the characteristic information of images to be recognized is input in character area identification model,
The first probability and the second probability of each first image-region in the images to be recognized are obtained, then, according to each first image
First probability in region, filter out include word image-region, then, filter out include word image-region
On the basis of, the second probability of the image-region is further analyzed, is judged whether image-region image district adjacent thereto
Domain mutually merges, and finally, according to the image-region after merging, determines the character area in images to be recognized.Therefore, it is waited for by determination
The first probability and the second probability for identifying each first image-region in image, can directly calculate character area, provide
The character area recognition methods that a kind of accuracy is high and recognition speed is fast.
It should be understood that above general description and following detailed description is only exemplary and explanatory, not
The disclosure can be limited.
Description of the drawings
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is a kind of flow chart of the method for identification character area shown according to an exemplary embodiment.
Fig. 2 is a kind of another flow chart of the method for identification character area shown according to an exemplary embodiment.
Fig. 3 is a kind of another flow chart of the method for identification character area shown according to an exemplary embodiment.
Fig. 4 is a kind of another flow chart of the method for identification character area shown according to an exemplary embodiment.
Fig. 5 is a kind of block diagram of the device of identification character area shown according to an exemplary embodiment.
Fig. 6 is a kind of another block diagram of the device of identification character area shown according to an exemplary embodiment.
Fig. 7 be a kind of identification character area shown according to an exemplary embodiment device in merging module block diagram.
Fig. 8 is a kind of another block diagram of the device of identification character area shown according to an exemplary embodiment.
Fig. 9 is a kind of block diagram of the device of character area for identification shown according to an exemplary embodiment.
Specific implementation mode
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all implementations consistent with this disclosure.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the disclosure.
The embodiment of the present disclosure provides a kind of method of identification character area.Referring to FIG. 1, Fig. 1 is exemplary according to one
Implement a kind of flow chart of the method for the identification character area exemplified, as shown in Figure 1, the identification text that the embodiment of the present disclosure provides
The method in block domain includes the following steps.
In step s 11, the characteristic information of images to be recognized is inputted into character area identification model, obtained described to be identified
The first probability and the second probability of each first image-region in image, the first probability characterization described first image region are
The probability of character area, the probability that the second probability characterization described first image region is connected with adjacent image-region,
In, the character area identification model character area in image for identification.
In step s 12, the described first image region for the first probability being more than to the first probability threshold value is determined as the second figure
As region.
In step s 13, two adjacent second image-regions of the second probability threshold value are all higher than to the second probability
It merges.
In step S14, according to the image-region after merging, the character area in the images to be recognized is determined.
CNN(convolutional neural network;Convolutional neural networks) it is a kind of feedforward neural network, mainly
It being made of convolutional layer and articulamentum, wherein articulamentum requires the size of input picture, and in practical application,
Its possible size of different images is also inconsistent, cannot be satisfied the requirement of articulamentum in CNN, being likely to result in CNN can not
Therefore the consequence of identification image needs to zoom in and out the image before images to be recognized is inputted full convolutional network, with full
The requirement of articulamentum in the foot CNN.
In image processing field, due to the pixel value of each pixel in image be only capable of characterizing basic protochrome and its
The basic coding of gray scale, not high-rise semantic information, wherein high-rise semantic information is for indicating which specifically has in image
Object, the object where etc..Therefore, after zooming in and out images to be recognized, literal field can not directly be carried out
Domain identifies, but the images to be recognized after the scaling can be specifically input to by the high-rise semantic information that need to obtain the image
In CNN, an eigenmatrix (hereinafter referred to as characteristic pattern) for corresponding to the image is obtained, is identified from this feature figure to be identified
The high-layer semantic information of image, and then can further identify the region where word.
In the embodiments of the present disclosure, the characteristic information of images to be recognized can be extracted from the characteristic pattern of images to be recognized
Come, includes the high-layer semantic information of the images to be recognized in this feature information.First, in step s 11, by images to be recognized
Characteristic information input character area identification model, obtain in the images to be recognized the first probability of each first image-region and
Second probability, the first probability of each first image-region characterize the probability that first image-region is character area, Mei Ge
Second probability of one image-region characterize first image-region whether the probability being connected with adjacent image-region.First image
Region is any image region in images to be recognized.
Specifically, multiple anchor points are pre-set in images to be recognized, wherein some anchor point is for indicating the image
In region, referred to as region anchor point, another part anchor point is for indicating whether adjacent area needs to connect in the image, referred to as
To connect anchor point.The rectangular area that multiple band rotations are pre-set in artwork, multiple images area is divided by images to be recognized
Domain, wherein the rectangular area with rotation is represented by (x, y, w, h, θ), and x, y are that the central point of the rectangular area of band rotation is sat
Mark, w, h are the width and height of the rectangular area of band rotation, and θ is that the rectangular area of band rotation is relative to horizontal direction
Rotation angle, each region anchor point correspond to a rectangular area with rotation, that is to say, one in corresponding images to be recognized
Image-region.Optionally, the rectangular area of preset band rotation can overlap each other.
After the characteristic information of images to be recognized is inputted character area identification model, the character area identification model is to this
The first image-region of each of images to be recognized is identified, and includes in the recognition result of each first image-region
One probability and the second probability, wherein the first probability is used to characterize the probability that first image-region is character area, the second probability
The probability being connected with adjacent image regions for characterizing first image-region.
Optionally, the character area identification model in step S11 is the model of the character area in image for identification, can
With by obtaining after being trained to full convolutional network.Specifically, character area identification can be obtained by executing step S17
Model.Fig. 2 is a kind of another flow chart of the method for identification character area shown according to an exemplary embodiment.Such as Fig. 2 institutes
Show, the method further includes step S17 other than including step S11-S14.
In step S17, according to the character area in the characteristic information of sample image and the sample image, to convolution
Neural network is trained, and obtains the character area identification model.
Under normal conditions, the coefficient in convolutional neural networks is randomly generated, and the coefficient generated at random using this is treated
The accuracy that the character area of identification image is identified can not ensure, therefore, word carried out using the convolutional neural networks
Before region recognition, demand that can be according to user to Text region accuracy is trained the convolutional network, to adjust convolution
Coefficient in neural network keeps the character area identified by the convolutional neural networks more accurate.
Specifically, sample image to be identified can be inputted in convolutional neural networks, it can by the processing of convolutional neural networks
To export result sample image, at least one rectangular area in the result sample image, the rectangular area is known to characterize
The character area not gone out, since rectangular area is identified according to the coefficient generated at random, the possible rectangular area
The character area in sample image to be identified can not accurately be covered, which (is had with target sample image
It is the character area in sample image to be identified to have the sample image to be identified of rectangle frame, the region of rectangle frame delineation) comparison,
According to error between the two, the coefficient in convolutional neural networks is adjusted, to reduce result sample image and target sample image
Error.The above method is repeated several times, until the mistake of the result sample image and target sample image of convolutional neural networks output
Until difference meets preset requirement, the convolutional neural networks after coefficient adjustment finishes are character area identification model, wherein in advance
If it is required that being demand of the user to Text region accuracy and pre-set.
After executing the step S11, step S12 is executed, according in the images to be recognized determined in step S11 each the
First probability of one image-region screens each first image-region in the images to be recognized.Specifically, in literal field
In the recognition result of domain identification model output, the numerical value of the first probability of the first image-region is larger, then it is believed that first figure
As containing word in region, the numerical value of the first probability is smaller, then it is assumed that therefore can without containing word in first image-region
One the first probability threshold value of default setting, is compared with the numerical value of the first probability of each first image-region, will be greater than this
The first image-region corresponding to the numerical value of first probability of the first probability threshold value be determined as include word region, and will be true
Fixed includes that the region of word screens, as the second image-region.It optionally, will be less than the of first probability threshold value
The first image-region corresponding to the numerical value of one probability is determined as not including the region for having word, is not processed to the region.
Wherein, when which is more than the first probability threshold value to the numerical value for being characterized in the first probability, this first
Contain word in the first image-region corresponding to the numerical value of probability, when the numerical value of the first probability is less than the first probability threshold value,
Word is not contained in the first image-region corresponding to the numerical value of first probability.
It includes text that the image-region (namely second image-region) filtered out in step s 12, which is only capable of characterizing in the region,
Word, but the size of word is uncertain, and what preset rectangular area was to determine, which might not completely include
The word, that is to say, the image-region filtered out may include only a part for word, the image adjacent with the image-region
Region includes the remainder of word.Therefore, in the embodiments of the present disclosure, need filter out include word image
On the basis of region, the part whether image-region only contains word is further analyzed by executing step S13.
In step s 13, include that further to analyze it second general for the image-region of word to what is filtered out in step S12
Rate, and the image-region filtered out is merged according to second probability.Specifically, in the output of character area identification model
Include multiple probability numbers in the second probability of each first image-region in recognition result, multiple probability numbers difference
Indicate in the multiple images region adjacent with first image-region whether include the word remainder, wherein probability
Numerical value is larger, then it is believed that also including a part for the word in adjacent image-region corresponding with the probability numbers,
The two is connected, and probability numbers are smaller, then it is believed that not including in adjacent image-region corresponding with the probability numbers has this
A part for word, the two are not attached to.Therefore, second probability threshold value can be pre-set, with the figure filtered out in step S12
As the numerical value of second probability in region is compared, the adjacent figure corresponding to the probability numbers of second probability threshold value will be greater than
Be determined as region include a part for the word region, and the region is merged with the image-region filtered out.
For screen in step s 12 include word each image-region, be performed both by above-mentioned steps, most
Afterwards, in step S14, by the image-region after merging, you can the character area being determined as in images to be recognized.
In the embodiments of the present disclosure, first, the characteristic information of images to be recognized is input in character area identification model,
The first probability and the second probability of each first image-region in the images to be recognized are obtained, then, according to each first image
First probability in region, filter out include word image-region, then, filter out include word image-region
On the basis of, the second probability of the image-region is further analyzed, is judged whether image-region image district adjacent thereto
Domain mutually merges, and finally, according to the image-region after merging, determines the character area in images to be recognized.Therefore, it is waited for by determination
The first probability and the second probability for identifying each first image-region in image, can directly calculate character area, provide
The character area recognition methods that a kind of accuracy is high and recognition speed is fast.
Optionally, Fig. 3 is a kind of another flow of the method for identification character area shown according to an exemplary embodiment
Figure.As shown in figure 3, after step s 12, the method is further comprising the steps of.
In step S15, second image-region is inputted into the character area identification model, is obtained described to be identified
The position offset of each second image-region in image.
In step s 16, according to the position offset of each second image-region, to second image-region
Position be adjusted.
Correspondingly, step S13 specifically includes step S131.
In step S131, the second probability is all higher than after the two adjacent and position adjustment of the second probability threshold value the
Two image-regions merge.
In the embodiments of the present disclosure, the size of the rectangular area of pre-set multiple bands rotation and position in the picture
It sets and is to determine, each of be divided into the size of the first image-region to be also to determine images to be recognized, due to different figures
As in, position where word is simultaneously different, and therefore, the image-region filtered out in step s 12 may include character portion
The region area divided is smaller, as long as at this point, slightly adjusting the position of the image-region, you can it includes this article to make the image-region
The region area of character segment increases, convenient for quickly recognizing character area.Therefore, the second image-region is obtained in step s 12
Afterwards, character area identification model need to be entered into, with obtain second image-region position offset (Δ x, Δ y, Δ w,
Δ h, Δ θ), and according to the position offset, the position of second image-region is adjusted, make larger in the image-region
Region area in can include word, wherein the position of the image-region after adjustment be (x+ Δs x, y+ Δ y, w+ Δ w, h+
Δh,θ+Δθ)。
After image-region is adjusted, the second probability of the image-region after being adjusted according to the position, to position tune
Image-region after whole merges, and specific implementation mode is as it was noted above, details are not described herein again.
Using above-mentioned technical proposal, it is contemplated that the problem that the region area comprising word may be smaller in image-region,
Before judging whether the image-region merges with adjacent image-region, the image-region is finely tuned first, in the image district
The area for increasing covering word in domain, reduces the number of operations of region merging technique, further improves the speed of character area identification.
Optionally, Fig. 4 is a kind of another flow of the method for identification character area shown according to an exemplary embodiment
Figure.As shown in figure 4, step S14 specifically includes following steps in Fig. 1.
In step s 141, the minimum enclosed rectangle of the image-region after merging is determined.
In step S142, the region where the minimum enclosed rectangle is determined as the word in the images to be recognized
Region.
Under normal conditions, due to the adjacent image-region of the image-region be located at the image-region it is upper and lower,
In eight orientation before left, rear, left front, left back, right and behind the right side, which can be with above-mentioned appointing in its eight orientation
The image-region in one orientation merges, and the character area after may merging not is rule, and in character recognition technology
In, usually character area is indicated with a rectangle, therefore, after image-region merging, in order to ensure that word is in same
It in rectangle frame, needs to handle the image-region after the merging, to obtain a rectangular area.
Specifically, first, according to the position of the image-region after merging, determine that the minimum of graphics field after the merging is outer
Rectangle is connect, includes the word in images to be recognized in the minimum enclosed rectangle, then, by the area where the minimum enclosed rectangle
Domain determines the character area in images to be recognized.
Based on same inventive concept, the embodiment of the present disclosure additionally provides a kind of device of identification character area.Fig. 5 is basis
A kind of block diagram of the device of identification character area shown in one exemplary embodiment.With reference to Fig. 5, which includes:
Probability obtains module 501, is configured as the characteristic information of images to be recognized inputting character area identification model, obtain
To the first probability and the second probability of each first image-region in the images to be recognized, the first probability characterization described the
One image-region is the probability of character area, and the second probability characterization described image region is connected with adjacent image-region
Probability, wherein the character area identification model character area in image for identification;
First determining module 502 is configured as the first probability being more than the described first image region of the first probability threshold value
It is determined as the second image-region;
Merging module 503 is configured as being all higher than the second probability two adjacent described the second of the second probability threshold value
Image-region merges;
Second determining module 504 is configured as determining the text in the images to be recognized according to the image-region after merging
Block domain.
Optionally, Fig. 6 is a kind of another block diagram of the device of identification character area shown according to an exemplary embodiment.
As shown in fig. 6, described device 500 further includes:
Offset obtains module 505, is configured as second image-region inputting the character area identification model,
Obtain the position offset of each second image-region in the images to be recognized;
Module 506 is adjusted, the position offset according to each second image-region is configured as, to second image
The position in region is adjusted;
The merging module 503 includes:
Merge submodule 5031, is configured as being all higher than the second probability the two adjacent and position of the second probability threshold value
The second image-region after adjustment merges.
Optionally, Fig. 7 be a kind of identification character area shown according to an exemplary embodiment device in merging module
Block diagram.As shown in fig. 7, the merging module 503 includes:
First determination sub-module 5032 is configured to determine that the minimum enclosed rectangle of the image-region after merging;
Second determination sub-module 5033 is configured as being determined as described waiting knowing by the region where the minimum enclosed rectangle
Character area in other image.
Optionally, Fig. 8 is a kind of another block diagram of the device of identification character area shown according to an exemplary embodiment.
As shown in figure 8, described device 500 further includes:
Training module 507 is configured as according to the literal field in the characteristic information of sample image and the sample image
Domain is trained convolutional neural networks, obtains the character area identification model.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, explanation will be not set forth in detail herein.
The disclosure also provides a kind of computer readable storage medium, is stored thereon with computer program instructions, which refers to
The step of enabling the method for the identification character area for realizing that the disclosure provides when being executed by processor.
Fig. 9 is a kind of block diagram of the device 800 of character area for identification shown according to an exemplary embodiment.Example
Such as, device 800 can be mobile phone, computer, digital broadcast terminal, messaging devices, game console, and tablet is set
It is standby, Medical Devices, body-building equipment, personal digital assistant etc..
With reference to Fig. 9, device 800 may include following one or more components:Processing component 802, memory 804, electric power
Component 806, multimedia component 808, audio component 810, the interface 812 of input/output (I/O), sensor module 814, and
Communication component 816.
The integrated operation of 802 usual control device 800 of processing component, such as with display, call, data communication, phase
Machine operates and record operates associated operation.Processing component 802 may include that one or more processors 820 refer to execute
It enables, to complete all or part of step of the method for above-mentioned identification character area.In addition, processing component 802 may include one
A or multiple modules, convenient for the interaction between processing component 802 and other assemblies.For example, processing component 802 may include more matchmakers
Module, to facilitate the interaction between multimedia component 808 and processing component 802.
Memory 804 is configured as storing various types of data to support the operation in device 800.These data are shown
Example includes instruction for any application program or method that are operated on device 800, contact data, and telephone book data disappears
Breath, picture, video etc..Memory 804 can be by any kind of volatibility or non-volatile memory device or their group
It closes and realizes, such as static RAM (SRAM), electrically erasable programmable read-only memory (EEPROM) is erasable to compile
Journey read-only memory (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic memory, flash
Device, disk or CD.
Electric power assembly 806 provides electric power for the various assemblies of device 800.Electric power assembly 806 may include power management system
System, one or more power supplys and other generated with for device 800, management and the associated component of distribution electric power.
Multimedia component 808 is included in the screen of one output interface of offer between described device 800 and user.One
In a little embodiments, screen may include liquid crystal display (LCD) and touch panel (TP).If screen includes touch panel, screen
Curtain may be implemented as touch screen, to receive input signal from the user.Touch panel includes one or more touch sensings
Device is to sense the gesture on touch, slide, and touch panel.The touch sensor can not only sense touch or sliding action
Boundary, but also detect duration and pressure associated with the touch or slide operation.In some embodiments, more matchmakers
Body component 808 includes a front camera and/or rear camera.When device 800 is in operation mode, such as screening-mode or
When video mode, front camera and/or rear camera can receive external multi-medium data.Each front camera and
Rear camera can be a fixed optical lens system or have focusing and optical zoom capabilities.
Audio component 810 is configured as output and/or input audio signal.For example, audio component 810 includes a Mike
Wind (MIC), when device 800 is in operation mode, when such as call model, logging mode and speech recognition mode, microphone by with
It is set to reception external audio signal.The received audio signal can be further stored in memory 804 or via communication set
Part 816 is sent.In some embodiments, audio component 810 further includes a loud speaker, is used for exports audio signal.
I/O interfaces 812 provide interface between processing component 802 and peripheral interface module, and above-mentioned peripheral interface module can
To be keyboard, click wheel, button etc..These buttons may include but be not limited to:Home button, volume button, start button and lock
Determine button.
Sensor module 814 includes one or more sensors, and the state for providing various aspects for device 800 is commented
Estimate.For example, sensor module 814 can detect the state that opens/closes of device 800, and the relative positioning of component, for example, it is described
Component is the display and keypad of device 800, and sensor module 814 can be with 800 1 components of detection device 800 or device
Position change, the existence or non-existence that user contacts with device 800,800 orientation of device or acceleration/deceleration and device 800
Temperature change.Sensor module 814 may include proximity sensor, be configured to detect without any physical contact
Presence of nearby objects.Sensor module 814 can also include optical sensor, such as CMOS or ccd image sensor, at
As being used in application.In some embodiments, which can also include acceleration transducer, gyro sensors
Device, Magnetic Sensor, pressure sensor or temperature sensor.
Communication component 816 is configured to facilitate the communication of wired or wireless way between device 800 and other equipment.Device
800 can access the wireless network based on communication standard, such as WiFi, 2G or 3G or combination thereof.In an exemplary implementation
In example, communication component 816 receives broadcast singal or broadcast related information from external broadcasting management system via broadcast channel.
In one exemplary embodiment, the communication component 816 further includes near-field communication (NFC) module, to promote short range communication.Example
Such as, NFC module can be based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra wide band (UWB) technology,
Bluetooth (BT) technology and other technologies are realized.
In the exemplary embodiment, device 800 can be believed by one or more application application-specific integrated circuit (ASIC), number
Number processor (DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array
(FPGA), controller, microcontroller, microprocessor or other electronic components are realized, for executing above-mentioned identification character area
Method.
In the exemplary embodiment, it includes the non-transitorycomputer readable storage medium instructed, example to additionally provide a kind of
Such as include the memory 804 of instruction, above-metioned instruction can be executed by the processor 820 of device 800 to complete above-mentioned identification literal field
The method in domain.For example, the non-transitorycomputer readable storage medium can be ROM, random access memory (RAM), CD-
ROM, tape, floppy disk and optical data storage devices etc..
Those skilled in the art will readily occur to other embodiment party of the disclosure after considering specification and putting into practice the disclosure
Case.This application is intended to cover any variations, uses, or adaptations of the disclosure, these modifications, purposes or adaptability
Variation follows the general principles of this disclosure and includes the undocumented common knowledge in the art of the disclosure or usual skill
Art means.The description and examples are only to be considered as illustrative, and the true scope and spirit of the disclosure are by following claim
It points out.
It should be understood that the present disclosure is not limited to the precise structures that have been described above and shown in the drawings, and
And various modifications and changes may be made without departing from the scope thereof.The scope of the present disclosure is only limited by the accompanying claims.
Claims (10)
1. a kind of method of identification character area, which is characterized in that including:
The characteristic information of images to be recognized is inputted into character area identification model, obtains each first figure in the images to be recognized
As first probability and the second probability in region, the first probability characterization described first image region is the probability of character area,
The probability that second probability characterization described first image region is connected with adjacent image-region, wherein the character area
The identification model character area in image for identification;
The described first image region that first probability is more than to the first probability threshold value is determined as the second image-region;
Two adjacent second image-regions that the second probability threshold value is all higher than to the second probability merge;
According to the image-region after merging, the character area in the images to be recognized is determined.
2. according to the method described in claim 1, it is characterized in that, the first probability is more than described the of the first probability threshold value
One image-region is determined as after the second image-region, and the method further includes:
Second image-region is inputted into the character area identification model, is obtained each described the in the images to be recognized
The position offset of two image-regions;
According to the position offset of each second image-region, the position of second image-region is adjusted;
Two adjacent second image-regions that the second probability threshold value is all higher than to the second probability merge, including:
The second image-region being all higher than after the two adjacent and position adjustment of the second probability threshold value to the second probability closes
And.
3. method according to claim 1 or 2, which is characterized in that the image-region according to after merging, determine described in
Character area in images to be recognized, including:
Determine the minimum enclosed rectangle of the image-region after merging;
Region where the minimum enclosed rectangle is determined as the character area in the images to be recognized.
4. according to the method described in claim 1, it is characterized in that, the method further includes:
According to the character area in the characteristic information of sample image and the sample image, convolutional neural networks are instructed
Practice, obtains the character area identification model.
5. a kind of device of identification character area, which is characterized in that including:
Probability obtains module, is configured as the characteristic information of images to be recognized inputting character area identification model, obtains described
The first probability and the second probability of each first image-region in images to be recognized, first probability characterize described first image
Region is the probability of character area, the probability that the second probability characterization described image region is connected with adjacent image-region,
Wherein, the character area identification model character area in image for identification;
First determining module, the described first image region for being configured as the first probability being more than the first probability threshold value are determined as the
Two image-regions;
Merging module is configured as being all higher than the second probability two adjacent second image-regions of the second probability threshold value
It merges;
Second determining module is configured as determining the character area in the images to be recognized according to the image-region after merging.
6. device according to claim 5, which is characterized in that described device further includes:
Offset obtains module, is configured as second image-region inputting the character area identification model, obtains institute
State the position offset of each second image-region in images to be recognized;
Module is adjusted, the position offset according to each second image-region is configured as, to second image-region
Position is adjusted;
The merging module includes:
Merge submodule, is configured as after being all higher than the two adjacent and position adjustment of the second probability threshold value to the second probability
Second image-region merges.
7. device according to claim 5 or 6, which is characterized in that the merging module includes:
First determination sub-module is configured to determine that the minimum enclosed rectangle of the image-region after merging;
Second determination sub-module is configured as the region where the minimum enclosed rectangle being determined as in the images to be recognized
Character area.
8. device according to claim 5, which is characterized in that described device further includes:
Training module is configured as according to the character area in the characteristic information of sample image and the sample image, to volume
Product neural network is trained, and obtains the character area identification model.
9. a kind of device of identification character area, which is characterized in that including:
Processor;
Memory for storing processor-executable instruction;
Wherein, the processor is configured as:
The characteristic information of images to be recognized is inputted into character area identification model, obtains each first figure in the images to be recognized
As first probability and the second probability in region, the first probability characterization described first image region is the probability of character area,
The probability that second probability characterization described first image region is connected with adjacent image-region, wherein the character area
The identification model character area in image for identification;
The described first image region that first probability is more than to the first probability threshold value is determined as the second image-region;
Two adjacent second image-regions that the second probability threshold value is all higher than to the second probability merge;
According to the image-region after merging, the character area in the images to be recognized is determined.
10. a kind of computer readable storage medium, is stored thereon with computer program instructions, which is characterized in that the program instruction
The step of any one of claim 1-4 the methods are realized when being executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810367675.8A CN108717542B (en) | 2018-04-23 | 2018-04-23 | Method and device for recognizing character area and computer readable storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810367675.8A CN108717542B (en) | 2018-04-23 | 2018-04-23 | Method and device for recognizing character area and computer readable storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108717542A true CN108717542A (en) | 2018-10-30 |
CN108717542B CN108717542B (en) | 2020-09-15 |
Family
ID=63899363
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810367675.8A Active CN108717542B (en) | 2018-04-23 | 2018-04-23 | Method and device for recognizing character area and computer readable storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108717542B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109685055A (en) * | 2018-12-26 | 2019-04-26 | 北京金山数字娱乐科技有限公司 | Text filed detection method and device in a kind of image |
CN109766885A (en) * | 2018-12-29 | 2019-05-17 | 北京旷视科技有限公司 | A kind of character detecting method, device, electronic equipment and storage medium |
CN110321892A (en) * | 2019-06-04 | 2019-10-11 | 腾讯科技(深圳)有限公司 | A kind of picture screening technique, device and electronic equipment |
CN112101308A (en) * | 2020-11-11 | 2020-12-18 | 北京云测信息技术有限公司 | Method and device for combining text boxes based on language model and electronic equipment |
CN117935296A (en) * | 2024-02-06 | 2024-04-26 | 广东度才子集团有限公司 | Employment quality report generation system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102855478A (en) * | 2011-06-30 | 2013-01-02 | 富士通株式会社 | Method and device for positioning text areas in image |
CN105574513A (en) * | 2015-12-22 | 2016-05-11 | 北京旷视科技有限公司 | Character detection method and device |
US20170083772A1 (en) * | 2015-09-18 | 2017-03-23 | Samsung Electronics Co., Ltd. | Apparatus and method for object recognition and for training object recognition model |
-
2018
- 2018-04-23 CN CN201810367675.8A patent/CN108717542B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102855478A (en) * | 2011-06-30 | 2013-01-02 | 富士通株式会社 | Method and device for positioning text areas in image |
US20170083772A1 (en) * | 2015-09-18 | 2017-03-23 | Samsung Electronics Co., Ltd. | Apparatus and method for object recognition and for training object recognition model |
CN105574513A (en) * | 2015-12-22 | 2016-05-11 | 北京旷视科技有限公司 | Character detection method and device |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109685055A (en) * | 2018-12-26 | 2019-04-26 | 北京金山数字娱乐科技有限公司 | Text filed detection method and device in a kind of image |
CN109685055B (en) * | 2018-12-26 | 2021-11-12 | 北京金山数字娱乐科技有限公司 | Method and device for detecting text area in image |
CN109766885A (en) * | 2018-12-29 | 2019-05-17 | 北京旷视科技有限公司 | A kind of character detecting method, device, electronic equipment and storage medium |
CN109766885B (en) * | 2018-12-29 | 2022-01-18 | 北京旷视科技有限公司 | Character detection method and device, electronic equipment and storage medium |
CN110321892A (en) * | 2019-06-04 | 2019-10-11 | 腾讯科技(深圳)有限公司 | A kind of picture screening technique, device and electronic equipment |
CN110321892B (en) * | 2019-06-04 | 2022-12-13 | 腾讯科技(深圳)有限公司 | Picture screening method and device and electronic equipment |
CN112101308A (en) * | 2020-11-11 | 2020-12-18 | 北京云测信息技术有限公司 | Method and device for combining text boxes based on language model and electronic equipment |
CN112101308B (en) * | 2020-11-11 | 2021-02-09 | 北京云测信息技术有限公司 | Method and device for combining text boxes based on language model and electronic equipment |
CN117935296A (en) * | 2024-02-06 | 2024-04-26 | 广东度才子集团有限公司 | Employment quality report generation system |
Also Published As
Publication number | Publication date |
---|---|
CN108717542B (en) | 2020-09-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108717542A (en) | Identify the method, apparatus and computer readable storage medium of character area | |
CN106454336B (en) | The method and device and terminal that detection terminal camera is blocked | |
CN106651955A (en) | Method and device for positioning object in picture | |
EP3133527A1 (en) | Human face recognition method, apparatus and terminal | |
CN106650575A (en) | Face detection method and device | |
CN107239535A (en) | Similar pictures search method and device | |
CN104731688B (en) | Point out the method and device of reading progress | |
CN105205494B (en) | Similar pictures recognition methods and device | |
CN105069426B (en) | Similar pictures judgment method and device | |
CN108062547A (en) | Character detecting method and device | |
CN107944447A (en) | Image classification method and device | |
CN106980840A (en) | Shape of face matching process, device and storage medium | |
CN107392166A (en) | Skin color detection method, device and computer-readable recording medium | |
CN107742120A (en) | The recognition methods of bank card number and device | |
CN105335714B (en) | Photo processing method, device and equipment | |
CN105208284B (en) | Shoot based reminding method and device | |
CN104933700B (en) | A kind of method and apparatus carrying out picture material identification | |
CN107967459A (en) | convolution processing method, device and storage medium | |
CN106598429A (en) | Method and device for adjusting window of mobile terminal | |
CN107911576A (en) | Image processing method, device and storage medium | |
CN105094364B (en) | Vocabulary display methods and device | |
CN104156476B (en) | The method and device of image synthesis | |
CN105426904B (en) | Photo processing method, device and equipment | |
CN108020374A (en) | Atmospheric pressure value determines method and device | |
CN107292306A (en) | Object detection method and device |
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 |