Fig. 1. Tibetan document image

Fig. 2. Flow chart of layout analysis method

Fig. 3. RLSA samples. (a) Pixels before smoothing; (b) pixels after smoothing

Fig. 4. ARLSA process of text lines in Tibetan document images. (a) Binary figures; (b) ARLSA processing results

Fig. 5. Filtering results in connected domains. (a) ARLSA processing result; (b) rectangular outer box for connected domains; (c) filtering result

Fig. 6. Vowel attribution separated from baseline. (a) ARLSA processing result of text line; (b) centroids of connected components; (c) vertical distance between centroids; (d) filtering result

Fig. 7. Cluster analysis graphs of random segmentation block sample data. (a) Random sample data distribution; (b) K=3 cluster

Fig. 8. Structural diagram of layout data

Fig. 9. Word segmentation and recognition. (a) Separation of vowels and base words, and word adhesion; (b) segmentation result; (c) recognition result

Fig. 10. Layout analysis results. (a) Original image; (b) target connected region; (c) classification result of layout elements; (d) layout description

Fig. 11. Wrong classification images. (a)(c) Original images; (b)(d) wrong classification results