55.Chapter 8: Off-line Text-independent Writer Identification for Chinese Handwriting

Encouraged by the strong requirements of information security, the rapid development of biometrics becomes a new focus in both academic and industrial research. Writer identification is a branch of behavioral biometrics using handwriting with a natural writing attitude as the individual characteristic for identification. Off-line text-independent writer identification is to identify a person based on the static handwritten data with unrestricted text content. We propose an effective method using the contour-directional feature (CDF) combined with the modified SIFT for Chinese writer identification. The investigation demonstrates that both features are capable of describing the characteristic of handwriting. In the stage of the modified SIFT extraction, a simple connected-component based segmentation algorithm is used to segment the handwriting image into character regions, and the modified SIFT descriptors are extracted from the character regions. Then, a codebook is constructed by K-means clustering. With the codebook, the occurrence histogram of the modified SIFT for each handwriting image is calculated. Both features are concatenated together to represent the characteristic of handwriting. Experimental results show that the proposed method is able to improve the performance and is superior to other methods in terms of identification accuracy. On the HIT-MW Chinese handwriting dataset involving 240 writers, the Top-1 accuracy is 96.3%, and the Top-10 accuracy is 99.2%.

Chapter 8:Off-line Text-independent Writer Identification for Chinese Handwriting, Y.-J. Xiong and Y. Lu*, Advances in Chinese document and text processing, Series on Language Processing, Pattern Recognition, and Intelligent Systems, 2017, 2 (8): 215-234