Back to Homepage |
I. Analyzing the faithfulness of our algorithm's facsimile capability
The columns from left to right respectively correspond to
the original handwriting by a calligraphist (Xo), facsimiled results when
our algorithm has no access to any characters written by the calligraphist (X0); and facsimiled results when our algorithm has learned 10, 20 and 40 characters written by the calligraphist as training samples, denoted as X10, X20, X40 respectively. |
Back to Top | Back to Homepage |
II. Results of phrase-facsimileing experiments
The sample characters in the personal handwriting database, shown for the first example below.
The facsimile result in the second row, compared with the original handwriting in the first row |
The facsimile result in the second row, compared with the original handwriting in the first row |
The facsimile result in the second row, compared with the original handwriting in the first row |
The facsimile result in the second row, compared with the original handwriting in the first row |
Back to Top | Back to Homepage |
III. Results of poem-facsimileing experiments
(I) First half part of a poem, originally written by a calligraphist and used as our training samples |
(II) Second half part of the poem, facsimiled by our algorithm |
(III) Second half part of the poem, originally written by the same calligraphist, as groundtruth data for comparison |
Back to Top | Back to Homepage |
IV. The User surveys to verify our algorithm's performance
Please go to the websites for the user surveys:
Back to Top | Back to Homepage |