A writer identification method based on XGabor and LCS

Abstract

Writer identification is a popular research field in many languages such as English, Persian, Chinese, etc. The approaches of writer identification methods are dependent on the language because different languages letters have different pattern. In this paper, we have presented XGabor filter and proposed a language independent writer identification system. In the feature extraction phase of proposed method, Gabor and XGabor filters are used while in the classification phase, a new classification method is defined that is not based on any kind of distances among feature vectors. The proposed classifier uses the sequence similarity of the Sorted Order of Features (SOF). To measure this similarity, the Longest Common Subsequence (LCS) algorithm is employed. In simulation phase, two databases in different languages have been used. First one consisted of 100 people's Persian handwritings and second one had 30 people's English handwritings. The accuracy of the system was satisfactory in both of them.

Off-line Arabic Character-Based Writer Identification – a Survey

“…SYSTEMS-STATE of the ART B. Helli et al proposed four systems [5,13,20,21] using PD100 data set for testing, which contains 500 samples from 100 writers and one system [20] using 350 samples written by 70 people .…”

“…In [13], new classification methods proposed that measures the sequence similarity of sorted order of features, then the Longest Common Subsequence (LCS) method is used to compare sequence similarity. While the feature vectors were extracted using Gabor and XGabor filters.…”

“…ability to determine the differences between the handwriting styles that are used to detect the correct writer's of the text image precisely. (writer discrimination) concern by researchers always interest by researchers [8] extremely interest in the last decade [10][11][12][13][14][15][16][17] According to [4] writer identification capable to minimize the ambiguities in pattern recognition process if writer's writing behaviour is available to the handwriting recognition system.…”

A Survey on Writer Identification Schemes

“…In another recent work proposed an LCS (longest common subsequence) based classifier to classify features that are extracted by Gabor and XGabor filters [53,54]. This classifier improved the system accuracy up to 95% on PD100.…”

A text-independent Persian writer identification based on feature relation graph (FRG)

“…In another recent work, we proposed an LCS (longest common subsequence) based classifier to classify features that are extracted by Gabor and XGabor filters [45,46]. This classifier improved the system accuracy up to 95% on PD100.…”