Claims Analysis
Total Claims: 6
Claim 1
Independent Claim
224 words
In a device which comprises: means for scanning an unknown character pattern and providing a video signal corresponding to an image of the unknown pattern, binary coding means for dividing the image into an array of pixels and converting the video signals into binary values using a threshold level, feature extracting means for converting the binary values corresponding to the pixels into segments corresponding to the image, memory means for storing the segments corresponding to the image, and operating means for performing reprogramming operations on the data stored in the memory means; an improved method comprising the steps of: obtaining repeatedly a bit matrix of an unknown character in a background pattern and decomposing it into a character component and a background component; modifying a binary coded level in sequential order with respect to a dictionary character pattern; statistically processing separately said character component and said background component of said individual bit matrices for obtaining horizontal, vertical and mask bit matrices for each category of characters; storing said individual bit-matrices as dictionary patterns in a given memory; obtaining evaluation indices on the basis of said individual bit-matrices of said character component and said background component, and of said horizontal, vertical and mask bit-matrices according to components of said dictionary patterns; and recognizing said unknown character pattern from the results of processing the evaluation indices.
Claim 2
Dependent Claim
224 words
In a device which comprises: means for scanning an unknown character pattern and providing a video signal corresponding to an image of the unknown pattern, binary coding means for dividing the image into an array of pixels and converting the video signals into binary values using a threshold level, feature extracting means for converting the binary values corresponding to the pixels into segments corresponding to the image, memory means for storing the segments corresponding to the image, and operating means for performing reprogramming operations on the data stored in the memory means; an improved method comprising the steps of: obtaining repeatedly a bit matrix of an unknown character in a background pattern and decomposing it into a character component and a background component; modifying a binary coded level in sequential order with respect to a dictionary character pattern; statistically processing separately said character component and said background component of said individual bit matrices for obtaining horizontal, vertical and mask bit matrices for each category of characters; storing said individual bit-matrices as dictionary patterns in a given memory; obtaining evaluation indices on the basis of said individual bit-matrices of said character component and said background component, and of said horizontal, vertical and mask bit-matrices according to components of said dictionary patterns; and recognizing said unknown character pattern from the results of processing the evaluation indices.
Claim 3
Dependent Claim
49 words
An improvement in the improvement of claim 1, further comprising means for totalizing data on said unknown character patterns according to a category of characters and means for modifying an already-learned dictionary pattern by performing processing on a new dictionary pattern on the basis of the thus totalized results.
Claim 4
Dependent Claim
28 words
An improvement in the improvement of claim 2, wherein said evaluation indices comprise: ##EQU11## where: B D is a stain bit-matrix, and B K is a cut bit-matrix.
Claim 5
Dependent Claim
42 words
An improvement in the improvement of claim 3, wherein there is included means for rejecting during said recognizing if equations D TI ≦D T3 and |D T2 -D T1 |≧D T4 are not satisfied, where D T is a total dissimilarity quantity.
Claim 6
Dependent Claim
96 words
An improvement in the improvement of claim 4 wherein said means for totalizing data comprises means for totalizing D MC , D CC , D C , D MB , D CB , D B and D T with respect to character K 1 , where: D MC is a character stain quantity, D CC is a character cut quantity, D C is a character dissimilarity quantity, D MB is a background stain quantity, D CB is a background character quantity, D B is a background dissimilarity quantity, and D T is a total dissimilarity quantity.