Summary of the invention
In view of this; Fundamental purpose of the present invention is to provide a kind of recognition methods and system based on the part input; Not only save time but also be convenient to identification; Thereby the problem that the system resource of having avoided the prior art appearance to need the spent overlong time of system identification Chinese character to cause is taken in a large number, and the low problem of the system recognition rate that handwriting input caused.
For achieving the above object, technical scheme of the present invention is achieved in that
A kind of recognition methods based on part input, this method comprises: whole hand-written zoning is divided into a plurality of identified regions, is based on stroke that identified region partly imports and constructs and identify the target candidate speech.
Wherein, this method also comprises: dispose the corresponding relation between the stroke structure of said identified region and said part input;
Saidly be based on stroke structure that identified region partly imports and identify the target candidate speech and specifically comprise: identify the target candidate speech according to configured corresponding relation.
Wherein, under the handwriting input state, the stroke structure of said part input does not receive under the situation of screen limitations affect, saidly is based on stroke structure that identified region partly imports and identifies the target candidate speech and also comprise:
Judge the identified region that is complementary in the position in hand-written district with the stroke of partly importing;
According to said configured corresponding relation, judge the stroke structure of the part input corresponding with the identified region that is complementary, obtain candidate word, from candidate word, identify the target candidate speech.
Wherein, under the handwriting input state, the stroke structure of said part input receives under the situation of screen limitations affect, saidly is based on stroke structure that identified region partly imports and identifies the target candidate speech and also comprise:
Gather preparatory input operation; The identified region that the stroke that said preparatory input operation is used to select and part is imported is complementary in the position in hand-written district;
According to the preparatory input operation of gathering, judge the identified region that is complementary in the position in hand-written district with the stroke of partly importing;
Be captured in the stroke of the part input of whole hand-written district input,, judge the stroke structure of the part input corresponding, obtain candidate word, from candidate word, identify the target candidate speech with the identified region that is complementary according to said configured corresponding relation.
A kind of recognition system based on the part input, this system comprises: division unit and recognition unit; Wherein,
Division unit is used for whole hand-written zoning is divided into a plurality of identified regions;
Recognition unit is used to be based on the stroke structure that identified region partly imports and identifies the target candidate speech.
Wherein, said recognition unit, the corresponding relation between the stroke structure that is further used for importing according to the said identified region and the said part of configuration identifies the target candidate speech.
Wherein, said recognition unit is further used under the handwriting input state, and the stroke structure of said part input does not receive under the situation of screen limitations affect, judges the identified region that is complementary in the position in hand-written district with the stroke of partly importing; According to said configured corresponding relation, judge the stroke structure of the part input corresponding with the identified region that is complementary, obtain candidate word, from candidate word, identify the target candidate speech.
Wherein, said recognition unit is further used under the handwriting input state, and the stroke structure of said part input receives to gather preparatory input operation under the situation of screen limitations affect; The identified region that the stroke that wherein said preparatory input operation is used to select and part is imported is complementary in the position in hand-written district; According to the preparatory input operation of gathering, judge the identified region that is complementary in the position in hand-written district with the stroke of partly importing; Be captured in the stroke of the part input of whole hand-written district input,, judge the stroke structure of the part input corresponding, obtain candidate word, from candidate word, identify the target candidate speech with the identified region that is complementary according to said configured corresponding relation.
The present invention is divided into a plurality of identified regions with whole hand-written zoning, is based on the stroke structure that identified region partly imports and identifies the target candidate speech.Adopt the present invention; Input just can reach the effect of the whole word input of prior art based on part; Not only save time but also be convenient to identification; Thereby the problem that the system resource of having avoided the prior art appearance to need the spent overlong time of system identification Chinese character to cause is taken in a large number, and the low problem of the system recognition rate that handwriting input caused.
Embodiment
Basic thought of the present invention is: whole hand-written zoning is divided into a plurality of identified regions, is based on the stroke structure that identified region partly imports and identifies the target candidate speech.
For making the object of the invention, technical scheme and advantage clearer, below lift embodiment and with reference to accompanying drawing, to further explain of the present invention.
A kind of recognition methods based on the part input, as shown in Figure 2, the principle process of this method may further comprise the steps:
Step 101, whole hand-written zoning is divided into a plurality of identified regions.
Step 102, be based on the stroke structure that identified region partly imports and identify the target candidate speech.
Here; As far as the area dividing in hand-written district, it all is feasible being divided into several zones, and its principle is just passable as long as can distinguish the part of input; Because all examples can't be enumerated; Therefore, following example through 9 area dividing and 4 area dividing describes, and the recognition effect that reaches when wherein adopting 9 area dividing is more excellent.
The Chinese character of China has the monomer structure of being divided into (like " greatly " word); Delta structure, left and right sides structure (like " Liu "), up-down structure (like " merchant " word); Left, center, right structure (like " street " word); Structure (like " meaning " word) under in going up, full investing mechanism (like " state " word), semi-surrounding structure (like " celebrating " word).
In view of above-mentioned these several class formations, for instance, can hand-written zoning be divided into several zones during 9 area dividing: 1,2,3,4,5,6,7,8,9, be illustrated in fig. 3 shown below.
Here, this method also comprises: the corresponding relation between the stroke structure of configuration identified region and part input.Being based on stroke structure that identified region partly imports identifies the target candidate speech and specifically comprises: identify the target candidate speech according to configured corresponding relation.For instance, for the information of part input, can be according to the input of the stroke structure position under it at specific identified region, through the positional information of part input and identified region confirm the user the Chinese character that will import, reach the purpose of quick identification.
Below be configured corresponding relation, as shown in Figure 3, comprising:
Delta structure: top is imported in No. 2 zones, and bottom left section is imported in 4 and No. 7 zones, and lower right-most portion is imported in 6 and No. 9 zones.
Left and right sides structure: left half is imported 1,4, No. 7 zones, and right half is imported 3,6, No. 9 zones.
Up-down structure: top is imported 1,2, and No. 3 zones, lower part are imported 7,8, No. 9 zones.
The left, center, right structure: left half is imported 1,4, No. 7 zones, and middle part is imported 2,5, No. 8 zones, right half is imported 3,6, No. 9 zones.
Structure under in going up: top is imported in 1,2, No. 3 zones, and middle part input is in 4,5, No. 6 zones, and the lower part is imported in 7,8, No. 9 zones.
Full investing mechanism: peripheral part is imported 1,2, and 3,6,9,8,7, No. 4 zones, center section is imported in No. 5 zones.
The semi-surrounding structure: peripheral part is imported 1,2, and 3,6,9,8,7, No. 4 zones, center section is imported in No. 5 zones.
In view of above-mentioned these several class formations, for instance, can hand-written zoning be divided into several zones during 4 area dividing: 1,2,3,4.As shown in Figure 7.
Be configured corresponding relation below, comprise:
Delta structure: top is imported in 1 and No. 2 zone, and bottom left section is imported in No. 3 zones, and lower right-most portion is imported in No. 4 zones.
Left and right sides structure: left half is imported in 1, No. 3 zone, and right half is imported in 2, No. 4 zones.
Up-down structure: top is imported in 1, No. 2 zone, and the lower part is imported in 3, No. 4 zones.
The left, center, right structure: left half is imported in 1, No. 3 zone, and the input of middle part is at No. 1 and No. 2, the neighboring region of No. 3 and No. 4, and right half is imported in 2, No. 4 zones.
Structure down in going up: top import in 1, No. 2 zone, and middle part input is at No. 1 and No. 3, the neighboring region of No. 2 and No. 4, and the lower part is imported in 3, No. 4 zones.
Full investing mechanism: peripheral part import 1,2,3, No. 4 zones, and center section is imported at No. 1 and No. 3, the neighboring region of No. 2 and No. 4.
The semi-surrounding structure: peripheral part is imported 1,2,3, No. 4 zones, and center section input is at No. 1 and No. 3, the neighboring region of No. 2 and No. 4.
The present invention designs two kinds of different portions input identifying schemes also to different screen types.
One, under the handwriting input state; The stroke structure of part input does not receive under the situation of screen limitations affect, is based on stroke structure that identified region partly imports and identifies the target candidate speech and also comprise: judge the identified region that the stroke with the part input is complementary in the position in hand-written district; According to said configured corresponding relation, judge the stroke structure of the part input corresponding with the identified region that is complementary, obtain candidate word, from candidate word, identify the target candidate speech.
For instance, utilize the input information in the subregion to carry out the identification that part is imported.
When the user imports a part of hand-written information; At first judge the identified region in the hand-written district at input stroke information place; Positional information according to identified region obtains the information that this part is imported corresponding place stroke structure, comes further to judge that the user wants the Chinese character of importing.Want "
" word such as the user; He can be 1; 2; Input " no " in No. 3 the zone; " no " that can obtain user's input like this is the top input of structure down during up-down structure is perhaps gone up; Can obtain "
" like this; Candidates such as " bad " supplies the user to select; Even the user does not know how "
" writes, as long as but remembering that there are " no ", this Chinese character of input that also can be correct in the top.As shown in Figure 4, the candidate word after obtaining to import is selected from candidate word, to identify the target candidate speech of input.
Two, under the handwriting input state, the stroke of part input structure receives under the situation of screen limitations affect, is based on stroke structure that identified region partly imports and identifies the target candidate speech and also comprise: gather preparatory input operation; The identified region that the stroke that input operation is used to select and part is imported in advance is complementary in the position in hand-written district; According to the preparatory input operation of gathering, judge the identified region that is complementary in the position in hand-written district with the stroke of partly importing; Be captured in the stroke of the part input of whole hand-written district input,, judge the stroke structure of the part input corresponding, obtain candidate word, from candidate word, identify the target candidate speech with the identified region that is complementary according to said configured corresponding relation.
This solution is actually for fear of the solution of the limiting factor of mobile phone screen, because the screen of some mobile phone is less, if be 9 zones with screen divider, each regional area can be very little, is not easy like this to write according to the zone.
For instance, to this type situation, entire identification process mainly comprises following content:
1) process of selection input area: the user only need slip over stroke and want the importation The corresponding area to get final product.For example we will import " prosperous ", and we only need the top of this word of input get final product, and 1,2,3 zones of correspondence slipped over and got final product on top this moment, were illustrated in fig. 5 shown below.
2) process of stroke input behind the selected input of user position, just can be imported the stroke that will import in whole hand-written district.Behind the position of selected " prosperous ", we can import the top " gold " of this word on whole screen.As shown in Figure 6.
3) utilize the stroke tectonic information in the process of positional information and the input of above-mentioned stroke of the identified region in the process of above-mentioned selection input area to confirm that the user wants the Chinese character of importing, the candidate word after the input is provided.From candidate word, select, to identify the target candidate speech of input.
In sum, contrast the present invention and prior art can be known: prior art is based on the recognition technology of whole word, owing to must in input area, intactly import each stroke of whole word; Just can identify this word of input; Therefore, on the one hand, when handwriting input; Some word too complex, thus can cause the too loaded down with trivial details waste plenty of time of input that system resource is taken in a large number; The screen of adding is too small, and the input identification that makes handwriting input must import whole Chinese character implements very difficulty; On the other hand, when the user does not remember the stroke structure that the residence has to some rarely used word, and only remember the part of stroke structure, input identification cannot correctly be accomplished in the time of also can causing handwriting input, causes the system identification inefficiency.And the present invention utilizes the part input to discern Chinese character through hand-written district being carried out area dividing, just can reach the effect of the whole word recognition of prior art; That is: employing the present invention; The stroke structure that only needs the input that part of input that the user knew or certain input area to have characteristic just can obtain the target candidate speech of user expectation, and the stroke that need not user's memory and import whole word is constructed, thereby has not only promoted the user experience of handwriting input; Main has carried out improving greatly to the internal system performance; Because identification is convenient in the present invention, recognition efficiency is high, judges by accident low; Save time, thereby system need not to spend the plenty of time and resource is done input identification processing and judgement.
A kind of recognition system based on the part input, this system comprises: division unit and recognition unit.Wherein, division unit is used for whole hand-written zoning is divided into a plurality of identified regions.Recognition unit is used to be based on the stroke structure that identified region partly imports and identifies the target candidate speech.
Here, recognition unit is further used for identifying the target candidate speech according to the corresponding relation between the stroke structure of the identified region of configuration and part input.
The recognition unit of below importing to different two kinds of different portions of screen type design.
One: under the handwriting input state, partly the stroke of input structure does not receive under the situation of screen limitations affect, and recognition unit is further used for judging the identified region that is complementary in the position in hand-written district with the stroke of partly importing; According to said configured corresponding relation, judge the stroke structure of the part input corresponding with the identified region that is complementary, obtain candidate word, from candidate word, identify the target candidate speech.
Two: under the handwriting input state, the stroke structure of said part input receives under the situation of screen limitations affect, and recognition unit is further used for gathering preparatory input operation; The identified region that the stroke that wherein input operation is used to select and part is imported in advance is complementary in the position in hand-written district; According to the preparatory input operation of gathering, judge the identified region that is complementary in the position in hand-written district with the stroke of partly importing; Be captured in the stroke of the part input of whole hand-written district input,, judge the stroke structure of the part input corresponding, obtain candidate word, from candidate word, identify the target candidate speech with the identified region that is complementary according to said configured corresponding relation.
Here, the abbreviation and the Key Term that relate in the above literal are explained as follows:
Hand-written district refers to: the zone that is used to utilize touch manner input characters stroke on the touch-screen.
Candidate regions refers to: the list area that is used for the identified Chinese character of stroke that explicit user slides on the touch-screen on touch-screen.
The input stroke refers to: the sliding trace that the user slides on touch-screen and produces through finger or felt pen.
The above is merely preferred embodiment of the present invention, is not to be used to limit protection scope of the present invention.