Image analysis – Pattern recognition – On-line recognition of handwritten characters
Reexamination Certificate
2000-06-19
2004-02-24
Couso, Yon J. (Department: 2625)
Image analysis
Pattern recognition
On-line recognition of handwritten characters
C382S189000, C345S467000
Reexamination Certificate
active
06697524
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an information processing method and apparatus, and particularly, to an information processing method and apparatus which is capable of outputting a recognition result by comparing an input pattern with a dictionary pattern.
2. Related Background Art
Conventionally, information processing apparatuses have been known in which pattern information memorized for recognition purpose can be displayed upon a designation from the user.
However, the conventional apparatuses as above noted had the following drawbacks because registration pattern was simply displayed in the form of a list.
1) If the user registered numerous character patterns, the use condition of each pattern could not be found although some patterns were registered but unused or forgotten.
2) Though a single character could be registered multiply which may be a character or registered pattern in basic dictionary, the use condition of each input pattern could not be found.
3) When a plurality of same patterns were registered in like writing manner, it could not be found which pattern was effective.
Conventionally, character recognition apparatuses have been well known for recognizing an input character based on hand data of the character drawn by handwriting on the coordinate input plane such as an input tablet. Such character recognition apparatuses are provided with a registration function of registering track (hand) data specific to the user of which the recognition ratio will be low, corresponding to a particular character code. Such a registration process is carried out by a special application for registration which is activated fully independently of the process of character recognition input upon the start of registration process being designated by the user.
Typically, there are largely two cases where the user uses such a registration function. The first case occurs when a subject character is not contained in the dictionary of recognition device in use, and newly registered and added thereto. The second case occurs when the track of a character entered by the user is not matched with data of the dictionary provided in recognition device so that there is a high probability that the character is recognized falsely. In either case, the user has a very high probability that the entered character is not totally recognized or erroneously recognized. Therefore, in this case, the user tries to register the character code corresponding to the track of character to enhance the input efficiency. However, with the conventional method, it was not clear what character the user tried to register until the recognition result for handwritten character first entered was examined. Also, there was a drawback that the registration processing, which had to be by necessity performed out of the character input mode, not only would cause the user a lot of difficulty to operate, but would also require the same character as in the input process to be entered again in the registration process, whereby the registration might not take effect if there was a large difference between actually input character and registered character.
The present invention has been achieved in the light of the above-described conventional drawbacks, and its object is to provide a character recognition device which allows the character registration to be simply made in the character input operation, and a registration method in said device.
With the above configuration, the subject character is recognized by referring to a dictionary memory, based on the input hand data, and the recognition character candidates thus recognized are displayed. The displayed candidate characters can be registered into the dictionary memory by changing the order of candidates.
Also, according to another invention, the input hand data is memorized, and a character code designating and corresponding to the memorized hand data is registered into the dictionary memory. Thus, it operates so that the subject character is recognized by referring to the dictionary memory based on the input hand data.
Conventionally, handwriting recognition devices have relied on a method in which a registration dictionary for each user (called a user dictionary) is created to enhance the character recognition ratio. With this method, it is indispensable to enter user's hand data into the recognition device, select a standard feature pattern from the hand data, and register it in the user dictionary. Also, some devices may include a function of reproducing the track of input pattern to allow for the confirmation of the pattern of a character which was registered as the standard feature pattern.
However, the conventional reproduction methods of input patterns involved displaying the track at constant speed as in the input process, or displaying only the shape of pattern at a time.
Conventionally, in proper recognition means, when it was desired that the hand data of user's originality, which would not result in any aimed recognition result, be reflected on the recognizing means, a feature of hand data of user's originality was stored for registration in a user dictionary, and the character recognition process was performed using the user dictionary to obtain an aimed recognition result. Accordingly, if like hand data was presented by different users, the recognition result was sometimes different in some instances, so that it was necessary to provide a plurality of user dictionaries corresponding to a plurality of users. To this end, when a particular user tried to make character recognition, a method has been used in which a user dictionary is selected by preselecting the user name from a list of user dictionary holders which is displayed on the operation screen for user dictionary switching means, and the user dictionary of the particular user is set, as shown in FIG.
25
.
However, with the above configuration, when a plurality of users having the user dictionary registered made use of the same character recognition device alternately, it was necessary to call out the operation screen for executing user dictionary switch means to switch user dictionary even though the user was engaged in some other application, whereby if there were a great number of user dictionary holders, the selection of user dictionary resulted in a very troublesome operation.
For example, an on-line character recognition device has been known which can allow the user to confirm which hand data the character stroke information registered in recognition dictionary is based on in creation by drawing sequentially the character stroke data within the recognition dictionary to reproduce the hand data.
However, the above conventional example had a drawback that since a character was drawn at constant speed for all the strokes composing the character, the character made up of a great number of strokes would be time-consuming, and the user desiring the confirmation only for a part of the character might spend a wasteful amount of time.
Conventionally, character recognition devices, particularly, on-line handwritten character recognition devices have involved comparing a character written by the user on the tablet with the recognition dictionary within the character recognition devices and outputting a character code having the highest degree of similarity as the recognition result. However, none of the methods or devices have been publicized having a recognition ratio of 100% in the present character recognition techniques. Therefore, research institutes and companies have made various attempts to suppress the false recognition ratio to lower value.
One of them includes reproducing the standard pattern of character stored in the recognition dictionary in practice. This method includes reproducing the hand data by sequentially drawing the character data (coordinate data) within the recognition dictionary, so that the user can cause user's hand data to approximate to the standard pattern while referring to reproduced hand data to obtain
Arai Tsunekazu
Harada Takashi
Matsubayashi Kazuhiro
Mori Shigeki
Sakaguchi Katsuhiko
Canon Kabushiki Kaisha
Couso Yon J.
LandOfFree
Information processing method and apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Information processing method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Information processing method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3303342