CN104715253A - Method and server for obtaining test question analysis information - Google Patents
Method and server for obtaining test question analysis information Download PDFInfo
- Publication number
- CN104715253A CN104715253A CN201510155055.4A CN201510155055A CN104715253A CN 104715253 A CN104715253 A CN 104715253A CN 201510155055 A CN201510155055 A CN 201510155055A CN 104715253 A CN104715253 A CN 104715253A
- Authority
- CN
- China
- Prior art keywords
- gray
- image
- examination question
- recognition result
- external frame
- 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.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a method for obtaining test question analysis information, which is used for improving the efficiency of obtaining test question analysis information. According to the embodiment, the method comprises the following steps: a server obtains a test question image from a mobile terminal; the server distinguishes test questions in the image to obtain distinguished results; the server finds analysis information corresponding to the test questions according to the distinguished results; and the server returns the analysis information to the mobile terminal. The embodiment of the invention further provides the server for obtaining test question analysis information, which is used for improving the efficiency of obtaining test question analysis information.
Description
Technical field
The present invention relates to data processing field, particularly relate to a kind of method and the server that obtain examination question resolving information.
Background technology
In the middle of daily study, usually adopting the mode doing examination question to consolidate the knowledge of study, along with the development of Internet technology, when running into the examination question that can not answer, obtaining the resolving information corresponding with examination question by internet.
In prior art, in order to obtain the resolving information corresponding with examination question by internet, adopt the following two kinds scheme:
1, utilize mobile phone that examination question is taken into picture, then picture is uploaded to forum by internet and carries out question and answer.
2, input and examination question in keyword, so utilize the information that internet hunt is relevant to the examination question of Keywords matching.
But, in above-mentioned two schemes, scheme 1 needs to wait for artificial treatment, because artifact effect is slow, cannot obtain examination question resolving information in time, scheme 2 needs manually to input test question information search, because artificial input efficiency is low, also will cause cannot obtaining examination question resolving information in time, efficiency is low.
Summary of the invention
The invention provides a kind of method and the server that obtain examination question resolving information, can solve in prior art when adopting internet to obtain examination question correspondence resolving information, artificial input efficiency is low, cannot obtain the problem of examination question resolving information in time.
First aspect present invention provides a kind of method obtaining examination question resolving information, comprising:
Server is from the image of acquisition for mobile terminal examination question;
Described server identifies the examination question in described image, obtains recognition result;
Described server, according to described recognition result, searches the resolving information that described examination question is corresponding;
Described server returns described resolving information to described mobile terminal.
In conjunction with the first aspect of the embodiment of the present invention, in the first implementation of first aspect present invention, examination question identifies in described image, before obtaining recognition result, also comprises:
When determine described image blurring time, again obtain image from described mobile terminal;
And/or,
When determining that described image there occurs rotation, be forward by described Image Adjusting;
And/or,
When the gray level image determining that described image is corresponding needs to carry out gray inversion, gray inversion is carried out to described gray level image.
In conjunction with the first implementation of the first aspect of the embodiment of the present invention, in the second implementation of first aspect present invention, determine described image blurringly to comprise:
The first word edge in described image is determined by gradient operator;
Detect the quantity at described first word edge;
When the pixel quantity at described first word edge is less than threshold value, determine described image blurring;
Or,
The first word edge in described image is determined by gradient operator;
Detect the first gray-scale value and first quantity of the first pixel, described first pixel is the pixel at the first word edge;
Determine the second gray-scale value according to described first gray-scale value, the difference between described first gray-scale value and described second gray-scale value is within the scope of first threshold;
Determine the second pixel that described second gray-scale value is corresponding and the second quantity of described second pixel, described second pixel forms the second word edge;
When the ratio of described first quantity and the second quantity is within the scope of Second Threshold, determine described image blurring.
In conjunction with the first implementation of the first aspect of the embodiment of the present invention, in the third implementation of first aspect present invention, describedly determine that described image there occurs rotation and comprises:
Determine the external frame on described image, described external frame is used for character corresponding to external described examination question;
Determine the horizontal group number in groups of described external frame and longitudinal group number in groups, transversal displacement between adjacent external frame in described transverse direction group number is in groups not more than the half of described external frame height, and the vertical misalignment amount between the adjacent external frame in described longitudinal direction group number is in groups not more than the half of described external frame height;
When the difference of described transverse direction group number in groups and described longitudinal direction group number is in groups in the 3rd threshold range, determine that described image there occurs rotation.
In conjunction with the first implementation of the first aspect of the embodiment of the present invention, in the 4th kind of implementation of first aspect present invention, the described gray-scale map determining that described image is corresponding needs to carry out gray inversion and comprises:
Described gray-scale map is carried out gray inversion and obtain the gray-scale map after reversing;
Determine the external frame quantity of the gray-scale map after described gray-scale map and reversion, described external frame is used for character corresponding to external described examination question;
When the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
In conjunction with the first aspect of the embodiment of the present invention or the first implementation of first aspect, in the 5th kind of implementation of first aspect present invention, examination question identifies in described image, before obtaining recognition result, also comprises:
The character corresponding to examination question on described image is split.
In conjunction with the first aspect of the embodiment of the present invention, in the 6th kind of implementation of first aspect present invention, described examination question in described image to be identified, obtains recognition result and comprise:
Adopt convolutional neural networks or recurrent neural network to described examination question identification, obtain recognition result.
In conjunction with the 6th kind of implementation of the first aspect of the embodiment of the present invention or the 4th kind of implementation of the second implementation of the first implementation of first aspect or first aspect or the third implementation of first aspect or first aspect or five kinds of implementations of first aspect or first aspect, in the 7th kind of implementation of first aspect present invention, examination question in described image is being identified, after obtaining recognition result, according to described recognition result, before searching resolving information corresponding to described examination question, also comprise:
Determine that described recognition result meets natural language model.
In conjunction with the first aspect of the embodiment of the present invention, in the 8th kind of implementation of first aspect present invention, described according to described recognition result, the resolving information searching described examination question corresponding comprises:
Determine the keyword in described recognition result, described keyword for searching resolving information corresponding to described examination question, described keyword comprise adjacent word to and/or adjacent word group;
Utilize the resolving information that described in described keyword lookup, examination question is corresponding.
In conjunction with the first aspect of the embodiment of the present invention, in the 9th kind of implementation of first aspect present invention, describedly determine that the keyword in described recognition result comprises:
Obtain the scoring of each participle in described recognition result;
The highest participle of determining to mark is described keyword.
Embodiment of the present invention second aspect provides a kind of server obtaining examination question resolving information, comprising:
Acquiring unit, for the image from acquisition for mobile terminal examination question;
Recognition unit, for identifying the examination question in described image, obtains recognition result;
Search unit, for according to described recognition result, search the resolving information that described examination question is corresponding;
Processing unit, for returning described resolving information to described mobile terminal.
In conjunction with the second aspect of the embodiment of the present invention, in the first implementation of the present invention second, examination question identifies in described image, before obtaining recognition result, described processing unit also for:
When determine described image blurring time, again obtain image from described mobile terminal;
And/or,
When determining that described image there occurs rotation, be forward by described Image Adjusting;
And/or,
When the gray level image determining that described image is corresponding needs to carry out gray inversion, gray inversion is carried out to described gray level image.
In conjunction with the first implementation of the second aspect of the embodiment of the present invention, in the second implementation of the present invention second, described processing unit also for:
The first word edge in described image is determined by gradient operator;
Detect the quantity at described first word edge;
When the pixel quantity at described first word edge is less than threshold value, determine described image blurring;
Or,
The first word edge in described image is determined by gradient operator;
Detect the first gray-scale value and first quantity of the first pixel, described first pixel is the pixel at the first word edge;
Determine the second gray-scale value according to described first gray-scale value, the difference between described first gray-scale value and described second gray-scale value is within the scope of first threshold;
Determine the second pixel that described second gray-scale value is corresponding and the second quantity of described second pixel, described second pixel forms the second word edge;
When the ratio of described first quantity and the second quantity is within the scope of Second Threshold, determine described image blurring.
In conjunction with the first implementation of the second aspect of the embodiment of the present invention, in the third implementation of the present invention second, described processing unit specifically for:
Determine the external frame on described image, described external frame is used for character corresponding to external described examination question;
Determine the horizontal group number in groups of described external frame and longitudinal group number in groups, transversal displacement between adjacent external frame in described transverse direction group number is in groups not more than the half of described external frame height, and the vertical misalignment amount between the adjacent external frame in described longitudinal direction group number is in groups not more than the half of described external frame height;
When the difference of described transverse direction group number in groups and described longitudinal direction group number is in groups in the 3rd threshold range, determine that described image there occurs rotation.
In conjunction with the first implementation of the second aspect of the embodiment of the present invention, in the 4th kind of implementation of the present invention second, described processing unit specifically for:
Described gray-scale map is carried out gray inversion and obtain the gray-scale map after reversing;
Determine the external frame quantity of the gray-scale map after described gray-scale map and reversion, described external frame is used for character corresponding to external described examination question;
When the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
In conjunction with the first implementation of the second aspect of the embodiment of the present invention or the second implementation of second aspect, in the 5th kind of implementation of the present invention second, examination question identifies in described image, before obtaining recognition result, described processing unit also for:
Equalization processing is carried out to the gray scale of described image;
Described image is carried out binary conversion treatment, splits with the character corresponding to examination question on described image.
In conjunction with the second aspect of the embodiment of the present invention, in the 6th kind of implementation of the present invention second, described recognition unit, specifically for adopting convolutional neural networks or recurrent neural network to described examination question identification, obtains recognition result.
In conjunction with the 6th kind of implementation of the second aspect of the embodiment of the present invention or the 4th kind of implementation of the second implementation of the first implementation of second aspect or second aspect or the third implementation of second aspect or second aspect or five kinds of implementations of second aspect or second aspect, in the 7th kind of implementation of second aspect present invention, examination question in described image is being identified, after obtaining recognition result, according to described recognition result, before searching resolving information corresponding to described examination question, described processing unit is also for determining that described recognition result meets natural language model.
In conjunction with the second aspect of the embodiment of the present invention, in the 8th kind of implementation of the present invention second, described in search unit specifically for:
Determine the keyword in described recognition result, described keyword for searching resolving information corresponding to described examination question, described keyword comprise adjacent word to and/or adjacent word group;
Utilize the resolving information that described in described keyword lookup, examination question is corresponding.
In conjunction with the second aspect of the embodiment of the present invention, in the 9th kind of implementation of the present invention second, described in search unit specifically for:
Obtain the scoring of each participle in described recognition result;
The highest participle of determining to mark is described keyword.
The application embodiment of the present invention has following beneficial effect:
The image of server by getting from mobile terminal identification, obtain recognition result, utilize recognition result from test item bank, search resolving information corresponding to examination question, then resolving information is back to mobile terminal, relatively there is prior art, without the need to waiting for artificial treatment, also without the need to manually inputting test question information, examination question resolving information can be obtained, thus raising efficiency.
Accompanying drawing explanation
Fig. 1 is a kind of embodiment schematic diagram obtaining examination question resolving information method in the embodiment of the present invention;
Fig. 2 is a kind of another embodiment schematic diagram obtaining examination question resolving information method in the embodiment of the present invention;
Fig. 3 is a kind of embodiment schematic diagram obtaining examination question resolving information server in the embodiment of the present invention.
Embodiment
Embodiments provide a kind of method and the relevant apparatus that obtain examination question resolving information, for improving the efficiency obtaining examination question resolving information.
Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those skilled in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
In the embodiment of the present invention, said mobile terminal can be taken pictures and the smart mobile phone of function of surfing the Net or panel computer etc. for having.
Refer to Fig. 1, in the embodiment of the present invention, a kind of embodiment obtaining examination question resolving information comprises:
101, server is from the image of acquisition for mobile terminal examination question.
Be understandable that, mobile terminal can use camera function that examination question is taken into image, and is deposited in the memory device of mobile phone terminal, and server initiatively can obtain this image from mobile phone, also can by mobile phone by image uploading in server.
102, described server identifies the examination question in described image, obtains recognition result.
Wherein, server can adopt optical recognition to identify the examination question in image, thus obtains recognition result, can comprise writing text information corresponding to examination question or illustration information in this recognition result.
103, described server is according to described recognition result, searches the resolving information that described examination question is corresponding.
Be understandable that, the database storing examination question resolving information is set in server, when server is after obtaining recognition result to the identification of image, this recognition result can be utilized from database, search the resolving information that examination question is corresponding.
104, described server returns described resolving information to described mobile terminal.
Be understandable that, this resolving information, after finding out resolving information corresponding to this examination question, can be returned to mobile terminal by server.
In the embodiment of the present invention, the image of server by getting from mobile terminal identification, obtain recognition result, utilize recognition result from test item bank, search resolving information corresponding to examination question, then resolving information is back to mobile terminal, relatively has prior art, without the need to waiting for artificial treatment, also without the need to manually inputting test question information, examination question resolving information can be obtained, thus raising efficiency.
Below in conjunction with Fig. 2 in the embodiment of the present invention, a kind of another embodiment obtaining the method for examination question resolving information is described, and specifically comprises:
201, server is from the image of acquisition for mobile terminal examination question.
The concrete manner of execution of the step 201 in the present embodiment can step 101 in reference diagram 1 embodiment, repeats no more herein.
Alternatively, server is identifying the examination question in image, before obtaining recognition result, can carry out pre-service to this image.
202, pre-service is carried out to image.
In order to make server identify image sooner, more accurately, pre-service can be carried out to image, wherein pretreated mode being carried out to image and comprising at least following a kind of:
One, when determine described image blurring time, again obtain image from described mobile terminal.
Be understandable that, by the number of the pixel at word edge in image, can determine that whether image is fuzzy, alternatively, determine by following steps image blurring:
A, the first word edge determined by gradient operator in described image;
Wherein, gradient operator can be Laplace operator (Laplacian derivatives), utilizes partial gradient value in Laplace operator detected image, is word edge by the part of Grad first threshold.
B, detect the quantity at described first word edge.
C, when the pixel quantity at described first word edge is less than threshold value, determine described image blurring.
Alternatively, also determine by following steps image blurring:
A, the first word edge determined by gradient operator in described image.
B, the first gray-scale value detecting the first pixel and the first quantity, described first pixel is the pixel at the first word edge.
C, determine the second gray-scale value according to described first gray-scale value, the difference between described first gray-scale value and described second gray-scale value is within the scope of first threshold.
D, determine the second pixel that described second gray-scale value is corresponding and the second quantity of described second pixel, described second pixel forms the second word edge.
E, when the ratio of described first quantity and the second quantity is within the scope of Second Threshold, determine described image blurring.
Two, when determining that described image there occurs rotation, be forward by described Image Adjusting.
Main detection be image whether occur 90 degree rotate or 180 degree of rotations.Wherein, can determine that image there occurs rotation as follows:
A, the external frame determined on described image, described external frame is used for character corresponding to external described examination question.
Wherein, first can do binary conversion treatment to image, the binaryzation of image, sets threshold value exactly, and the gray-scale value of the pixel in image is converted into 0 or 255; Because character is prospect, thus gray-scale value is less than threshold value transfer 255 to, and the pixel that gray-scale value is greater than threshold value becomes 0.For prospect character, the region that pixel value is 255 can be extracted, this region is connected region, thus the external frame for this connected region external can be determined, then by length breadth ratio and the area of the external frame of connected region, from the external frame of connected region, the external frame for external character is determined.
B, determine described external frame laterally group number in groups and longitudinal group number in groups, transversal displacement between adjacent external frame in described transverse direction group number is in groups not more than the half of described external frame height, and the vertical misalignment amount between the adjacent external frame in described longitudinal direction group number is in groups not more than the half of described external frame height.
With the upper left corner of the external frame of character for initial point, being laterally to the right x-axis, is longitudinally y-axis downwards.First external frame is polymerized to row by y coordinate, checks adjacent external frame line by line in order, if adjacent external frame side-play amount is greater than the half of external frame height, then think that original row terminates, external frame starts new one group thus.The horizontal group number in groups of all external frames can be obtained like this.
In like manner, external frame is polymerized to row by x coordinate, checks adjacent external frame by column according to the order of sequence, if side-play amount is greater than the half of external width of frame, then thinks that former row terminate, newly arrange beginning.Obtain the longitudinal group number in groups of all external frames thus.
C, when the difference of described transverse direction group number in groups and described longitudinal direction group number is in groups in the 3rd threshold range, determine that described image there occurs rotation.
Transverse direction group number is in groups compared with longitudinal group number in groups, obtains transverse direction group number in groups and the difference of described longitudinal direction group number in groups, when this difference is in the 3rd threshold range, then determine that described image there occurs rotation.
Three, when the gray level image determining that described image is corresponding needs to carry out gray inversion, gray inversion is carried out to described gray level image.
Be understandable that, on gray level image, word is generally dark, and background is light color.But under special circumstances, when such as taking the exercise question on blackboard, word is white, and background is dark.Therefore need the color detecting font, reverse gray level image if desired, wherein, can determine that the gray level image that image is corresponding needs to carry out gray inversion as follows:
A, described gray-scale map carried out gray inversion and obtain the gray-scale map after reversing.
Wherein, gray-scale map obtains after can carrying out binary conversion treatment by image, and gray-scale map is the image presenting black and white effect, and described reversion is exchanged by the gray-scale value of the pixel in gray-scale map.
B, determine the external frame quantity of gray-scale map after described gray-scale map and reversion, described external frame is used for character corresponding to external described examination question.
The external frame former figure gray level image and the gray level image after turning being carried out to the connected domain of prospect detects, and obtains the quantity of external frame on the gray-scale map after the quantity of external frame on gray-scale map and reversion.
C, when the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
The quantity of external frame on gray-scale map after the quantity of frame external on the gray-scale map obtained and reversion is compared, can detect that owing to there being the gray-scale map of normal font color more multiword accords with external frame, according to this logic, when the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, then determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
Alternatively, in this enforcement, step 203 can also be comprised after step 202.
203, corresponding to examination question on described image character is split.
After picture pre-service, can start to carry out the character cutting of picture, the cutting of character is the correct basis identified, character cutting needs on the basis of image binaryzation more accurately, find connected domain as initial cutting result, then connected domain be arranged in rows, the information of recycling row carries out the correction of cutting, can obtain final cutting result, concrete step is as follows:
A, equalization processing is carried out to the gray-scale value of pixel in image.
Local gray level equalization can make image carry out binaryzation more accurately, and the formula of local equalization is:
By the distributed areas of pixel value in former figure by [p
min, p
max] be transformed into [p
min, 255], the equalization of local make use of the intensity profile of the next balanced local of pixel value statistical information of local, avoid uneven illumination even time overall situation distribution and the inconsistent and equilibrium result of mistake that causes of local distribution, too small for some regional area, and have too concentrated pixel value distribution (as one piece of pure background or pure character portion), the distribution that we can use its peripheral region replaces, distribution that is that cause is biased because adding up not enough to avoid it, multi-level local gray level equalization can be from coarse to fine to former figure, divide less regional area from level to level, after all local gray level equalization is carried out to every block regional area, results on average all again obtains final equalization result.
B, result according to gray balance, carry out binary conversion treatment to image.
After carrying out equalization processing, obtain original character seed by Laplace operator.From these seeds, the average pixel value of character is estimated to calculate.From remaining background area, we can estimate the average pixel value of background equally.Can threshold value be calculated from these two values, split the character made new advances for the region of search in certain limit around character seed.Character segmentation result after renewal more accurately and fully can obtain character portion.Through several take turns renewal after, final Character segmentation result will converge on real character position, and binaryzation is accomplished.
Also to revise according to their respective rules further for Chinese character and mathematic sign.We can make introductions all round concrete method below.
C, the result of binaryzation is utilized to carry out Character segmentation.
Be understandable that, to the order that the character in image cuts be: first the character of line of text cut, then the Chinese character in line of text cut, finally the formula in line of text cut.
Respectively each step cutting is introduced below.
1, the character of line of text is cut.
The result of binaryzation tentatively can provide the connected domain of each character, and based on connected domain, each character can by primary segmentation; Primary segmentation obtains the Characters Stuck phenomenon that line of text there will be mistake, therefore the character of initial segmentation first can be formed line of text, recycling line of text Optimized Segmentation.
After obtaining line of text, solve the problem of adhesion by crossing cutting.Cross cutting and can split each line of text more equably.In general, when block length with height large percentage time, then determine the phenomenon creating adhesion.For such situation, current text guild reexamines possible cut-off.The scoring of cut-off is based on the close degree of bottom profiled and the distance of distance front and back cut-off on character in the number of pixels of vertical projection, vertical direction.Carry out cutting again at most probable cut-off, can adhesion problems be solved.
2, to the cutting of Chinese character.
Because most of Chinese character is all made up of multiple connected domain, and Chinese character has the feature of Chinese characters, according to the feature of Chinese character, may be used for merging the single character that these Chinese character fragments become complete, particularly, we search for this optimum character fragments by dynamic programming algorithm and combine.
3, the carrying out of formula is cut.
Formula comprises complicated structure, such as subscript, subscript, fraction structure, the structure such as to comprise, and may inter-bank.Formula segmentation not only will obtain each symbol, and will construct the structural relation between symbol.Each symbol can obtain based on its connected domain.And the structure of structure, then more complex.The method that we build is constantly horizontal and vertical projection.Formula, first by longitudinal projection, can be divided into each several part with horizontal neighbouring relations according to result.Then, then carry out transverse projection, identify the immanent structure of the part of multiply connected domain further.With equal sign and radical sign for example, in transverse projection, equal sign identifiable design goes out upper and lower two parts, but it does not meet the Up-Center-Down Structure of fraction, so will not split further.And radical sign and the connected domain under it are also indivisible in transverse projection, can judge that it is " comprising " structure thus.Its part comprised carries out longitudinal direction, transverse projection further.Thus, the structure of formula is built, and each leaf node is independently symbol.
204, described server identifies the examination question in described image, obtains recognition result.
Alternatively, server, when identifying the examination question in image, can adopt degree of depth learning art to identify, particularly, convolutional neural networks and recurrent neural network can be adopted to identify.
Alternatively, in order to promote the speed of identification, can Force Integrated Graphics Processor (IGP) Nforce (GPU) in the server, utilize GPU to identify.
Alternatively, step 205 can also be comprised in the present embodiment.
205, determine that described recognition result meets natural language model.
Be understandable that, after server identifies the examination question in image, natural language model can also be applied and this recognition result is verified, when determining that recognition result does not meet natural language model, recognition result is corrected, thus promotes recognition correct rate.
Wherein, natural language model can be ternary language model, compare particular by the rule set in the contextual information in recognition result and natural language model, when the contextual information in recognition result is identical with the rule set in natural language model, then determine that recognition result meets natural language model.
206, described server is according to described recognition result, searches the resolving information that described examination question is corresponding.
Be understandable that, in service, be provided with the database of test question information, after obtaining recognition result, this recognition result can be utilized in resolving information storehouse to search the resolving information corresponding with this examination question.
Wherein, server can determine the keyword in described recognition result, described keyword comprise adjacent word to and/or adjacent word group, then utilize the resolving information that described in described keyword lookup, examination question is corresponding.
Alternatively, server is determining that the keyword mode in recognition result is: the scoring obtaining each participle in described recognition result, and the highest participle of then determining to mark is described keyword.
Wherein, the formula obtaining the scoring of each participle in recognition result is:
Wherein, wherein score represents final scoring, and base represents basic score, boost
irepresent the weight score according to presetting rule, u
irepresent other weighted term scores.
Wherein, weight score is from weighted term, and main weighted term comprises the weighting of maximum coupling subsequence, Window match, formula coupling weighting and figure and searches figure weighting.
Be understandable that, in the exercise question and exam pool of inquiry, exercise question presses the situation of identical sequence order coupling.Suppose that inquiry exercise question is by sequence participle { x
1, x
2..., x
nrepresent, and a retrieving information in exam pool is with sequence participle { y
1, y
2... y
mrepresent.With d (x
i, y
j) represent as alignment x
iand y
jtime, the score of maximum coupling subsequence before this.Therefore, this score can according to method below, according to x
iand y
jwhether identical and iterative computation:
Wherein be weight according to text type.After having calculated, and take back in above formula.
A part for entirely and just complete exercise question likely do not clapped by exercise question picture due to user's shooting, so be used based on the coupling assessment of moving window.If the text size of inquiry is l, with a wide window for 1.5l shiding matching on exercise question in exam pool of the distance of 0.5l.In every section of window, above formula is followed in the calculating of coupling, selects the highest window score as boost
2carry out the basic score of weighting.
In exercise question, formula is often more important than text, and therefore, we will to formula partial weighting score.The method of weighting does adjacent word to the participle rolled into a ball with adjacent word to formula part, and one is used from and does matching primitives with the above-mentioned formula of the participle of the formula of related topic in exam pool.
Wherein, if the exercise question of inquiry contains illustration, the score of the mode weighting illustration to scheme to search figure is also needed.Illustration is using the vector representation based on image characteristics extraction as fingerprint, and in exam pool, the illustration of related topic is retrieved, and the distance both calculating with Hamming distance, distance is nearer, higher to the weighting of former basic score.
In the embodiment of the present invention, server, before carrying out image recognition, first carries out pre-service to image, the effect promoting image recognition efficiency and image recognition accuracy can be played, such as when determining image blurring, reappearing and obtaining image, blurred picture can be avoided to affect the accuracy identified; When carrying out image recognition, adopting degree of depth learning method to identify, further, the GPU in server can be adopted to identify, the effect promoting recognition efficiency can be played; Before image recognition, the character in image is split, reduce the difficulty to image recognition, thus promote recognition correct rate; Adopt adjacent word to or adjacent word group carry out searching of resolving information as keyword, can play and promote the effect of resolving information of searching.
Above a kind of method obtaining examination question resolving information in the embodiment of the present invention is described, below a kind of server obtaining examination question resolving information in the embodiment of the present invention is described, refer to Fig. 3, in the embodiment of the present invention, a kind of server embodiment obtaining examination question resolving information comprises:
Acquiring unit 301, for the image from acquisition for mobile terminal examination question;
Recognition unit 302, for identifying the examination question in described image, obtains recognition result;
Search unit 303, for according to described recognition result, search the resolving information that described examination question is corresponding;
Processing unit 304, for returning described resolving information to described mobile terminal.
Optionally, on the basis of the embodiment corresponding to above-mentioned Fig. 1, in another one embodiment of the present invention:
Wherein, examination question identifies in described image, before obtaining recognition result, described processing unit 304 also for: when determine described image blurring time, again obtain image from described mobile terminal; And/or, when determining that described image there occurs rotation, be forward by described Image Adjusting; And/or, when the gray level image determining that described image is corresponding needs to carry out gray inversion, gray inversion is carried out to described gray level image.
Alternatively, described processing unit 304 is determined described image blurring, determines the first word edge in described image especially by gradient operator; Detect the quantity at described first word edge; When the pixel quantity at described first word edge is less than threshold value, determine described image blurring;
Alternatively, described processing unit 304 is determined described image blurring, determines the first word edge in described image especially by gradient operator; Detect the first gray-scale value and first quantity of the first pixel, described first pixel is the pixel at the first word edge; Determine the second gray-scale value according to described first gray-scale value, the difference between described first gray-scale value and described second gray-scale value is within the scope of first threshold; Determine the second pixel that described second gray-scale value is corresponding and the second quantity of described second pixel, described second pixel forms the second word edge; When the ratio of described first quantity and the second quantity is within the scope of Second Threshold, determine described image blurring.
Alternatively, described processing unit 304 determines that described image there occurs rotation, and especially by the external frame determined on described image, described external frame is used for character corresponding to external described examination question; Determine the horizontal group number in groups of described external frame and longitudinal group number in groups, transversal displacement between adjacent external frame in described transverse direction group number is in groups not more than the half of described external frame height, and the vertical misalignment amount between the adjacent external frame in described longitudinal direction group number is in groups not more than the half of described external frame height; When the difference of described transverse direction group number in groups and described longitudinal direction group number is in groups in the 3rd threshold range, determine that described image there occurs rotation.
Alternatively, described processing unit 304 determines that described image needs to carry out gray inversion, obtains the gray-scale map after reversing especially by described gray-scale map being carried out gray inversion; Determine the external frame quantity of the gray-scale map after described gray-scale map and reversion, described external frame is used for character corresponding to external described examination question; When the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
Alternatively, described processing unit 304 examination question in described image identifies, before obtaining recognition result, also for carrying out equalization processing to the gray scale of described image, and described image is carried out binary conversion treatment, split with the character corresponding to examination question on described image.
Alternatively, described recognition unit 302, specifically for adopting convolutional neural networks or recurrent neural network to described examination question identification, obtains recognition result.
Alternatively, described processing unit 304, is identifying the examination question in described image, after obtaining recognition result, according to described recognition result, before searching resolving information corresponding to described examination question, also for determining that described recognition result meets natural language model.
Alternatively, described in search unit 303 specifically for: determine the keyword in described recognition result, described keyword for searching resolving information corresponding to described examination question, described keyword comprise adjacent word to and/or adjacent word group; Utilize the resolving information that described in described keyword lookup, examination question is corresponding.
Alternatively, described in search the keyword that unit 303 determines in described recognition result, especially by the scoring obtaining each participle in described recognition result, and the highest participle of determining to mark is described keyword.
Those skilled in the art can be well understood to, and for convenience and simplicity of description, the system of foregoing description, the specific works process of device and unit, with reference to the corresponding process in preceding method embodiment, can not repeat them here.
In several embodiments that the application provides, should be understood that, disclosed system, apparatus and method, can realize by another way.Such as, device embodiment described above is only schematic, such as, the division of described unit, be only a kind of logic function to divide, actual can have other dividing mode when realizing, such as multiple unit or assembly can in conjunction with or another system can be integrated into, or some features can be ignored, or do not perform.Another point, shown or discussed coupling each other or direct-coupling or communication connection can be by some interfaces, and the indirect coupling of device or unit or communication connection can be electrical, machinery or other form.
The described unit illustrated as separating component or can may not be and physically separates, and the parts as unit display can be or may not be physical location, namely can be positioned at a place, or also can be distributed in multiple network element.Some or all of unit wherein can be selected according to the actual needs to realize the object of the present embodiment scheme.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, also can be that the independent physics of unit exists, also can two or more unit in a unit integrated.Above-mentioned integrated unit both can adopt the form of hardware to realize, and the form of SFU software functional unit also can be adopted to realize.
If described integrated unit using the form of SFU software functional unit realize and as independently production marketing or use time, can be stored in a computer read/write memory medium.Based on such understanding, the part that technical scheme of the present invention contributes to prior art in essence in other words or all or part of of this technical scheme can embody with the form of software product, this computer software product is stored in a storage medium, comprising some instructions in order to make a computer equipment (can be personal computer, server, or the network equipment etc.) perform all or part of step of method described in each embodiment of the present invention.And aforesaid storage medium comprises: USB flash disk, portable hard drive, ROM (read-only memory) (ROM, Read-OnlyMemory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. various can be program code stored medium.
The above, above embodiment only in order to technical scheme of the present invention to be described, is not intended to limit; Although with reference to previous embodiment to invention has been detailed description, those of ordinary skill in the art is to be understood that: it still can be modified to the technical scheme described in foregoing embodiments, or carries out equivalent replacement to wherein portion of techniques feature; And these amendments or replacement, do not make the essence of appropriate technical solution depart from the spirit and scope of various embodiments of the present invention technical scheme.
Claims (20)
1. obtain a method for examination question resolving information, it is characterized in that, comprising:
Server is from the image of acquisition for mobile terminal examination question;
Described server identifies the examination question in described image, obtains recognition result;
Described server, according to described recognition result, searches the resolving information that described examination question is corresponding;
Described server returns described resolving information to described mobile terminal.
2. method according to claim 1, is characterized in that, examination question identifies in described image, before obtaining recognition result, also comprises:
When determine described image blurring time, again obtain image from described mobile terminal;
And/or,
When determining that described image there occurs rotation, be forward by described Image Adjusting;
And/or,
When the gray level image determining that described image is corresponding needs to carry out gray inversion, gray inversion is carried out to described gray level image.
3. method according to claim 2, is characterized in that, determines described image blurringly to comprise:
The first word edge in described image is determined by gradient operator;
Detect the quantity at described first word edge;
When the pixel quantity at described first word edge is less than threshold value, determine described image blurring;
Or,
The first word edge in described image is determined by gradient operator;
Detect the first gray-scale value and first quantity of the first pixel, described first pixel is the pixel at the first word edge;
Determine the second gray-scale value according to described first gray-scale value, the difference between described first gray-scale value and described second gray-scale value is within the scope of first threshold;
Determine the second pixel that described second gray-scale value is corresponding and the second quantity of described second pixel, described second pixel forms the second word edge;
When the ratio of described first quantity and the second quantity is within the scope of Second Threshold, determine described image blurring.
4. method according to claim 2, is characterized in that, describedly determines that described image there occurs rotation and comprises:
Determine the external frame on described image, described external frame is used for character corresponding to external described examination question;
Determine the horizontal group number in groups of described external frame and longitudinal group number in groups, transversal displacement between adjacent external frame in described transverse direction group number is in groups not more than the half of described external frame height, and the vertical misalignment amount between the adjacent external frame in described longitudinal direction group number is in groups not more than the half of described external frame height;
When the difference of described transverse direction group number in groups and described longitudinal direction group number is in groups in the 3rd threshold range, determine that described image there occurs rotation.
5. method according to claim 2, is characterized in that, the described gray-scale map determining that described image is corresponding needs to carry out gray inversion and comprises:
Described gray-scale map is carried out gray inversion and obtain the gray-scale map after reversing;
Determine the external frame quantity of the gray-scale map after described gray-scale map and reversion, described external frame is used for character corresponding to external described examination question;
When the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
6. method according to claim 1 and 2, is characterized in that, examination question identifies in described image, before obtaining recognition result, also comprises:
The character corresponding to examination question on described image is split.
7. method according to claim 1, is characterized in that, describedly identifies the examination question in described image, obtains recognition result and comprises:
Adopt convolutional neural networks or recurrent neural network to described examination question identification, obtain recognition result.
8. method according to any one of claim 1 to 7, is characterized in that, is identifying the examination question in described image, after obtaining recognition result, according to described recognition result, before searching resolving information corresponding to described examination question, also comprises:
Determine that described recognition result meets natural language model.
9. method according to claim 1, is characterized in that, described according to described recognition result, and the resolving information searching described examination question corresponding comprises:
Determine the keyword in described recognition result, described keyword for searching resolving information corresponding to described examination question, described keyword comprise adjacent word to and/or adjacent word group;
Utilize the resolving information that described in described keyword lookup, examination question is corresponding.
10. method according to claim 9, is characterized in that, describedly determines that the keyword in described recognition result comprises:
Obtain the scoring of each participle in described recognition result;
The highest participle of determining to mark is described keyword.
11. 1 kinds of servers obtaining examination question resolving information, is characterized in that, comprising:
Acquiring unit, for the image from acquisition for mobile terminal examination question;
Recognition unit, for identifying the examination question in described image, obtains recognition result;
Search unit, for according to described recognition result, search the resolving information that described examination question is corresponding;
Processing unit, for returning described resolving information to described mobile terminal.
12. servers according to claim 11, is characterized in that, examination question identifies in described image, before obtaining recognition result, described processing unit also for:
When determine described image blurring time, again obtain image from described mobile terminal;
And/or,
When determining that described image there occurs rotation, be forward by described Image Adjusting;
And/or,
When the gray level image determining that described image is corresponding needs to carry out gray inversion, gray inversion is carried out to described gray level image.
13. servers according to claim 11, is characterized in that, described processing unit also for:
The first word edge in described image is determined by gradient operator;
Detect the quantity at described first word edge;
When the pixel quantity at described first word edge is less than threshold value, determine described image blurring;
Or,
The first word edge in described image is determined by gradient operator;
Detect the first gray-scale value and first quantity of the first pixel, described first pixel is the pixel at the first word edge;
Determine the second gray-scale value according to described first gray-scale value, the difference between described first gray-scale value and described second gray-scale value is within the scope of first threshold;
Determine the second pixel that described second gray-scale value is corresponding and the second quantity of described second pixel, described second pixel forms the second word edge;
When the ratio of described first quantity and the second quantity is within the scope of Second Threshold, determine described image blurring.
14. servers according to claim 12, is characterized in that, described processing unit specifically for:
Determine the external frame on described image, described external frame is used for character corresponding to external described examination question;
Determine the horizontal group number in groups of described external frame and longitudinal group number in groups, transversal displacement between adjacent external frame in described transverse direction group number is in groups not more than the half of described external frame height, and the vertical misalignment amount between the adjacent external frame in described longitudinal direction group number is in groups not more than the half of described external frame height;
When the difference of described transverse direction group number in groups and described longitudinal direction group number is in groups in the 3rd threshold range, determine that described image there occurs rotation.
15. servers according to claim 12, is characterized in that, described processing unit specifically for:
Described gray-scale map is carried out gray inversion and obtain the gray-scale map after reversing;
Determine the external frame quantity of the gray-scale map after described gray-scale map and reversion, described external frame is used for character corresponding to external described examination question;
When the quantity of external frame is greater than the quantity of external frame on described gray-scale map on the gray-scale map after reversing, determine that the gray-scale map that described image is corresponding needs to carry out gray inversion.
16. servers according to claim 11 or 12 are put, and it is characterized in that, examination question identifies in described image, before obtaining recognition result, described processing unit also for:
Equalization processing is carried out to the gray scale of described image;
Described image is carried out binary conversion treatment, splits with the character corresponding to examination question on described image.
17. servers according to claim 11, is characterized in that, described recognition unit, specifically for adopting convolutional neural networks or recurrent neural network to described examination question identification, obtains recognition result.
18. according to claim 11 to the server according to any one of 17, it is characterized in that, examination question in described image is being identified, after obtaining recognition result, according to described recognition result, before searching resolving information corresponding to described examination question, described processing unit is also for determining that described recognition result meets natural language model.
19. servers according to claim 11, is characterized in that, described in search unit specifically for:
Determine the keyword in described recognition result, described keyword for searching resolving information corresponding to described examination question, described keyword comprise adjacent word to and/or adjacent word group;
Utilize the resolving information that described in described keyword lookup, examination question is corresponding.
20. servers according to claim 19, is characterized in that, described in search unit specifically for:
Obtain the scoring of each participle in described recognition result;
The highest participle of determining to mark is described keyword.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510155055.4A CN104715253A (en) | 2015-04-02 | 2015-04-02 | Method and server for obtaining test question analysis information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510155055.4A CN104715253A (en) | 2015-04-02 | 2015-04-02 | Method and server for obtaining test question analysis information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104715253A true CN104715253A (en) | 2015-06-17 |
Family
ID=53414563
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510155055.4A Pending CN104715253A (en) | 2015-04-02 | 2015-04-02 | Method and server for obtaining test question analysis information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104715253A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105677872A (en) * | 2016-01-08 | 2016-06-15 | 广东小天才科技有限公司 | Question searching method and device and learning equipment |
| CN105975552A (en) * | 2016-04-29 | 2016-09-28 | 广东小天才科技有限公司 | Topic search method and apparatus for intelligent device |
| CN105975550A (en) * | 2016-04-29 | 2016-09-28 | 广东小天才科技有限公司 | Examination question search method and device of intelligent device |
| CN107403130A (en) * | 2017-04-19 | 2017-11-28 | 北京粉笔未来科技有限公司 | A kind of character identifying method and character recognition device |
| CN107609195A (en) * | 2017-10-18 | 2018-01-19 | 广东小天才科技有限公司 | One kind searches topic method and device |
| CN108932508A (en) * | 2018-08-13 | 2018-12-04 | 杭州大拿科技股份有限公司 | A kind of topic intelligent recognition, the method and system corrected |
| CN110084099A (en) * | 2019-03-14 | 2019-08-02 | 北京暄暄科技有限公司 | Information feedback method and system based on image recognition technology |
| CN110163204A (en) * | 2019-04-15 | 2019-08-23 | 平安国际智慧城市科技股份有限公司 | Businessman's monitoring and managing method, device and storage medium based on image recognition |
| CN111666474A (en) * | 2019-03-08 | 2020-09-15 | 小船出海教育科技（北京）有限公司 | Method and terminal for searching questions in whole page |
| CN111814798A (en) * | 2020-07-14 | 2020-10-23 | 深圳中兴网信科技有限公司 | Method for digitizing titles and readable storage medium |
| CN111915635A (en) * | 2020-08-21 | 2020-11-10 | 广州云蝶科技有限公司 | Test question analysis information generation method and system supporting self-examination paper marking |
| CN112818947A (en) * | 2016-12-15 | 2021-05-18 | 网易（杭州）网络有限公司 | Test paper image processing method, test paper image processing device and computer readable storage medium |
| CN114333338A (en) * | 2020-09-30 | 2022-04-12 | 瑞昱新加坡有限公司 | License plate identification system and method |
| CN114723721A (en) * | 2022-04-18 | 2022-07-08 | 贝塔通科技(北京)有限公司 | Image definition determining method and system and image uploading method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060252023A1 (en) * | 2005-05-03 | 2006-11-09 | Lexmark International, Inc. | Methods for automatically identifying user selected answers on a test sheet |
| CN101122952A (en) * | 2007-09-21 | 2008-02-13 | 北京大学 | Picture words detecting method |
| CN102045503A (en) * | 2009-10-15 | 2011-05-04 | 索尼公司 | Information processing apparatus, display control method, and display control program |
| CN103678637A (en) * | 2013-12-19 | 2014-03-26 | 北京快乐学网络科技有限公司 | Method and device for acquiring test question information |
| CN103839062A (en) * | 2014-03-11 | 2014-06-04 | 东方网力科技股份有限公司 | Image character positioning method and device |
| CN103914567A (en) * | 2014-04-23 | 2014-07-09 | 北京奇虎科技有限公司 | Objective test question answer matching method and objective test question answer matching device |
| CN103927552A (en) * | 2014-04-23 | 2014-07-16 | 北京奇虎科技有限公司 | Method and device for matching answers of target test questions |
| CN103955525A (en) * | 2014-05-09 | 2014-07-30 | 北京奇虎科技有限公司 | Method and client for searching answer to test question |
-
2015
- 2015-04-02 CN CN201510155055.4A patent/CN104715253A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060252023A1 (en) * | 2005-05-03 | 2006-11-09 | Lexmark International, Inc. | Methods for automatically identifying user selected answers on a test sheet |
| CN101122952A (en) * | 2007-09-21 | 2008-02-13 | 北京大学 | Picture words detecting method |
| CN102045503A (en) * | 2009-10-15 | 2011-05-04 | 索尼公司 | Information processing apparatus, display control method, and display control program |
| CN103678637A (en) * | 2013-12-19 | 2014-03-26 | 北京快乐学网络科技有限公司 | Method and device for acquiring test question information |
| CN103839062A (en) * | 2014-03-11 | 2014-06-04 | 东方网力科技股份有限公司 | Image character positioning method and device |
| CN103914567A (en) * | 2014-04-23 | 2014-07-09 | 北京奇虎科技有限公司 | Objective test question answer matching method and objective test question answer matching device |
| CN103927552A (en) * | 2014-04-23 | 2014-07-16 | 北京奇虎科技有限公司 | Method and device for matching answers of target test questions |
| CN103955525A (en) * | 2014-05-09 | 2014-07-30 | 北京奇虎科技有限公司 | Method and client for searching answer to test question |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105677872A (en) * | 2016-01-08 | 2016-06-15 | 广东小天才科技有限公司 | Question searching method and device and learning equipment |
| CN105677872B (en) * | 2016-01-08 | 2019-04-19 | 广东小天才科技有限公司 | A kind of topic searching method, topic searcher and facility for study |
| CN105975552B (en) * | 2016-04-29 | 2020-01-03 | 广东小天才科技有限公司 | Question searching method and device of intelligent equipment |
| CN105975552A (en) * | 2016-04-29 | 2016-09-28 | 广东小天才科技有限公司 | Topic search method and apparatus for intelligent device |
| CN105975550A (en) * | 2016-04-29 | 2016-09-28 | 广东小天才科技有限公司 | Examination question search method and device of intelligent device |
| CN105975550B (en) * | 2016-04-29 | 2020-01-14 | 广东小天才科技有限公司 | Question searching method and device of intelligent equipment |
| CN112818947A (en) * | 2016-12-15 | 2021-05-18 | 网易（杭州）网络有限公司 | Test paper image processing method, test paper image processing device and computer readable storage medium |
| CN107403130A (en) * | 2017-04-19 | 2017-11-28 | 北京粉笔未来科技有限公司 | A kind of character identifying method and character recognition device |
| CN107609195A (en) * | 2017-10-18 | 2018-01-19 | 广东小天才科技有限公司 | One kind searches topic method and device |
| CN108932508A (en) * | 2018-08-13 | 2018-12-04 | 杭州大拿科技股份有限公司 | A kind of topic intelligent recognition, the method and system corrected |
| CN108932508B (en) * | 2018-08-13 | 2022-03-18 | 杭州大拿科技股份有限公司 | Method and system for intelligently identifying and correcting subjects |
| US11508251B2 (en) | 2018-08-13 | 2022-11-22 | Hangzhou Dana Technology Inc. | Method and system for intelligent identification and correction of questions |
| CN111666474A (en) * | 2019-03-08 | 2020-09-15 | 小船出海教育科技（北京）有限公司 | Method and terminal for searching questions in whole page |
| CN111666474B (en) * | 2019-03-08 | 2023-08-25 | 北京百舸飞驰科技有限公司 | Whole page question searching method and terminal |
| CN110084099A (en) * | 2019-03-14 | 2019-08-02 | 北京暄暄科技有限公司 | Information feedback method and system based on image recognition technology |
| CN110163204A (en) * | 2019-04-15 | 2019-08-23 | 平安国际智慧城市科技股份有限公司 | Businessman's monitoring and managing method, device and storage medium based on image recognition |
| CN111814798A (en) * | 2020-07-14 | 2020-10-23 | 深圳中兴网信科技有限公司 | Method for digitizing titles and readable storage medium |
| CN111915635A (en) * | 2020-08-21 | 2020-11-10 | 广州云蝶科技有限公司 | Test question analysis information generation method and system supporting self-examination paper marking |
| CN114333338A (en) * | 2020-09-30 | 2022-04-12 | 瑞昱新加坡有限公司 | License plate identification system and method |
| CN114723721A (en) * | 2022-04-18 | 2022-07-08 | 贝塔通科技(北京)有限公司 | Image definition determining method and system and image uploading method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104715253A (en) | Method and server for obtaining test question analysis information | |
| CN111325203B (en) | American license plate recognition method and system based on image correction | |
| CN107403130A (en) | A kind of character identifying method and character recognition device | |
| RU2661750C1 (en) | Symbols recognition with the use of artificial intelligence | |
| CN107305630B (en) | Text sequence identification method and device | |
| US9697444B2 (en) | Convolutional-neural-network-based classifier and classifying method and training methods for the same | |
| US20190180154A1 (en) | Text recognition using artificial intelligence | |
| US9042648B2 (en) | Salient object segmentation | |
| CN110647829A (en) | Bill text recognition method and system | |
| CN105868758A (en) | Method and device for detecting text area in image and electronic device | |
| CN104463101A (en) | Answer recognition method and system for textual test question | |
| CN105631482A (en) | Convolutional neural network model-based dangerous object image classification method | |
| CN107085726A (en) | Oracle bone rubbing individual character localization method based on multi-method denoising and connected component analysis | |
| CN112418216A (en) | Method for detecting characters in complex natural scene image | |
| CN111626297A (en) | Character writing quality evaluation method and device, electronic equipment and recording medium | |
| CN113537227A (en) | Structured text recognition method and system | |
| CN110659398A (en) | Visual question-answering method based on mathematical chart data set | |
| CN113111716A (en) | Remote sensing image semi-automatic labeling method and device based on deep learning | |
| CN112801092B (en) | Method for detecting character elements in natural scene image | |
| CN111104941A (en) | Image direction correcting method and device and electronic equipment | |
| CN111832497B (en) | Text detection post-processing method based on geometric features | |
| CN116258861B (en) | Semi-supervised semantic segmentation method and segmentation device based on multi-label learning | |
| US11132572B2 (en) | Method and system for splicing and restoring shredded paper based on extreme learning machine | |
| JP7293658B2 (en) | Information processing device, information processing method and program | |
| Fornés et al. | The ICDAR/GREC 2013 music scores competition: Staff removal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| AD01 | Patent right deemed abandoned | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20190326 |