WO2015070825A2 - 连续滑动输入单词的方法和系统 - Google Patents

连续滑动输入单词的方法和系统 Download PDF

Info

Publication number
WO2015070825A2
WO2015070825A2 PCT/CN2015/070783 CN2015070783W WO2015070825A2 WO 2015070825 A2 WO2015070825 A2 WO 2015070825A2 CN 2015070783 W CN2015070783 W CN 2015070783W WO 2015070825 A2 WO2015070825 A2 WO 2015070825A2
Authority
WO
WIPO (PCT)
Prior art keywords
word
sensing device
information
input
trajectory
Prior art date
Application number
PCT/CN2015/070783
Other languages
English (en)
French (fr)
Other versions
WO2015070825A3 (zh
Inventor
王佳梁
张瞰
邹林
Original Assignee
上海触乐信息科技有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 上海触乐信息科技有限公司 filed Critical 上海触乐信息科技有限公司
Publication of WO2015070825A2 publication Critical patent/WO2015070825A2/zh
Publication of WO2015070825A3 publication Critical patent/WO2015070825A3/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
    • G06F3/0487—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser
    • G06F3/0488—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures
    • G06F3/04886—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures by partitioning the display area of the touch-screen or the surface of the digitising tablet into independently controllable areas, e.g. virtual keyboards or menus
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/335—Filtering based on additional data, e.g. user or group profiles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
    • G06F3/0487—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser
    • G06F3/0488—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures
    • G06F3/04883—Interaction techniques based on graphical user interfaces [GUI] using specific features provided by the input device, e.g. functions controlled by the rotation of a mouse with dual sensing arrangements, or of the nature of the input device, e.g. tap gestures based on pressure sensed by a digitiser using a touch-screen or digitiser, e.g. input of commands through traced gestures for inputting data by handwriting, e.g. gesture or text

Definitions

  • the invention relates to the field of electronic device input control, in particular to the technical field of electronic device touch screen information input, in particular to a method and system for sliding input words based on a software keyboard on an electronic device.
  • the main input is done via a software keyboard.
  • the user completes the input by pressing a button on the software keyboard to select the input content.
  • the traditional software keyboard brings great inconvenience to the user. This is mainly because the touch screen area is small and there is more content to be selected, so the corresponding screen selection area is also small. This makes it impossible for the user to directly input through the selection area on the screen by the finger, and often requires other tools having a smaller contact resolution, such as a stylus.
  • Another way to enter text on the touch screen is to slide in.
  • User in a software Slide the keyboard over the position of each letter in the word and enter the desired word.
  • the user's finger or stylus does not need to leave the touch screen, so that the speed of the slide can be faster than the click speed, thus making the input more convenient.
  • the existing sliding input technology usually finds the corresponding letter according to the coordinates of the key points in the sliding process of the user, and when the contact area of the button and the sensing device is large, the screening efficiency can be high, and the pair is reduced. Subsequent calculation processing capability requirements.
  • the size of the software keyboard disposed on the terminal device is also reduced, and accordingly, the contact area between each button and the sensing device is also smaller, which causes the user to accidentally press or system error. The probability of identification and misjudgment is greatly increased. In this case, the selection of the coordinates of the key points, due to the user input habits and other factors in the sliding input process, is likely to cause a higher false positive rate, which greatly affects the accuracy of the prediction is not high enough .
  • the user may input the word “kate” by using the track shown in FIG. 1 , it is also possible to sequentially follow the “k” “a” according to the track shown in FIG. 2 .
  • "t" "e”, and the possibility of mis-pressing or mis-pressing during the input process For example, as shown in Figure 3, the user mistakenly enters the letter “t” when passing the letter "t”. In the "y" field of the character, or other mistypes.
  • the commonly used sliding input method generally obtains a corresponding candidate word by detecting an inflection point in the sliding trajectory as a key point and querying the characteristic vocabulary according to the key point, but the inventor finds that the input word is continuously slid.
  • the process of the above due to the above various possible accidents or different user input habits, in the process of detecting the inflection point in the sliding track, it is prone to omission, mis-increment, or misjudgment of the inflection point, which will be searched.
  • the accuracy of the results has an impact, and it also increases the amount of data calculation and increases the hardware burden.
  • the amount of data to be calculated is large, and the calculation complexity is high.
  • the calculation process can be realized by a CPU with higher computing power, a large amount of calculation usually consumes more power.
  • a large amount of calculation usually consumes more power.
  • it often imposes a large limitation on the endurance of the electronic device; It is often necessary to slide out the complete trajectory of a word to complete the calculation. For some longer words, it is difficult for the user to draw a complete correct trajectory.
  • a method for continuously sliding an input word includes: detecting a motion trajectory of the sensing device, recording input information, and acquiring feature information from the input information, the feature information including at least an angle At least one of information and a length of motion trajectory; searching in the feature vocabulary according to the feature information will meet at least one of: the angle information is adapted to the angle information in the feature information or the length of the motion track is a word corresponding to the length of the motion track in the feature information is used as a matched word, and a preliminary screening word set is obtained; and according to the actual motion track of each word in the preliminary screening word set, the first standard track corresponding to the word is calculated.
  • a distance measure filtering the words according to the distance measure to obtain a rough filter word set; calculating a similarity between the actual motion track of each word in the coarse filter word set and the corresponding second standard track, according to the similarity degree
  • the words are sorted and output to the user according to the sort.
  • the present invention also provides a system for continuously sliding input words, comprising: an input module adapted to receive a sliding input of the sensing device; an output module adapted to feed the output result to the user; a recording module, configured to detect the input of the sensing device and record the input information; the feature information acquiring module is adapted to obtain, from the recorded input information, feature information including at least one of the angle information and the actual motion track length; the preliminary screening module And being adapted to retrieve a matching word from the feature vocabulary according to the feature information to obtain a preliminary screening word set; the coarse filtering module is adapted to initially filter each word vocabulary between the word vocabulary and the standard trajectory The distance measure is filtered according to the distance measure to obtain a rough filtered word set; the fine filtering module is adapted to calculate a similarity between the standard track and the actual motion track for each word in the coarse filtered word set. Degree, and sort the words obtained by fine filtering, the words Output to the output module in the stated order.
  • a system for continuously sliding input words comprising: an interaction device adapted to receive a user's sliding input, to detect and record user input, and to adapt the input result Feedback to the user; the processor is adapted to perform the following operations: performing the retrieval in the feature vocabulary according to the feature information, and meeting at least one of the following: the angle information is adapted to the angle information in the feature information or the motion trajectory thereof a set of words whose length is adapted to the length of the motion track in the feature information is used as a preliminary screening word set; and according to the actual motion track of each word in the preliminary screening word set, the first standard track corresponding to the word is calculated The distance between the metrics is filtered according to the distance metric to obtain a rough filtered word set; the similarity between the actual motion trajectory of each word in the coarse filtered word set and the corresponding second standard trajectory is calculated, according to the similarity degree The words are sorted and output to the user in the sorted order.
  • the present invention provides a method and system for continuously sliding input words based on a software keyboard on an electronic device, wherein the interactive device records and transmits the input information input by the user to the processor, and the processor Entering the characteristic information such as the angle information and the actual track length in the information, quickly filtering out the preliminary screening word set of hundreds or thousands of similar words from the feature vocabulary, and then further using the position information to further perform the preliminary screening of the word set. Filtering to obtain a rough filter word set, on the basis of which the similarity between the actual motion trajectory of each word and its standard trajectory is calculated, and the weighted calculation is performed according to the priority of the word, the environmental word information, the historical input information, and whether it is a prediction result or the like.
  • the candidate words are sorted, and then output to the interactive device, and fed back to the user, so that various features in the sliding process can be utilized, and the user's misoperation and system misidentification can be effectively alleviated under the premise of ensuring the speed.
  • the adverse effects brought about by the The input speed and accuracy significantly reduce the user's operation complexity, the implementation process is simple, the use is convenient and fast, the work performance is stable and reliable, and the scope of application is wide, which greatly promotes the large-scale application of the electronic device touch screen information input method to people. Work and life bring great convenience.
  • FIG. 1 to FIG. 3 are schematic diagrams showing an implementation of inputting a word “kate” by using a software keyboard on an electronic device
  • FIG. 4 is a schematic structural view of an embodiment of a continuous sliding input word system of the present invention.
  • FIG. 5 is a schematic flow chart of an embodiment of a method for continuously sliding input words according to the present invention.
  • step S1 shown in FIG. 5;
  • FIG. 7 is a schematic diagram of a logical storage structure of a feature vocabulary in an embodiment of a method for continuously sliding input words according to the present invention.
  • FIG. 8 is a schematic diagram of calculating an abscissa distance metric in an actual motion trajectory in an embodiment of a method for continuously sliding input words according to the present invention
  • FIG. 9 is a schematic flow chart of calculating a standard trajectory of each word and its actual trajectory similarity in an embodiment of the method for continuously sliding input words according to the present invention.
  • FIG. 10 is a schematic diagram of a framework of an embodiment of a continuous sliding input word system of the present invention.
  • FIG. 11 is a schematic diagram of another embodiment of a continuous sliding input word system of the present invention.
  • keyboard is broadly defined and includes any input component having a defined area, including but not limited to a hardware mechanical keyboard, a hardware inductive keyboard, a software keyboard on a touch screen, and the like.
  • the embodiment of the present invention is exemplified by a software keyboard on a touch screen by default, but those skilled in the art can understand that the method and system mentioned in the present invention can also be applied to a hardware keyboard.
  • an embodiment of the continuous sliding input word system may include at least an interaction device 110 and a processor 120, wherein the interaction device 110 may include a keyboard 101 and a text display area on the mobile communication terminal device. 102, such as a touch display with a software keyboard; the processor 120 can be a CPU or an application specific integrated circuit chip.
  • the interaction device 110 may include a keyboard 101 and a text display area on the mobile communication terminal device. 102, such as a touch display with a software keyboard; the processor 120 can be a CPU or an application specific integrated circuit chip.
  • the keyboard 101 is a software keyboard, and the keyboard layout can adopt a QWERTY one-button keyboard layout, which includes 26 English letters, numbers, symbols and space keys. Each button represents a letter, and the number or symbol can be combined with the letter at the same button. Up; AZERTY's one-button keyboard layout is also available. In other embodiments, the keyboard 101 may also adopt other keyboard layouts, or may be applicable to languages other than English, and the specific contents and positions of letters, numbers, and symbols may be changed according to the requirements of a specific language or keyboard layout.
  • the keyboard 101 and the text display area 102 may be integrated on the same interactive device, or may be separately disposed on separate interactive devices.
  • the keyboard 101 is unfolded or stowed to obtain a text display area 102 of a variable size; the size of the keyboard 101 can also be adjusted.
  • the user sequentially uses the sensing device 130 such as a finger or a touch pen to cross the approximate area of the letters "k", "a", "t", and "e” in the keyboard 101 area, and interacts.
  • the device 110 detects the action of the sensing device 130, records the input information of the sensing device 130, such as the position where the sensing device 130 stays, the angle of motion, the length of the motion track, and the like, and obtains the motion track of the sensing device 130 through the input information;
  • the device 110 transmits the recorded input information and the motion trajectory to the processor 120, and the processor 120 further processes the trajectory of the sensing device 130 by performing operations such as searching, filtering, and matching in the feature vocabulary.
  • the matched set of words is then further sorted according to the set order, and the sorted result is presented in the text display area 102 for feedback to the user.
  • an embodiment of the system for continuously sliding input words may further comprise a storage device 140 adapted to store data generated or possibly generated in a method for implementing continuous sliding input of words, such as a feature vocabulary , input information, motion trajectories, etc.
  • the storage device 140 can be any memory medium such as random access memory (RAM), read only memory (ROM), flash memory, hard disk, and the like.
  • step S1 the motion of the sensing device is detected, and the input information is recorded.
  • step S110 is performed to detect whether the sensing device is in contact with the interaction device.
  • step S120 is performed to record the position of the sensing device, and by step S130, it is detected whether there is movement of the sensing device.
  • the position of the sensing device is continuously recorded; when it is detected that the sensing device stops moving, it is detected whether it stops contacting with the interactive device, its position is recorded, and the time at which it stops at the stop position can also be recorded.
  • step S140 Determine whether the corresponding word is output according to the input information by comparing the time when the sensing device stays at the stop position with the predetermined time threshold. For example, when the dwell time exceeds the predetermined time threshold, step S150 is performed to output the input track.
  • step S150 is performed to output the input track.
  • the sensing device may further include further recording angle information corresponding to the position, for example, calculating a slope of each position point to obtain angle information corresponding to the position; and further including further recording the actual motion track length of the sensing device. For example, according to the motion trajectory of the sensing device, the distance between each two consecutive positions is added, and the length of the trajectory is approximated, and the length of each user sliding trajectory can be easily extracted by using a similar integration method.
  • the size may be normalized to eliminate the size difference caused by the difference in the size of the touch screen of the mobile phone; and may include smoothing the trajectory by using a trajectory smoothing method, for example, using a mean filter to eliminate slippage caused by hand shake or the like.
  • Noise in the trajectory data may also include processing such as point distance normalization.
  • Step S2 acquiring feature information such as angle information and motion track length from the input information; and performing a search in the feature vocabulary according to the feature information to obtain a preliminary screening word set.
  • the input information recorded in step S1 does not include angle information, motion track length, and the like.
  • Step S2 performs calculation based on the recorded position information, and acquires feature information such as corresponding angle information and motion track length.
  • an angle of the starting position of the sensing device can be calculated based on the input information, and a word having a matching starting point angle can be found in the thesaurus. For example, when the user enters the word "what" through the keyboard of the qwerty layout, when the finger slides from the initial "w" to the second letter “h", the angle of the starting position is the same as the user inputting "snake”. The angle of the starting position is the same, so when the retrieval is performed in the feature vocabulary based on the angle information of the starting position, the word "snake” is also added to the initial screening word set.
  • an angle of the sensing device termination position can be calculated based on the input information, and a word having a matching termination position angle is searched in the thesaurus. For example, when the user enters the word "insist” through the keyboard of the azerty layout, when the finger slides from the letter “s” to the last letter “t", the difference between the end position angle and the end position angle when the user inputs "candy" is Within a predetermined threshold, therefore, when searching in the feature vocabulary based on the angle information of the end position, the word “candy” is also added to the preliminary screening word set.
  • the combination of the angle information of the angle information of the starting position and the angle information of the ending position, or the angle information of other specified positions, or the combination of the angle information of the starting position or the ending position may also be used.
  • a search is performed in the feature vocabulary to obtain a preliminary filtered word set.
  • the predetermined angle threshold range may be set, and when the corresponding angle of the word in the feature vocabulary does not exceed the predetermined angle threshold range, it may be considered to match the angle information; for example, the starting point of the word may be calculated The difference between the angle and the actual input starting point angle, when the difference is within a predetermined angle threshold range, then the word is a matching starting angle word.
  • the length of the actual motion trajectory of the sensing device may also be obtained from the input information, and the word with the matching motion trajectory length is searched in the lexicon.
  • the predetermined length threshold range may be set, and when the corresponding track length of the word in the feature vocabulary does not exceed the predetermined length threshold range, it may be considered to match the actual motion track length; for example, the word may be calculated The difference between the length of the track and the length of the actual motion track, when the difference is within a predetermined length threshold, then the word is a matching track length word.
  • the track length can be combined with the angle information for preliminary screening in the feature lexicon.
  • the first preliminary screening may be performed according to the length of the track, and then the first preliminary screening may be screened again according to the angle information, according to the sieve.
  • the result of the selection is to obtain a preliminary screening word set; or to change the order of the two screenings before and after, that is, the first preliminary screening is performed according to the angle information, and then the screening is performed according to the length of the track; or when the feature lexicon is constructed, the words are according to the length of the track.
  • the angle information establishes an index sequence, and then searches according to the angle information or the track length to obtain a preliminary screening word set.
  • the preliminary screening in the feature vocabulary reduces the requirements on the size of the device and reduces the requirement for the consistency of the actual motion input trajectory, and enhances the user's accidental sliding and the system.
  • the inaccuracy of misidentification greatly improves the accuracy of recognition.
  • the input information recorded in step S1 includes feature information such as angle information and motion track length, and step S2 extracts therefrom to obtain the required related feature information.
  • Step S2 may extract all the feature information and process the feature information with respect to the motion trajectory of the sensing device; or process only the feature information such as a predetermined number of angle information and track length, for example, when After the feature information is extracted according to all the input information, a predetermined number of feature information is used for subsequent retrieval, screening, and the like, wherein the predetermined number may be 75%-95% of the total data.
  • a large amount of data training shows that using 85% of the input information corresponding to the input information can achieve a good balance between accuracy and computational complexity, so that the amount of data can be reduced without affecting the accuracy. Reduce the requirements on hardware devices, improve processing efficiency, and provide users with a faster and more accurate input experience.
  • the feature vocabulary is a vocabulary of words established by indexing each feature information.
  • the standard feature information corresponding to the words in the dictionary can be extracted and stored in the feature lexicon module. For example, corresponding to the angle in the feature information, the ideal angle information corresponding to the word in the dictionary may be extracted; and corresponding to the length of the motion track in the feature information, the standard track length corresponding to the word in the dictionary may be extracted.
  • the feature vocabulary constructed may include a word feature table and a word standard trajectory.
  • the index table, the field in the word feature table may include a word identifier, a word content, and may further include at least one of a standard track length and a standard angle index, and the word standard track index table refers to a word in the word feature table.
  • the standard track of the word may be a large number of similar tracks of corresponding words extracted from the sliding track of the user according to a large amount of training data, or the standard track of the word may also be a straight line of the continuous letter center of the word in the screen software keyboard. line.
  • the logical storage structure of the final feature vocabulary is shown in Figure 7.
  • the vocabulary here can use existing small database storage, or design private storage to optimize data access speed and space.
  • step S3 the input information is processed, and the preliminary screening word set obtained in step S2 is filtered according to the processing result.
  • the processing of the preliminary screening word set may further include coarse filtering and fine filtering.
  • the inventors since the initial screening of word sets is obtained by using angle information, the number of words included in the initial screening word group is usually relatively large, and the inventors have found that if the initial screening of word sets is performed one by one to calculate and match similarities one by one, not only consumes more resources, but also consumes more resources. Moreover, the calculation time is relatively long. Therefore, the inventor first performs coarse filtering on the input information, and then performs fine filtering again according to the result of the coarse filtering, thereby saving resources and improving calculation efficiency.
  • an envelope algorithm is used to calculate a distance measure between the actual trajectory and the standard trajectory for each word in the word set obtained by the preliminary screening, and filter according to the calculated distance metric. Get a rough set of filtered words.
  • the following steps can be used to calculate the distance metric.
  • the envelope of the word is calculated according to the actual motion trajectory of each word. Specifically, the upper envelope and the lower envelope of each point in the actual motion trajectory are respectively calculated.
  • the abscissa or the ordinate of each point in the actual motion trajectory of the sensing device is extracted.
  • the corresponding upper envelope value and lower envelope value are calculated for the abscissa or ordinate of each point on the actual motion trajectory.
  • the abscissa or the ordinate of the Nth point and the set number of points are compared, and the maximum value is the upper envelope value of the Nth point, and the minimum value is the sub-package of the Nth point. Network value.
  • Nth point For example, comparing the Nth point with its first i points, ie, the Nith, ..., the N-1th point, and the subsequent i points, ie, the N+1th, ..., N+i points coordinate;
  • XupperboundN Xmax ⁇ N-i,...,N-1,N,N+1,...,N+i ⁇
  • XlowerboundN Xmin ⁇ N-i,...,N-1,N,N+1,...,N+i ⁇ ;
  • YupperboundN Ymax ⁇ N-i,...,N-1,N,N+1,...,N+i ⁇
  • YlowerboundN Ymin ⁇ N-i,...,N-1,N,N+1,...,N+i ⁇ .
  • N ⁇ i and N>j j is an arbitrary natural number between 1 and i, and the abscissa or ordinate of the Nith, ..., Nj-1th points can be estimated according to the motion trajectory; Or, when the track contains M points, and N+i>M and N+j ⁇ M, j is an arbitrary natural number between 1 and i, and according to the motion trajectory, the N+j+1, ..., The abscissa or ordinate of the N+ith point is estimated.
  • the upper envelope 211 and the lower envelope 212 respectively relative to the entire motion trajectory abscissa 210 are obtained, and the upper envelope and the lower envelope relative to the ordinate of the entire motion trajectory.
  • the first standard trajectory of each word is calculated.
  • the first standard trajectory is calculated according to the sampling frequency of the actual motion trajectory, so that the number of sampling points of the first standard trajectory to be obtained is the same as the number of sampling points of the actual motion trajectory, and Based on this, a first standard trajectory is obtained.
  • the first standard trajectory of each word may also be calculated first, or the first standard trajectory may be calculated simultaneously on multiple processors and the distance metric may be calculated if the actual computing power permits.
  • the distance metric is calculated based on the first standard trajectory of each word and its envelope. Specifically, referring to FIG. 8, when calculating the distance metric, respectively, the abscissa 230 of the first standard trajectory of the word and the corresponding actual motion trajectory abscissa, ie, the upper envelope 211 and the lower envelope 212, are compared. Calculating the sum of the abscissa of the first standard trajectory beyond the envelope; And the ordinate of the first standard trajectory of the word and its corresponding actual trajectory ordinate, ie, the upper envelope 221 and the lower envelope 222, are compared, and the ordinate of the first standard trajectory exceeds the sum of the envelopes. Then, calculate the sum of the two and use it as the distance measure for the word.
  • the number of settings may be set according to the actual computing power of the mobile terminal, for example, may be several tens to several thousand.
  • fine filtering is performed, that is, the similarity between the standard trajectory and the actual trajectory is calculated separately. Specifically, referring to FIG. 9, calculating the similarity between the standard trajectory of each word and its actual trajectory, the following steps can be taken:
  • a second standard trajectory corresponding to each word is calculated, and a corresponding actual motion trajectory is calculated.
  • the second standard trajectory corresponding to each word may be calculated according to the input information or the input information.
  • the trajectory point sequence corresponding to the second standard trajectory may be acquired by using an upsampling manner.
  • step S420 the similarity is calculated according to the second standard trajectory of each word and its actual motion trajectory.
  • it may include:
  • D[i,j] cost+min(D[i-1,j], D[i,j-1], D[i-1,j-1]);
  • the total number of points in the trajectory, min(D[i-1,j], D[i,j-1], D[i-1, J-1]) is the minimum value among D[i-1,j], D[i,j-1], D[i-1,j-1];
  • step (ii) repeating the above step (i) until traversing all the points in the second standard trajectory, and obtaining a similar distance D[n, m] between the accumulated calculated trajectory and the actual trajectory;
  • the similar distance is the similarity between the standard trajectory feature of the word and the actual trajectory.
  • step S4 is performed to further sort and output the set of candidate words obtained after filtering. Specifically, after obtaining the similarity of each candidate word, the priority of the candidate word, the environmental word information, the historical input information, and whether it is a prediction result may be further combined to perform a weighting calculation, thereby obtaining a score of the candidate word. Finally, the finely selected candidate word sets are sorted according to the scores, and the candidate words are sequentially output to the interactive device in order to implement feedback on the user input.
  • the interaction device 110 may at least include an input device adapted to the user for sliding input, such as a touch sensitive input device, a contactless inductive input device, and an output device that feeds the input result to the user. , for example, video output devices, audio output devices, and the like.
  • the input device can be integrated with the output device, for example, the interaction device 110 can be a touch display integrated with a software keyboard; in addition, the input device can also be separated from the output device, for example, the user inputs on the touch panel, The output is fed back to the user via the audio output device, or its output is displayed through a display separate from the touchpad.
  • the interaction device 110 can also detect and record the input of the user, for example, detecting whether the sensing device operated by the user has contact with the user; when there is contact, recording the location of the sensing device when contacting the device, and detecting the sensing device Whether there is movement; when detecting that there is movement of the sensing device, continuously recording the position of the sensing device; when detecting that the sensing device stops moving, detecting whether it stops contacting, recording its position and the time it stays at the position where the contact is stopped. When the dwell time exceeds the predetermined time threshold, the word corresponding to the input trajectory is output. In the above detection process, when it is detected that the sensing device stops contacting with the interaction device, the corresponding word set is obtained according to the input information, and the obtained word set is output to the user.
  • the interaction device 110 may also preprocess the input information.
  • the input information can be upsampled; for example, the input information can be smoothed. Due to the user's input, such as hand shake, etc., the defect in the sliding track data may be caused. Therefore, a series of preprocessing operations may be performed on the track, for example, the size may be normalized to eliminate the size of the track due to the difference in the size of the touch screen of the mobile phone. Differences may include smoothing the trajectory by using trajectory smoothing, for example, using a mean filter, eliminating noise caused by hand shake and the like, and may also include processing such as pitch normalization.
  • the interaction device 110 may further process the input information.
  • the corresponding angle information may be calculated according to the location of the sensing device; and the actual motion may be calculated according to the location of the sensing device. Track length.
  • the calculation may be performed according to the input trajectory that has not been preprocessed, or the smoothed input trajectory, or the input trajectory after the upsampling; or the angle information may be separately calculated according to different input information.
  • the actual motion track length for example, the angle information can be calculated according to the smoothed input track, and the motion track length can be calculated according to the input track after the upsampled.
  • the interaction device 110 may further process according to the input information, and extract a predetermined quantity for subsequent processing for the obtained feature information.
  • a predetermined quantity for subsequent processing for the obtained feature information.
  • the interaction device 110 uses 85% of the input information corresponding to the input information, so that a better balance between the accuracy and the calculation amount can be achieved, and the user can provide faster and more accurate input. Experience.
  • the interaction device 110 can record the input information and transmit the recorded input information to the processor 120, or the interaction device 110 records and pre-processes the input information and inputs the pre-processed input.
  • the information is transmitted to the processor 120, and the processor 120 extracts the pair from the received input information or the preprocessed input information.
  • Characteristic information The processor 120 may extract corresponding feature information according to all input information, or extract feature information according to a predetermined number of input information, for example, extracting 75%-95% of input information, especially 85% of input information, corresponding to Feature information.
  • the search is performed in the feature vocabulary to find a word that matches the feature information, and a preliminary filter word set is obtained; then, according to the input information, the preliminary filter word set is first coarsely filtered, and then Fine filtration is carried out.
  • a memory 140 may also be included, adapted to store a feature vocabulary, and a set of selected or filtered words. Memory 140 may also be adapted to store a computer program that operates the invention.
  • the memory 140 may include one or more magnetic storage media such as a hard disk; one or more optical storage media such as a compact disk (CD), a rewritable optical disk (CD-Rs), and a repeatable optical disk (CD ⁇ RWs). ), DVDs, rewritable DVD (DVD ⁇ Rs), re-writable DVD (DVD ⁇ RWs), Blu-ray DVD; one or more semiconductor storage media such as flash memory, SD card, memory stick, or other suitable Computer readable medium.
  • the interaction device 810 can include: an input module 801 adapted to receive a sliding input of the sensing device; and an output module 802 adapted to The output result is fed back to the user; the input information recording module 803 is adapted to detect the input of the sensing device and record the input information.
  • the processor 820 may include: a feature information acquiring module 821, configured to acquire feature information such as angle information and track length from the recorded input information; and a preliminary screening module 822 adapted to select, according to the feature information such as the angle information and the track length.
  • the matching vocabulary 830 retrieves the matched words to obtain a preliminary screening word set;
  • the coarse filtering module 823 is adapted to calculate the distance between the actual motion trajectory and the standard trajectory for each word in the preliminary screening word set by using an envelope algorithm.
  • the metric is filtered according to the calculated distance metric to obtain a rough filtered word set.
  • the fine filtering module 824 is adapted to calculate the similarity between the standard trajectory and the actual motion trajectory for each word in the coarse filtered word set.
  • the words obtained by the fine filtering are sorted and output to the output module 802.
  • the input information recording module 803 may be further adapted to pre-process the recorded input information, such as smoothing processing, such as upsampling, and output the pre-processed input information to the processor 820;
  • the input information recording module 803 records only the input information, and the feature information acquiring module 821 preprocesses the input information, and extracts feature information such as angle information and track length according to the preprocessed result, and The pre-processed results are transmitted to a preliminary screening module 822, a coarse filtering module 823, and a fine filtering module 824.
  • the interaction device 910 performs the extraction operation of the feature information.
  • the interaction device 910 may include the feature information acquiring module 904 .
  • the interaction device 910 transmits the feature information and the input information to the processor 920, and performs subsequent filtering and filtering through the preliminary screening module 921, the coarse filtering module 922, and the fine filtering module 923 in the processor 920 to obtain candidate words, and the interaction device 910.
  • the candidate words are sorted according to a predetermined rule and outputted in order, thereby being fed back to the user.
  • the process is as follows:
  • the interactive device records the position information and the angle information when the sliding starts;
  • the interactive device records the position information and the angle information when the finger or the sensing device stops sliding, and calculates when the finger or the sensing device stops moving. Length of the sliding track;
  • the interactive device transmits the recorded position information, the angle information, and the length of the sliding track to the processor, and the processor is long according to the position information, the angle information, and the sliding track. Degrees match the user's motion trajectory.
  • the matching process is as follows:
  • the user's finger or sensing device slides on the keyboard, and the interactive device continuously records the position information of the finger or the sensing device;
  • the interactive device records the position information when the finger or the sensing device stops sliding;
  • the interactive device transmits all the recorded position information to the processor, and the processor calculates angle information of the start position and the end position, and calculates a sliding track length to the time when the finger or the sensing device stops moving, and according to the position information
  • the angle information and the length of the sliding track match the motion track of the user.
  • the matching process is as follows:
  • the interactive device when the user's finger or the sensing device starts to slide from a point near the button q, the interactive device records the position information and the angle information when the sliding starts;
  • the user's finger or sensing device slides on the keyboard, and the interactive device continuously records the position information of the finger or the sensing device;
  • the interactive device When the finger or the sensing device is lifted or stayed for more than a predetermined time threshold near the button k, the interactive device records the position information and the angle information when the finger or the sensing device stops sliding, and calculates when the finger or the sensing device stops moving. Length of the sliding track;
  • the interaction device transmits the angle information of the start position, the length of the sliding track, and the predetermined number of pieces of position information to the processor, for example, the trajectory, and the processor moves the user according to the received angle information, the position information, and the length of the sliding track. Tracks are matched.
  • the method of continuously sliding input words of the present invention can also support languages of other languages and is not affected by the layout of the keyboard.
  • the languages supported by the method of sliding input text include, but are not limited to, Spanish (English, French, German, etc.), Latin (Spanish, Portuguese, etc.), and Asian languages (Arabic, Hindi, Chinese, Japanese, Korean). Wait).
  • the method of sliding input text supports the most popular QWERTY keyboard layout, the AZERTY keyboard layout commonly used in French, and other keyboard layouts.
  • the present invention filters in the feature vocabulary according to the angle information and the trajectory length input by the user, and performs coarse filtering by using the envelope and further calculating the similarity to complete the fine filtering, obtaining the candidate words, and effectively reducing the
  • the user's misoperation and system misidentification may bring about adverse effects, reduce the amount of data calculation, save energy consumption, and provide users with a more convenient and fast input experience.
  • the invention can effectively improve the input speed and accuracy of words, significantly reduce user operation complexity, realize simple process, convenient and fast use, low computational complexity, and wide application range, and realize sliding input based on screen software keyboard.
  • the system of text is a system of text.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • User Interface Of Digital Computer (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Input From Keyboards Or The Like (AREA)

Abstract

一种连续滑动输入单词的方法和系统，其中，所述方法包括：检测感应设备的运动轨迹，记录输入信息，获取至少包括角度信息和运动轨迹长度中至少一个的特征信息；根据所述特征信息在特征词库中进行检索，将符合角度信息相适应或运动轨迹长度相适应中至少一种的单词作为匹配单词，获得初步筛选单词集合；计算每个单词的实际运动轨迹与其对应的第一标准轨迹之间的距离量度，获得粗过滤单词集合；计算粗过滤单词集合中每一个单词的实际运动轨迹与其对应的第二标准轨迹的相似度，根据相似度对所述单词进行排序后输出。本发明有效地减轻了误操作的不良影响，提高了输入速度和准确度，明显降低用户操作复杂度，实现过程简单，使用方便快捷。

Description

连续滑动输入单词的方法和系统
本申请要求2013年11月15日提交中国专利局、申请号为201310574003.1、发明名称为“连续滑动输入单词的方法和系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明涉及电子设备输入控制领域，特别涉及电子设备触摸屏幕信息输入技术领域，尤其是电子设备上基于软件键盘实现滑动输入单词的方法及系统。
背景技术
现代生活中，随着集成电路技术的不断发展，已经出现了各种各样的触摸屏电子设备，在这些众多的配备触摸屏的电子设备(如掌上电脑、智能手机、嵌入式系统、袖珍型电脑等)上，由于受到体积，制造成本等因素的制约，无法配备完善的硬件键盘，因而在这类设备上进行输入通常是通过软件模拟完成的。
通常在配备有较小面积触摸屏的设备上，主要的输入是通过软件键盘来完成的。用户通过按压点击软件键盘上的按键，来选择输入内容，从而完成输入。在现有技术中，传统的软件键盘会给用户的使用带来很大的不便。这主要是由于触摸屏面积较小，供选择的内容较多，所以相应的屏幕选择区域也较小。这样就使得用户无法直接通过手指准确的点击屏幕上的选择区域来进行输入而往往需要借助于其他具有更小触点分辨率的工具，如手写笔。这就要求在大多数情况下，用户需要同时使用双手进行操作：一只手把持该电子设备，另一只手利用手写笔点击触摸屏进行输入。这样一来，不仅操作比较麻烦，而且，即使如此，在某些环境，如行驶的汽车中或其它较为颠簸的环境中，用户仍然可能由于各种原因产生例如错误点击按键等各种误操作，从而无法准确实现输入。
另一种在触摸屏上输入文本的方式是滑动输入。用户在一个软件 键盘上依次滑过单词中各个字母所在的位置，输入想要的单词。从理论上来说，用户的手指或手写笔不需要离开触摸屏，使得滑动的速度可以比点击速度更快，从而能够使输入更加便捷。
然而在实现上，目前现有的滑动输入技术通常根据用户滑动过程中关键点的坐标查找对应的字母，当按键与感应设备的接触面面积较大时确实能够具有较高的筛选效率，降低对后续计算处理能力的要求。然而随着终端设备的不断小型化，设置在终端设备上的软件键盘的尺寸也随之缩小，相应的，各个按键与感应设备的接触面积也变得更小，这使得用户误按或系统误识别、误判断的几率大大增加。在这种情况下，通过关键点的坐标进行筛选，由于滑动输入过程中受到用户输入习惯及其它因素的影响较多，反而容易造成较高的误判率，大大影响了预测的准确性不够高。
参考图1至图3，由于不同的用户具有不同的输入习惯，例如用户可能采用图1所示轨迹输入单词“kate”，也有可能按照图2所示轨迹依次划过字母“k”“a”“t”“e”，以及在输入的过程中有可能产生误按、错按的情况，例如，如图3所示，用户在经过字母“t”时，误划入字母“t”临近的字符“y”区域内，或者其它误按错按的情况。目前常用的滑行输入方法，一般通过检测滑动轨迹中的拐点，将其作为关键点并根据该关键点在特征词库中进行查询从而获得对应的候选词，然而发明人发现，在连续滑动输入单词的过程中，由于存在上述各种可能的意外或不同的用户输入习惯，在对滑动轨迹中的拐点进行检测的过程中，容易出现拐点的遗漏、或误增、或错判，这些都会对检索结果的准确性造成影响，同时也会增大数据计算量，增加硬件负担。
而且，通过关键点的坐标进行查找时，待计算的数据量大，计算复杂度高，尽管能够通过具有更高运算能力的CPU实现计算过程，然而庞大的计算量通常会耗费更多的电量，在电池容量一定的情况下，其往往会对电子设备的续航能力造成较大的限制；此外，现有技 术往往需要滑动出一个单词的完整轨迹才能完成计算，对于一些较长的单词，用户很难划出完整的正确的轨迹。
发明内容
本发明的目的是克服了上述现有技术中的缺点，提供一种更准确、更快速、更易用的滑动输入系统和方法。
根据本发明的一方面，本发明提供了一种连续滑动输入单词的方法，包括：检测感应设备的运动轨迹，记录输入信息，从所述输入信息中获取特征信息，所述特征信息至少包括角度信息和运动轨迹长度中至少一个；根据所述特征信息在特征词库中进行检索，将符合以下至少一项：其角度信息与所述特征信息中角度信息相适应或其运动轨迹长度与所述特征信息中运动轨迹长度相适应的单词作为匹配的单词，获得初步筛选单词集合；根据所述初步筛选单词集合中每一个单词的实际运动轨迹，计算其与所述单词对应的第一标准轨迹之间的距离量度，按照所述距离量度对单词进行过滤，获得粗过滤单词集合；计算粗过滤单词集合中每一个单词的实际运动轨迹与其对应的第二标准轨迹的相似度，根据相似度对所述单词进行排序，并按照所述排序输出至用户。
根据本发明的另一方面，本发明还提供了一种连续滑动输入单词的系统，包括：输入模块，适于接收感应设备的滑动输入；输出模块，适于将输出结果反馈至用户；输入信息记录模块，适于对感应设备的输入进行检测并记录输入信息；特征信息获取模块，适于从所记录的输入信息中获取至少包含角度信息和实际运动轨迹长度之一的特征信息；初步筛选模块，适于根据所述特征信息从特征词库中检索匹配的单词，获得初步筛选单词集合；粗过滤模块，适于对初步筛选单词集合中的每一个单词计算器实际运动轨迹与标准轨迹之间的距离量度，根据所述距离量度进行过滤，获得粗过滤单词集合；精过滤模块，适于对所述粗过滤单词集合中的每个单词，分别计算其标准轨迹与实际运动轨迹之间的相似度，并对精过滤所获得单词进行排序，将单词 按照所述顺序输出至所述输出模块。
根据本发明的另一方面，本发明还提供了一种连续滑动输入单词的系统，包括：交互设备，适于接收用户的滑动输入，对用户的输入进行检测和记录，以及适于将输入结果反馈给用户；处理器，适于执行以下操作：根据所述特征信息在特征词库中进行检索，将符合以下至少一项：其角度信息与所述特征信息中角度信息相适应或其运动轨迹长度与所述特征信息中运动轨迹长度相适应的单词的集合作为初步筛选单词集合；根据所述初步筛选单词集合中每一个单词的实际运动轨迹，计算其与所述单词对应的第一标准轨迹之间的距离量度，按照所述距离量度对单词进行过滤，获得粗过滤单词集合；计算粗过滤单词集合中每一个单词的实际运动轨迹与其对应的第二标准轨迹的相似度，根据相似度对所述单词进行排序，并按照所述排序输出至用户。
相较于现有技术，本发明提供了在电子设备上基于软件键盘连续滑动输入单词的方法和系统，其中，交互设备通过对用户滑动输入的输入信息进行记录并传输至处理器，处理器根据输入信息中的角度信息、实际轨迹长度等特征信息，从特征词库中快速过滤出有可能的数百或数千个相似单词的初步筛选单词集合，接着使用位置信息进一步对初步筛选单词集合进行过滤，获得粗过滤单词集合，在此基础上通过计算每个单词实际运动轨迹与其标准轨迹的相似度，并结合单词的优先级、环境词信息、历史输入信息以及是否是预测结果等进行加权计算，根据计算结果对候选词进行排序，进而输出至交互设备，反馈给用户，从而能够利用滑动过程中的各种特征，在保证速度的前提下，有效地减轻了用户误操作和系统误辨识可能带来的不良影响，极大地提高了单词的输入速度和准确度，明显降低用户操作复杂度，实现过程简单，使用方便快捷，工作性能稳定可靠，适用范围较为广泛，大大促进了电子设备触摸屏幕信息输入方法的大规模应用，给人们的工作和生活带来了很大的便利。
附图说明
图1至图3为在电子设备上采用软件键盘滑动输入单词“kate”的实现示意图；
图4为本发明连续滑动输入单词系统一种实施方式的结构示意图；
图5为本发明连续滑动输入单词方法一种实施方式的流程示意图；
图6为图5所示步骤S1一种实施方式的流程示意图；
图7为本发明连续滑动输入单词方法一种实施方式中特征词库的逻辑存储结构示意图；
图8为本发明连续滑动输入单词方法一种实施方式中计算实际运动轨迹中横坐标距离量度的示意图；
图9为本发明连续滑动输入单词方法一种实施方式中计算每个单词标准轨迹及其实际轨迹相似度的流程示意图；
图10为本发明连续滑动输入单词系统一种实施方式的框架示意图；以及
图11为本发明连续滑动输入单词系统另一种实施方式的框架示意图。
具体实施方式
在详细说明根据本发明的实施例前，应该注意到的是，所述实施例主要在于向电子设备中输入文本有关的方法步骤和设备组件的组合。因此，所属设备组件和方法步骤已经在附图中通过常规符号在适当的位置表示出来了，并且只示出了与理解本发明的实施例有关的细节，以免因对于得益于本发明的本领域普通技术人员而言显而易见的那些细节而模糊了本公开内容。
在本文中，诸如左和右，上和下，前和后，第一和第二之类的关系术语仅仅用来区分一个实体或动作与另一个实体或动作，而不一定要求或暗示这种实体或动作之间的任何实际的这种关系或顺序。术语“包括”、“包含”或任何其他变体旨在涵盖非排他性的包含，由此使得包括一系列要素的过程、方法、物品或者设备不仅包含这些要素，而且还包含没有明确列出的其他要素，或者为这种过程、方法、物品或者设备所固有的要素。
出于本公开的目的，术语“键盘”是广义定义的，包括任何具有确定区域的输入部件，包含但不局限于硬件机械式键盘、硬件感应式键盘、触摸屏上的软件键盘等。本发明的实施例默认以触摸屏上的软件键盘为例说明，但本领域的普通技术人员可以理解，本发明所提及的方法和系统也可以应用于硬件键盘。
参考图4，本发明所涉及的连续滑动输入单词系统的一种实施方式，可至少包括交互设备110以及处理器120，其中，交互设备110可包括移动通信终端设备上的键盘101和文本显示区域102，例如配备软件键盘的触摸式显示屏；处理器120可以是CPU或专用集成电路芯片。当用户在键盘101上输入后，其输入信息经由处理器120进行处理，处理后的文本在文本显示区域102中进行显示，从而反馈给用户。
键盘101为软件键盘，其键盘布局可采用QWERTY的单键键盘布局，包含了26个英文字母，数字、符号和空格键，每个按键代表一个字母，数字或符号可与字母复合在同一个按键上；也可采用AZERTY的单键键盘布局。在其它实施方式中，键盘101也可采用其它键盘布局，或适用于除英语以外的其它语种，且字母、数字、符号的具体内容和位置可根据具体语言的需求或键盘排布而改变。
在不同的移动通信终端设备上，键盘101和文本显示区域102可集成在同一个交互设备上，也可分别设置在分离的交互设备上。当键盘101和文本显示区域102集成在同一个交互设备上时，可根据需要 将键盘101展开或收起，从而获得大小可变化的文本显示区域102；也可对键盘101的大小尺寸进行调节。
在连续滑动输入单词的过程中，例如，结合图1，用户采用手指、触摸笔等感应设备130依次在键盘101区域划过字母“k”“a”“t”“e”的大致区域，交互设备110检测感应设备130的动作，记录感应设备130的输入信息，比如感应设备130停留的位置、运动的角度、运动的轨迹长度等等，并通过这些输入信息获得感应设备130的运动轨迹；交互设备110将所记录的输入信息以及运动轨迹传送至处理器120，处理器120进一步对其进行处理，通过在特征词库中进行检索、过滤和匹配等操作，获得与感应设备130的运动轨迹相匹配的单词集合，接着，进一步根据设定顺序对匹配的单词集合进行排序，并将排序结果呈现在文本显示区域102中，反馈给用户。
此外，本发明所涉及的连续滑动输入单词的系统的一种实施方式还可包括存储设备140，适于存储为了实现连续滑动输入单词的方法中所产生或有可能产生的数据，例如特征词库、输入信息、运动轨迹等。存储设备140可以是任何记忆介质，例如随机访问内存(RAM)，只读内存(ROM)，闪存(Flash)，硬盘等。
参考图5，在本发明连续滑动输入单词的方法的一种实施方式中，可包括以下步骤：
步骤S1，检测感应设备的运动，记录输入信息。
其中，参考图6，在一种实施方式中，首先，执行步骤S110，可检测感应设备是否与交互设备相接触。当存在接触时，执行步骤S120，记录感应设备的位置，并且通过步骤S130，检测感应设备是否存在移动。当检测到感应设备存在移动时，持续记录感应设备的位置；当检测到感应设备停止移动时，检测其是否与交互设备停止接触，记录其位置并且也可记录其在停止位置停留的时间。具体来说，当感应设备停止移动之后仍保持与交互设备进行接触时，用户可能是在犹豫后续待输入字母或观望已输入轨迹对应的单词，此时可执行步骤 S140，通过将感应设备在停止位置停留的时间与预定时间阈值进行比较，判断是否根据已输入的信息输出对应的单词，例如当停留时间超过预定时间阈值时，执行步骤S150，输出与已输入轨迹对应的单词。在上述检测过程中，当检测到感应设备停止与交互设备进行接触时，根据已输入信息获得对应的单词集合，并将所获得的单词集合输出给用户。
在记录感应设备所在位置时，还可包括进一步记录其位置对应的角度信息，例如计算各个位置点的斜率，以获得与该位置对应的角度信息；还可包括进一步记录感应设备实际的运动轨迹长度，例如根据感应设备的运动轨迹，对每2个连续位置之间的距离相加，近似计算轨迹的长度，采用类似积分的方式，可以很容易地提取出每一段用户滑动轨迹的长度。
在实际处理过程中，由于用户输入时手抖动等原因会造成滑动轨迹数据中的嗓点，因此可对输入信息进行一系列预处理工作，并对预处理后的结果进行记录。例如可包括通过大小归一化，消除轨迹因为手机触摸屏大小差异而带来的尺寸差异；可包括通过轨迹平滑，例如采用均值滤波器等方法对轨迹进行平滑，消除因手抖动等原因会造成滑动轨迹数据中的噪点；还可包括点距归一化等处理。
步骤S2，从所述输入信息中获取角度信息、运动轨迹长度等特征信息；根据所述特征信息，在特征词库中进行检索，获得初步筛选单词集合。
在一种实施方式中，步骤S1所记录的输入信息中并不包含角度信息、运动轨迹长度等，步骤S2根据所记录的位置信息进行计算，获取对应的角度信息、运动轨迹长度等特征信息。
在一种具体实施例中，可根据所述输入信息计算感应设备起始位置的角度，并在词库中查找具有匹配起始点角度的单词。例如，当用户通过qwerty布局的键盘输入单词“what”时，其手指从首字母“w”滑至第二个字母“h”时，其起始位置的角度与用户在输入“snake” 时起始位置的角度相同，因此，当根据起始位置的角度信息在特征词库中进行检索时，单词“snake”也会被加入至初步筛选单词集合中。
在另一种具体实施例中，可根据所述输入信息计算感应设备终止位置的角度，并在词库中查找具有匹配终止位置角度的单词。例如，当用户通过azerty布局的键盘输入单词“insist”时，其手指从字母“s”滑至末字母“t”时，其终止位置角度与用户在输入“candy”时终止位置角度之差在预定阈值以内，因此，当根据终止位置的角度信息在特征词库中进行检索时，单词“candy”也会被加入至初步筛选单词集合中。
在另一种具体实施方式中，还可根据起始位置的角度信息和终止位置的角度信息的组合，或其它指定位置的角度信息，或其与起始位置或终止位置的角度信息的组合，在特征词库中进行检索，以获得初步筛选单词集合。
其中，可设置预定角度阈值范围，当特征词库中单词的对应角度与实际输入角度信息相差值不超过预定角度阈值范围时，可认为其与角度信息相匹配；例如，可计算单词的起始点角度与实际输入起始点角度之间的差值，当所述差值在预定角度阈值范围内时，则该单词为匹配起始角度单词。
在又一种具体实施例中，也可从所述输入信息中获取感应设备实际运动轨迹的长度，在词库中查找具有匹配运动轨迹长度的单词。其中，可设置预定长度阈值范围，当特征词库中单词的对应轨迹长度与实际运动轨迹长度相差值不超过预定长度阈值范围时，可认为其与实际运动轨迹长度相匹配；例如，可计算单词的轨迹长度与实际运动轨迹长度之间的差值，当所述差值在预定长度阈值范围内时，则该单词为匹配轨迹长度单词。
在又一种具体实施例中，还可将轨迹长度与角度信息相结合，在特征词库中进行初步筛选。例如，可先根据轨迹长度进行第一次初步筛选，然后再根据角度信息对第一次初步筛选进行再次筛选，根据筛 选结果获得初步筛选单词集合；或者调换前后两次筛选的次序，即先根据角度信息进行第一次初步筛选，再根据轨迹长度进行再次筛选；或者在构建特征词库时，将单词按照轨迹长度或者角度信息建立索引序列，再根据角度信息或轨迹长度对其进行检索，获得初步筛选单词集合。
采用指定位置的角度信息或轨迹长度作为特征信息，在特征词库中进行初步筛选，降低了对设备尺寸的要求以及降低了对实际运动输入轨迹一致性的要求，增强了对用户误滑动和系统误识别的包容性，大大提高了识别的准确率。
在本发明另一种实施方式中，步骤S1所记录的输入信息中包括角度信息、运动轨迹长度等特征信息，步骤S2从中进行提取，即可获得所需要的相关特征信息。其中，步骤S2可提取出相对于感应设备的运动轨迹而言，全部的特征信息并对这些特征信息进行处理；也可仅对预定数量的角度信息、轨迹长度等特征信息进行处理，例如，当根据所有输入信息提取了特征信息之后，采用预定数量的特征信息进行后续的检索、筛选等操作，其中，预定数量可为总数据的75％-95％。大量数据训练表明，采用85％的输入信息所对应的特征信息，能够在精确性和运算量两者之间达到较好的平衡，从而能够在不影响精确性的前提下，减少运算的数据量，降低对硬件设备的要求，提高处理效率，为用户提供更加快捷、准确的输入体验。
此外，还可包括构建特征词库，该步骤并非是本发明的方法中的必要步骤，也可以采用其它任何能够达到相同或者相应技术效果的处理方式。具体来说，特征词库是以各特征信息为索引建立的单词的词库。在构建过程中，可提取出字典中单词所对应的标准特征信息并将其存储于特征词库模块中。例如，与特征信息中的角度相对应的，可提取出字典中单词所对应的理想角度信息；与特征信息中的运动轨迹长度相对应的，可提取出字典中单词所对应的标准轨迹长度。在一种实施方式中，所构建的特征词库中可包括单词特征表及单词标准轨迹 索引表，所述的单词特征表中的字段可包括单词标识、单词内容，还可包括标准轨迹长度和标准角度信息中至少一项，所述的单词标准轨迹索引表引用单词特征表中的单词标识；该单词标准轨迹可以为根据大量训练数据从用户的滑动轨迹中所提取出的对应单词的大量相似轨迹，或者该单词标准轨迹也可以为单词的连续字母中心在屏幕软件键盘中的直线连线。最终特征词库的逻辑存储结构如图7所示。这里的词库可以使用现有小型的数据库存储，也可以设计私有的存储方式来优化数据的访问速度及使用空间。
步骤S3，对所述输入信息进行处理，根据处理结果对步骤S2所获得的初步筛选单词集合进行过滤。其中，对所述初步筛选单词集合进行处理又可进一步包括对其进行粗过滤和精过滤。
如前文所述，由于采用角度信息进行检索获得初步筛选单词集合，其所包含的单词数量通常比较多，发明人发现，如果对初步筛选单词集合逐一进行相似度计算和匹配，不仅耗费资源较多，而且计算时间相对较长，因此，发明人通过先对输入信息进行粗过滤，然后根据粗过滤的结果再次进行精过滤，从而节省资源，提高计算效率。
具体来说，在粗过滤过程中，采用包络算法，对初步筛选所获得的单词集合中的每一个单词计算其实际轨迹与标准轨迹之间的距离量度，根据计算获得的距离量度进行过滤，获得粗过滤单词集合。
在一种实施方式中，可采用以下步骤计算距离量度。首先，根据每个单词的实际运动轨迹计算该单词的包络，具体来说，分别计算实际运动轨迹中每个点的上包络和下包络。在一种具体实施方式中，参考图8，首先，根据输入信息，提取感应设备实际运动轨迹中各个点的横坐标或纵坐标。接着，对实际运动轨迹上每个点的横坐标或纵坐标计算其对应的上包络值和下包络值。
具体来说，比较第N个点与其前后设定数目的点的横坐标或纵坐标，取其最大值为第N个点的上包络值，取其最小值为第N个点的下包络值。
例如，比较第N个点和其前i个点，即第N-i个，…，第N-1个点，以及其后i个点，即第N+1，…，N+i个点的横坐标；
XupperboundN＝Xmax{N-i,…,N-1,N,N+1,…,N+i}，XlowerboundN＝Xmin{N-i,…,N-1,N,N+1,…,N+i}；
以及，比较第N个点和其前i个点，即第N-i个，…，第N-1个点，以及其后i个点，即第N+1，…，N+i个点的纵坐标：
YupperboundN＝Ymax{N-i,…,N-1,N,N+1,…,N+i}，YlowerboundN＝Ymin{N-i,…,N-1,N,N+1,…,N+i}。
其中，当N<i且N>j时，j为1至i之间的任意自然数，可根据运动轨迹，对第N-i个，…，第N-j-1个点的横坐标或纵坐标进行估算；或者，当轨迹包含M个点，以及N+i>M且N+j<M时，j为1至i之间的任意自然数，可根据运动轨迹，对第N+j+1个，…，第N+i个点的横坐标或纵坐标进行估算。
参考图8，遍历实际运动轨迹上所有的点之后，获得分别相对于整个运动轨迹横坐标210的上包络211和下包络212，以及相对于整个运动轨迹纵坐标的上包络和下包络。
接着，计算每个单词的第一标准轨迹。其中，在计算每个单词的第一标准轨迹时，按照实际运动轨迹的采样频率计算第一标准轨迹，使得待获得的第一标准轨迹的采样点数目与实际运动轨迹的采样点数目相同，并在此基础上进行拟合，获得第一标准轨迹。
在其它实施例中，也可以先计算每个单词的第一标准轨迹，或者在实际运算能力允许的情况下，也可在多个处理器上同时计算第一标准轨迹以及计算距离量度。
接着，根据每个单词的第一标准轨迹及其包络，计算其距离量度。具体来说，参考图8，在计算距离量度时，分别将该单词第一标准轨迹的横坐标230与其对应的实际运动轨迹横坐标包络，即上包络211和下包络212，进行比较，计算其第一标准轨迹横坐标超出包络之和； 以及将该单词第一标准轨迹的纵坐标与其对应的实际轨迹纵坐标包络，即上包络221和下包络222，进行比较，计算其第一标准轨迹纵坐标超过包络之和。然后，计算两者的总和，并将其作为该单词的距离量度。
最后，根据所计算出的距离量度，按照从小到大的顺序对其进行排序，选择在先的设定个数的单词，获得粗过滤单词集合。其中，可根据移动终端的实际运算能力对设定个数进行设置，例如，可为几十至几千个。
接下来，对于粗过滤单词集合中的每个单词，进行精过滤，即分别计算其标准轨迹与实际轨迹的相似度。具体来说，参考图9，计算每个单词的标准轨迹与其实际轨迹的相似度，可采用下述步骤：
首先，步骤S410，计算每个单词对应的第二标准轨迹，以及对应的实际运动轨迹。其中，可根据所述输入信息或对所述输入信息进行处理，计算每个单词对应的第二标准轨迹，例如可采用增采样的方式获取第二标准轨迹对应的轨迹点序列。
接着，步骤S420，根据每个单词的第二标准轨迹及其实际运动轨迹，计算其相似度。
例如，在一种具体实施方式中，可包括：
(i)对于所述轨迹点序列中的每一个点s[i]，根据以下公式将第二标准轨迹中的所有点t[j]与该点s[i]之间的距离进行逐一累加计算，并得到相似距离D[i,j]:
D[i,j]＝cost+min(D[i-1,j],D[i,j-1],D[i-1,j-1])；
其中，cost＝distance(s[i],t[j])。distance(s[i],t[j])为s[i]和t[j]两点之间的距离，
D[0,0]＝0,i＝1,2,3,..,n,n为用户轨迹点序列中点的总数，j＝1,2,3,...，m，m为标准轨迹中点的总数，min(D[i-1,j],D[i,j-1],D[i-1, j-1])为D[i-1,j]、D[i,j-1]、D[i-1,j-1]三者中的最小值；
(ii)重复上述步骤(i)，直至遍历所述第二标准轨迹中的所有点，并得到累加计算后的标准轨迹与实际轨迹之间的相似距离D[n,m]；
其中，上述相似距离即为该单词的标准轨迹特征与实际轨迹的相似度。
接着，执行步骤S4，进一步对过滤后所获得的候选词集合进行排序并输出。具体来说，获得每个候选词的相似度之后，还可进一步再结合候选词的优先级、环境词信息、历史输入信息以及是否是预测结果等进行加权计算，从而获得该候选词的分数，最后根据分数对经过精筛选的候选词集合进行排序，并按顺序依次将候选词输出至交互设备，实现对用户输入的反馈。
在本发明连续滑动输入单词系统中，交互设备110至少可包括适于用户进行滑动输入的输入设备，例如触摸感应输入设备、无接触式感应式输入设备，以及将输入结果反馈给用户的输出设备，例如视频输出设备、音频输出设备等。其中，输入设备可与输出设备集成在一起，例如交互设备110可为集成了软件键盘的触摸式显示屏；此外，输入设备也可与输出设备相分离，例如用户在触摸板上进行输入，其输出通过音频输出设备反馈至用户，或者其输出通过与触摸板分离的显示屏进行显示。
具体来说，交互设备110还可对用户的输入进行检测和记录，例如检测用户所操作的感应设备是否与其存在接触；当存在接触时，记录感应设备与其接触时的所在位置，并且检测感应设备是否存在移动；当检测到感应设备存在移动时，持续记录感应设备的位置；当检测到感应设备停止移动时，检测其是否停止接触，记录其位置以及其在停止接触的位置所停留的时间。当停留时间超过预定时间阈值时，输出与已输入轨迹对应的单词。在上述检测过程中，当检测到感应设备停止与交互设备进行接触时，根据已输入信息获得对应的单词集合，并将所获得的单词集合输出给用户。
在本发明另一种实施方式中，交互设备110还可对输入信息进行预处理。例如，可对输入信息进行增采样；又例如可对输入信息进行平滑处理。由于用户输入时手抖动等原因会造成滑动轨迹数据中的嗓点，因此可对轨迹进行一系列预处理工作，例如可包括通过大小归一化，消除轨迹因为手机触摸屏大小差异而带来的尺寸差异；可包括通过轨迹平滑，例如采用均值滤波器等方法对轨迹进行平滑，消除因手抖动等原因会造成滑动轨迹数据中的噪点；还可包括点距归一化等处理。
在本发明又一种实施方式中，交互设备110还可对输入信息进一步进行处理，例如，可根据感应设备所在位置，计算其对应的角度信息；还可根据感应设备所在位置，计算其实际运动轨迹长度。当计算角度信息以及实际运动轨迹长度时，可根据未进行预处理的输入轨迹，或平滑后的输入轨迹，或增采样后的输入轨迹，进行计算；也可根据不同的输入信息分别计算角度信息或实际运动轨迹长度，例如可根据平滑后的输入轨迹计算角度信息，以及根据增采样后的输入轨迹计算运动轨迹长度。
在本发明又一种实施方式中，交互设备110还可根据输入信息进行处理，对于所获得的特征信息，提取预定数量进行后续处理。在一种实施方式中，例如根据输入信息计算所对应的特征信息后，提取75％-95％的特征信息进行后续运算。在一种具体实施例中，交互设备110采用85％的输入信息所对应的特征信息，从而能够在精确性和运算量两者之间达到较好的平衡，为用户提供更加快捷、准确的输入体验。
在其它的实施方式中，交互设备110可对输入信息进行记录并将所记录的输入信息传输至处理器120，或者，交互设备110对输入信息进行记录和预处理并将经预处理后的输入信息传输至处理器120，处理器120从所接收的输入信息或经预处理的输入信息中提取出对 应的特征信息。其中，处理器120可根据全部输入信息提取对应的特征信息，也可根据预定数量的输入信息提取特征信息，例如提取75％-95％的输入信息，尤其是85％的输入信息，所对应的特征信息。
当处理器120获得特征信息之后，在特征词库中进行检索，找到与特征信息相匹配的单词，获得初步筛选单词集合；接着，进一步根据输入信息，对初步筛选单词集合先进行粗过滤，然后再进行精过滤。
在本发明一种实施方式中，还可包括存储器140，适于存储特征词库，以及筛选或过滤的单词集合。存储器140还可适于存储运行本发明的计算机程序。其中，存储器140可包括一个或多个磁性存储介质，例如硬盘；一个或多个光学存储介质，例如光盘(CD)，可擦写光盘(CD-Rs)，可反复擦写光盘(CD±RWs)，DVDs，可擦写DVD(DVD±Rs)，可反复擦写DVD(DVD±RWs)，蓝光DVD；一个或多个半导体存储介质，例如闪存，SD卡，记忆棒，或其它适配的电脑可读介质。
参考图10，图示为本发明连续滑动输入单词系统一种实施方式的框架示意图，其中，交互设备810可包括：输入模块801，适于接收感应设备的滑动输入；输出模块802，适于将输出结果反馈至用户；输入信息记录模块803，适于对感应设备的输入进行检测并记录输入信息。
处理器820可包括：特征信息获取模块821，适于从所记录的输入信息中获取角度信息、轨迹长度等特征信息；初步筛选模块822，适于根据所述角度信息、轨迹长度等特征信息从特征词库830中检索匹配的单词，获得初步筛选单词集合；粗过滤模块823，适于采用包络算法，对初步筛选单词集合中的每一个单词计算其实际运动轨迹与标准轨迹之间的距离量度，根据计算获得的距离量度进行过滤，获得粗过滤单词集合；精过滤模块824，适于对所述粗过滤单词集合中的每个单词，分别计算其标准轨迹与实际运动轨迹的相似度，并对精过滤所获得的单词进行排序后输出至输出模块802。
其中，在一种实施方式中，输入信息记录模块803还可适于对所记录的输入信息进行预处理，例如平滑处理，例如增采样，并将预处理后的输入信息输出至处理器820；在另一种实施方式中，输入信息记录模块803仅对输入信息进行记录，由特征信息获取模块821对输入信息进行预处理，根据预处理后的结果提取角度信息、轨迹长度等特征信息，并将预处理后的结果传输至初步筛选模块822、粗过滤模块823以及精过滤模块824。
参考图11，图示为本发明连续滑动输入单词系统另一种实施方式的框架示意图，由交互设备910完成对特征信息的提取操作，其中交互设备910可包括特征信息获取模块904。交互设备910将特征信息和输入信息传输至处理器920，通过处理器920中的初步筛选模块921、粗过滤模块922以及精过滤模块923实现后续的筛选、过滤，以获得候选词，交互设备910通过将候选词按照预定的规则进行排序，并按照排序输出，从而反馈给用户。
下面结合本发明所公开的系统和方法，以单词quick的标准轨迹和用户滑动输入的轨迹为例，对其发明思路进行进一步阐述。
在一种实现过程中，过程如下：
(1)当用户的手指或感应设备从按键q附近的点开始滑动时，交互设备记录开始滑动时的位置信息以及角度信息；
(2)用户手指或感应设备在键盘上滑动，交互设备持续记录手指或感应设备的位置信息；
(3)当手指或感应设备在按键k附近被提起或停留时间超过预定时间阈值时，交互设备记录手指或感应设备停止滑动时的位置信息及角度信息，并计算至手指或感应设备停止运动时止的滑动轨迹长度；
(4)交互设备将所记录的位置信息、角度信息以及滑动轨迹长度传输至处理器，处理器根据所述位置信息、角度信息和滑动轨迹长 度对用户运动轨迹进行匹配。
在另一种实现过程中，匹配过程如下：
(21)当用户的手指或感应设备从按键q附近的点开始滑动时，交互设备记录开始滑动时的位置信息；
(22)用户手指或感应设备在键盘上滑动，交互设备持续记录手指或感应设备的位置信息；
(23)当手指或感应设备在按键k附近被提起或停留时间超过预定时间阈值时，交互设备记录手指或感应设备停止滑动时的位置信息；
(24)交互设备将所有记录的位置信息传输至处理器，处理器计算开始位置和终止位置的角度信息，以及计算至手指或感应设备停止运动时止的滑动轨迹长度，并根据所述位置信息、角度信息及滑动轨迹长度对用户运动轨迹进行匹配。
在另一种实现过程中，匹配过程如下：
(31)当用户的手指或感应设备从按键q附近的点开始滑动时，交互设备记录开始滑动时的位置信息及角度信息；
(32)用户手指或感应设备在键盘上滑动，交互设备持续记录手指或感应设备的位置信息；
(33)当手指或感应设备在按键k附近被提起或停留时间超过预定时间阈值时，交互设备记录手指或感应设备停止滑动时的位置信息和角度信息，并计算至手指或感应设备停止运动时止的滑动轨迹长度；
(34)交互设备将所述开始位置的角度信息、滑动轨迹长度及预定数量的位置信息传输至处理器，例如占轨迹，处理器根据所接收的角度信息、位置信息及滑动轨迹长度对用户运动轨迹进行匹配。
在本发明的上述说明中，其中示范了如何滑动输入英文单词。实际上，本发明的连续滑动输入单词的方法也可以支持其它语种的语言，且不受键盘布局的影响。该滑动输入文本的方法支持的语言包括但不局限于西语(英语、法语、德语等)、拉丁语(西班牙语、葡萄牙语等)和亚洲语言(阿拉伯语、印度语、中文、日文、韩文等)。在键盘布局上，该滑动输入文本的方法支持最流行的QWERTY键盘布局，法语中常用的AZERTY键盘布局，以及其他键盘布局。
相较于现有技术，本发明根据用户输入的角度信息、轨迹长度在特征词库中进行筛选，并采用包络进行粗过滤以及进一步计算相似度完成精过滤，获得候选词，有效地减轻了用户误操作和系统误辨识可能带来的不良影响，减少了数据运算量，节省了能耗，为用户提供了更为便利和快捷的输入体验。此外，本发明能够有效提高单词的输入速度和准确度、明显降低用户操作复杂度、实现过程简单、使用方便快捷、计算复杂度低、适用范围较为广泛的电子设备上基于屏幕软件键盘实现滑动输入文本的系统。
在此说明书中，本发明已参照其特定的实施例作了描述。但是，很显然仍可以作出各种修改和变换而不背离本发明的精神和范围。因此，说明书和附图仅作为本发明的示例性说明，并不应对本发明的发明思路造成限制。

Claims (17)

  1. 一种连续滑动输入单词的方法，其特征在于，包括：
    检测感应设备的运动轨迹，记录输入信息，从所述输入信息中获取特征信息，所述特征信息至少包括角度信息和运动轨迹长度中至少一个；
    根据所述特征信息在特征词库中进行检索，将符合以下至少一项：其角度信息与所述特征信息中角度信息相适应或其运动轨迹长度与所述特征信息中运动轨迹长度相适应的单词作为匹配的单词，获得初步筛选单词集合；
    根据所述初步筛选单词集合中每一个单词的实际运动轨迹，计算其与所述单词对应的第一标准轨迹之间的距离量度，按照所述距离量度对单词进行过滤，获得粗过滤单词集合；以及
    计算粗过滤单词集合中每一个单词的实际运动轨迹与其对应的第二标准轨迹的相似度，根据相似度对所述单词进行排序，并按照所述排序输出至用户。
  2. 如权利要求1所述的方法，其特征在于，所述角度信息包括起始点角度和终止点角度中至少一个。
  3. 如权利要求1所述的方法，其特征在于，所述检测感应设备的运动并记录输入信息包括以下至少一项：
    当检测到感应设备与交互设备存在接触，并相对于交互设备产生运动时，记录感应设备运动时的角度信息；
    当检测到感应设备与交互设备存在接触，并相对于交互设备产生运动时，记录感应设备实际运动轨迹长度。
  4. 如权利要求1所述的方法，其特征在于，所述检测感应设备的运动并记录输入信息包括以下至少一项：
    当检测感应设备与交互设备存在接触时，记录感应设备运动的起始位置；
    当检测到感应设备存在移动时，持续记录感应设备的位置；
    当检测到感应设备停止移动时，记录感应设备在停止位置停留的时间；
    当检测到感应设备停止移动时，记录感应设备的位置并检测感应设备是否离开交互设备。
  5. 如权利要求1所述的方法，其特征在于，从所述输入信息中获取特征信息包括：
    根据所述输入信息计算感应设备指定位置的角度信息或实际运动轨迹长度或其组合；
    根据预定数量的输入信息计算对应的特征信息；
    从所述输入信息计算获得的特征信息中，提取相对于感应设备的运动轨迹而言预定数量的特征信息。
  6. 如权利要求1所述的方法，其特征在于，所述根据初步筛选单词集合中每一个单词的实际运动轨迹，计算其与所述单词对应的第一标准轨迹之间的距离量度，按照所述距离量度对单词进行过滤，包括：
    根据每个单词的实际运动轨迹，分别计算上包络和下包络，以及按照实际运动轨迹的采样频率计算第一标准轨迹；
    分别计算第一标准轨迹的横坐标和纵坐标超出包络之和，将两者之和作为该单词的距离量度。
  7. 如权利要求1所述的方法，其特征在于，所述计算粗过滤单词集合中每一个单词的实际运动轨迹与其对应的第二标准轨迹的相似度，包括：
    根据所述输入信息，获得每个单词的第二标准轨迹；
    根据每个单词实际运动轨迹上的每个轨迹点，计算其与第二标准轨迹中各个轨迹点之间的相似距离；
    重复上述过程，直至遍历所述第二标准轨迹中的所有点，将计算获得的相似距离进行累加。
  8. 如权利要求1所述的方法，其特征在于，所述根据相似度对 单词进行排序还包括：将所述相似度与单词的优先级、环境词信息、历史输入信息以及是否是预测结果进行加权计算，获得所述单词的分数，按照分数将单词进行排序。
  9. 如权利要求1所述的方法，其特征在于，还包括对所述输入信息进行预处理。
  10. 一种连续滑动输入单词的系统，其特征在于，包括：
    输入模块，适于接收感应设备的滑动输入；
    输出模块，适于将输出结果反馈至用户；
    输入信息记录模块，适于对感应设备的输入进行检测并记录输入信息；
    特征信息获取模块，适于从所记录的输入信息中获取至少包含角度信息和实际运动轨迹长度之一的特征信息；
    初步筛选模块，适于根据所述特征信息从特征词库中检索匹配的单词，获得初步筛选单词集合；
    粗过滤模块，适于对初步筛选单词集合中的每一个单词计算器实际运动轨迹与标准轨迹之间的距离量度，根据所述距离量度进行过滤，获得粗过滤单词集合；以及
    精过滤模块，适于对所述粗过滤单词集合中的每个单词，分别计算其标准轨迹与实际运动轨迹之间的相似度，并对精过滤所获得单词进行排序，将单词按照所述顺序输出至所述输出模块。
  11. 如权利要求10所述的系统，其特征在于，所述输入信息记录模块对感应设备的输入进行以下至少一项检测：
    当检测到感应设备与交互设备存在接触，并相对于交互设备产生运动时，记录感应设备运动时的角度信息；
    当检测到感应设备与交互设备存在接触，并相对于交互设备产生运动时，记录感应设备实际运动轨迹长度。
  12. 如权利要求10所述的系统，其特征在于，所述输入信息记录模块对感应设备的输入进行以下至少一项检测：
    当检测感应设备与交互设备存在接触时，记录感应设备运动的起始位置；
    当检测到感应设备存在移动时，持续记录感应设备的位置；
    当检测到感应设备停止移动时，记录感应设备在停止位置停留的时间；
    当检测到感应设备停止移动时，记录感应设备的位置并检测感应设备是否离开交互设备。
  13. 如权利要求10所述的系统，其特征在于，所述特征信息获取模块通过以下至少一种方式获取特征信息：
    根据所述输入信息计算感应设备指定位置的角度信息或实际运动轨迹长度或其组合；
    根据预定数量的输入信息计算对应的特征信息；
    从所述输入信息计算获得的特征信息中，提取相对于感应设备的运动轨迹而言预定数量的特征信息。
  14. 如权利要求10所述的系统，其特征在于，所述粗过滤模块执行以下操作：
    根据每个单词的实际运动轨迹，分别计算上包络和下包络，以及按照实际运动轨迹的采样频率计算第一标准轨迹；
    分别计算第一标准轨迹的横坐标和纵坐标超出包络之和，将两者之和作为该单词的距离量度。
  15. 如权利要求10所述的系统，其特征在于，所述精过滤模块执行如下操作：
    根据所述输入信息，获得每个单词的第二标准轨迹；
    根据每个单词实际运动轨迹上的每个轨迹点，计算其与第二标准轨迹中各个轨迹点之间的相似距离；
    重复上述过程，直至遍历所述第二标准轨迹中的所有点，将计算获得的相似距离进行累加。
  16. 一种连续滑动输入单词的系统，其特征在于，包括：
    交互设备，适于接收用户的滑动输入，对用户的输入进行检测和记录，以及适于将输入结果反馈给用户；
    处理器，适于执行以下操作：
    根据所述特征信息在特征词库中进行检索，将符合以下至少一项：其角度信息与所述特征信息中角度信息相适应或其运动轨迹长度与所述特征信息中运动轨迹长度相适应的单词的集合作为初步筛选单词集合；
    根据所述初步筛选单词集合中每一个单词的实际运动轨迹，计算其与所述单词对应的第一标准轨迹之间的距离量度，按照所述距离量度对单词进行过滤，获得粗过滤单词集合；以及
    计算粗过滤单词集合中每一个单词的实际运动轨迹与其对应的第二标准轨迹的相似度，根据相似度对所述单词进行排序，并按照所述排序输出至用户。
  17. 如权利要求16所述的系统，其特征在于，还包括：存储器，适于存储特征词库，筛选或过滤的单词集合，以及运行处理器所述操作的计算机程序。
PCT/CN2015/070783 2013-11-15 2015-01-15 连续滑动输入单词的方法和系统 WO2015070825A2 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310574003.1 2013-11-15
CN201310574003.1A CN104346067B (zh) 2013-11-15 2013-11-15 连续滑动输入单词的方法和系统

Publications (2)

Publication Number Publication Date
WO2015070825A2 true WO2015070825A2 (zh) 2015-05-21
WO2015070825A3 WO2015070825A3 (zh) 2015-07-23

Family

ID=52501804

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/070783 WO2015070825A2 (zh) 2013-11-15 2015-01-15 连续滑动输入单词的方法和系统

Country Status (3)

Country Link
US (1) US10082952B2 (zh)
CN (1) CN104346067B (zh)
WO (1) WO2015070825A2 (zh)

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104346067B (zh) * 2013-11-15 2019-02-05 上海触乐信息科技有限公司 连续滑动输入单词的方法和系统
US10359929B2 (en) * 2015-11-09 2019-07-23 Analog Devices, Inc. Slider and gesture recognition using capacitive sensing
CN105824568B (zh) * 2016-03-30 2020-01-24 苏州佳世达电通有限公司 电子装置及其操作方法
CN106293128B (zh) * 2016-08-12 2019-06-28 清华大学 盲式文字输入方法、盲式文字输入装置和计算装置
CN107817942B (zh) * 2016-09-14 2021-08-20 北京搜狗科技发展有限公司 一种滑行输入方法、系统和一种用于滑行输入的装置
US11609692B2 (en) * 2017-04-07 2023-03-21 Hewlett-Packard Development Company, L.P. Cursor adjustments
KR102543867B1 (ko) * 2018-03-28 2023-06-14 사로니코스 트레이딩 앤드 서비스즈, 유니페쏘알 엘디에이 터치스크린 상의 터치의 신뢰성을 향상시키기 위한 모바일 디바이스 및 방법
CN111198644B (zh) * 2018-11-16 2023-10-27 西安易朴通讯技术有限公司 智能终端的屏幕操作的识别方法及系统
CN112261459B (zh) * 2020-10-23 2023-03-24 北京字节跳动网络技术有限公司 一种视频处理方法、装置、电子设备和存储介质
KR20220128892A (ko) 2021-03-15 2022-09-22 삼성전자주식회사 전자 장치 및 오타 보정 방법
CN113749620B (zh) * 2021-09-27 2024-03-12 广州医科大学附属第一医院（广州呼吸中心） 一种睡眠呼吸暂停检测方法、系统、设备及存储介质

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW338815B (en) * 1995-06-05 1998-08-21 Motorola Inc Method and apparatus for character recognition of handwritten input
US7382358B2 (en) * 2003-01-16 2008-06-03 Forword Input, Inc. System and method for continuous stroke word-based text input
US8225203B2 (en) * 2007-02-01 2012-07-17 Nuance Communications, Inc. Spell-check for a keyboard system with automatic correction
EP2545426A4 (en) * 2010-03-12 2017-05-17 Nuance Communications, Inc. Multimodal text input system, such as for use with touch screens on mobile phones
CN102736821B (zh) * 2011-03-31 2017-06-16 深圳市世纪光速信息技术有限公司 基于滑动轨迹确定候选词的方法和装置
CN102736825A (zh) * 2011-04-08 2012-10-17 许平 汉语拼音手势输入系统及其输入方法
US9182831B2 (en) * 2011-04-09 2015-11-10 Shanghai Chule (Cootek) Information Technology Co., Ltd. System and method for implementing sliding input of text based upon on-screen soft keyboard on electronic equipment
US20140189569A1 (en) * 2011-07-18 2014-07-03 Syntellia, Inc. User interface for text input on three dimensional interface
CN102520804A (zh) * 2011-12-08 2012-06-27 周志农 触摸屏软键盘及输入方法
CN107273022A (zh) * 2012-05-17 2017-10-20 联发科技（新加坡）私人有限公司 自动纠错方法及装置和移动终端
US9304595B2 (en) * 2012-10-19 2016-04-05 Google Inc. Gesture-keyboard decoding using gesture path deviation
CN104346067B (zh) * 2013-11-15 2019-02-05 上海触乐信息科技有限公司 连续滑动输入单词的方法和系统

Also Published As

Publication number Publication date
US10082952B2 (en) 2018-09-25
WO2015070825A3 (zh) 2015-07-23
US20160004433A1 (en) 2016-01-07
CN104346067A (zh) 2015-02-11
CN104346067B (zh) 2019-02-05

Similar Documents

Publication Publication Date Title
WO2015070825A2 (zh) 连续滑动输入单词的方法和系统
EP2698692B1 (en) System and method for implementing sliding input of text based upon on-screen soft keyboard on electronic equipment
EP2874383B1 (en) System and method for controlling slide operation auxiliary input in portable terminal devices
KR102460737B1 (ko) 공중 필기 인식을 위한 방법, 장치, 기기 및 컴퓨터 판독 가능한 저장 매체
US8315482B2 (en) Integrated platform for user input of digital ink
US7778464B2 (en) Apparatus and method for searching for digital ink query
US8041120B2 (en) Unified digital ink recognition
KR101354663B1 (ko) 수기 기호 인지 방법 및 장치
US20120114245A1 (en) Online Script Independent Recognition of Handwritten Sub-Word Units and Words
KR20120011010A (ko) 수기 인식 방법 및 장치
WO2016107317A1 (zh) 输入法光标操作方法和装置
WO2015078383A1 (zh) 手写字符识别方法和系统
TW201201113A (en) Handwriting recognition method and device
CN103455527A (zh) 手写文件检索装置、方法以及记录介质
CN102207824A (zh) 一种基于触摸屏设备的滑行拼音输入法
JP2013025390A (ja) 手書き入力方法
KR20180030966A (ko) 중첩된 필기 인식 기술을 위한 시스템 및 방법
Nguyen et al. Semi-incremental recognition of on-line handwritten Japanese text
CN110858317B (zh) 手写识别方法及装置
Zafar et al. Neural nets for on-line isolated handwritten character recognition: A comparative study

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15721526

Country of ref document: EP

Kind code of ref document: A2

NENP Non-entry into the national phase in:

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC , EPO FORM 1205A DATED 27-09-16

122 Ep: pct application non-entry in european phase

Ref document number: 15721526

Country of ref document: EP

Kind code of ref document: A2